Based on: Bacteria and Viruses read through the Vacuum.Net Theory

Jump to the article here

Medicine is full of catalogues.

Diseases are sorted by organ. Or by pathogen. Or by mechanism.

But none of these sorting systems tells you what can actually be done.

This paper offers a different cut.

Every disease state can end in one of only four ways.

1. Let it close. The body finishes the job itself. No treatment needed.
2. Remove one party. Kill or clear the other participant. Antibiotics are the classic case.
3. Supply what is missing. Replace an absent component. Insulin is the classic case.
4. Restore proportion. Both parties stay. What is wrong is the relation between them. Nothing is removed. Nothing is supplied. The balance must change.

Three of these four operations are crowded. They have industries, trial designs, and regulators built around them. The fourth is almost untouched. Yet the fourth holds sepsis, ARDS, severe influenza, severe COVID-19, the autoimmune diseases, Ebola, and — on the evidence in the paper — atherosclerotic cardiovascular disease. That last one kills more people than anything else on earth.

The asymmetry is historical, not biological. Operation 2 produced the most spectacular results in the history of medicine. The whole field then reorganised around it: find the agent, kill the agent. Operation 3 followed. Genetic medicine has made it technically routine. Operation 4 never got a name. It never got a standard trial design. It never got a regulatory pathway. And that is exactly where the largest remaining burden of death now sits.

Three consequences follow, and the paper builds on them.

The operation determines the endpoint. An operation-2 trial asks: was the agent cleared? An operation-4 trial cannot ask that. Nothing is being cleared. Measuring clearance produces null results for drugs that work.

The operation determines who to enrol. Operations 1 to 3 enrol by diagnosis. Operation 4 cannot. The same diagnosis can contain patients on both sides of the line where a treatment reverses its effect.

The wrong operation is not neutral. It causes harm. Several trials below quantify that harm.

2. Operation 1: let it close

Some diseases are self-limiting. The body completes the process on its own. Mild influenza is one. Viral pneumonia is another. Most wounds heal by themselves. Intervention adds nothing here. It can subtract.

The research question is not a treatment. It is: where is the boundary? Treating a patient whose process is still completing causes measurable harm.

RECOVERY gave the number. Among hospitalised COVID-19 patients who needed no oxygen, 28-day mortality was 17.0% on dexamethasone against 13.2% with usual care. That is a rate ratio of 1.22. The drug made things worse in exactly the group whose disease was still resolving. The confidence interval includes 1, so this is a direction, not a proven fact. But the point estimate points 3.8 percentage points the wrong way.

Antibiotics for viral infections repeat the same error at a much larger scale. The bill for that arrives in the next section.

Research direction 1: build the boundary test, disease by disease. The right clinical question is not “does this drug work?” It is “has this patient’s process stopped completing?” Today, that question has a bedside answer in exactly one condition: COVID-19, where the answer is whether the patient needs oxygen. Even that was found by accident, in a subgroup analysis. For every other condition in this row, the boundary is unmeasured. Success would be a simple admission measurement. It should separate patients whose process will complete from those whose will not. It should be validated against outcome without treatment.

3. Operation 2: remove one party

This operation built modern medicine. It still works beautifully where it works at all. Tuberculosis is curable. So are malaria, bacterial pneumonia, and hepatitis C. But two facts should govern further investment.

It is running into a wall. Deaths directly attributable to bacterial antimicrobial resistance were 1.14 million in 2021. The forecast for 2050 is 1.91 million. That is a rise of 67.5%. Associated deaths rise from 4.71 million to 8.22 million, up 74.5%. Cumulatively, 39 million deaths are forecast between 2025 and 2050. Deaths in adults over 70 rose more than 80% between 1990 and 2021.

The wall is structural, not technical. The operation works by removing a self-replicating party. A self-replicating party selects for escape. Every success narrows the space of future successes. No other operation has this property.

Where it works, the drug is no longer the constraint. Hepatitis C is the cleanest case. About 47 million people live with chronic infection. Roughly 68 million were ever infected between 2015 and 2024. Only about 24 million knew their diagnosis — 35%. Only about 13 million had received the direct-acting antivirals by the end of 2024. That is 19% of the ever-infected. Around 44 million people were never diagnosed at all. A curative eight-week regimen exists. Four out of five people who need it have not had it. No new molecule closes that gap.

This leads to three directions:

• Direction 2: treat delivery as the research problem, not the molecule. Where a cure already exists, the remaining deaths are a diagnosis and logistics problem. That problem is bigger than any new compound.
• Direction 3: stop paying for removal where nothing needs removing. Antibiotic use in self-limiting viral illness drives the resistance curve above. The boundary test of Direction 1 is also the cheapest lever on antimicrobial resistance.
• Direction 4: for resistant organisms, change operation, not compound. Phage therapy, anti-virulence drugs, and host-directed approaches all abandon removal. The argument for them is not that they are promising. It is that removal has a mathematical ceiling, and these do not.

The same ceiling, inside one patient

The resistance curve above describes a population over decades. Oncology shows the identical structure inside one body, over months.

Osimertinib is the best first-line drug for EGFR-mutated lung cancer. In the FLAURA trial, median progression-free survival was 18.9 months against 10.2 months on the older agents. That is a hazard ratio of 0.46 — close to a doubling. Median overall survival was 38.6 months against 31.8 months. That is a gain of 6.8 months, a hazard ratio of 0.80. At three years, 28% of patients were still on the regimen. Which means 72% were not.

Look at the gap between those numbers. Time before return nearly doubles. Survival gains only a fifth. What is removed is removed. What remains is what the removal did not reach. And what remains keeps growing. This is the antimicrobial resistance curve, with one body replacing the population and eighteen months replacing the decades.

So cancer is not a second instance of operation 2 waiting for a better compound. Nothing foreign is present. Nothing is missing. What does not settle is division itself. That places cancer structurally in the fourth row. It is the one fourth-row condition with no established boundary marker. Naming that marker is the open problem. Finding the next inhibitor is not.

4. Operation 3: supply what is missing

Where a component is absent, supplying it ends the disease. Insulin and vitamin C are the historical cases. The modern form is genetic. And it now works.

Exagamglogene autotemcel (exa-cel) edits BCL11A to restore fetal haemoglobin. Across more than 100 treated patients — 46 with sickle cell disease, 56 with beta-thalassaemia — the results are categorical. In sickle cell disease, 92.3% of evaluable patients had no vaso-occlusive crisis for at least 12 consecutive months. The longest crisis-free period was 54.8 months. In thalassaemia, 94.2% became transfusion-independent for at least 12 consecutive months, the longest for 59.4 months. A single administration converted a lifelong disease into its absence. The effect is approaching five years old and still holding.

The constraint is not biology. Each therapy is built once, for one mutation. The cost can only be amortised over large patient populations. That excludes most of the conditions the operation could solve. There are thousands of single-gene disorders. Approved genetic therapies exist for only tens of them.

Direction 5: make the platform the product, not the therapy. The target is a delivery and editing platform where changing the target sequence does not restart development, manufacturing, or regulatory review. Judge every proposal by what it does to the cost of the second indication, not the first. Two developments bear on this. Targeted lipid nanoparticles can generate engineered cells inside the body, removing per-patient manufacturing. And computational protein design can now generate binders and antibodies against specified targets with atomic accuracy, shortening the pipeline’s front end.

Direction 6: audit the boundary between operations 3 and 4. Not every deficiency is a deficiency. When supplying the “missing” component fails to end the disease, the classification was wrong. The condition belongs in the fourth row. The audit is cheap. It uses trial results that already exist.

5. Operation 4: restore proportion

Here, both parties remain. Nothing is cleared. Nothing is missing. The relation between the parts is what is wrong. The reference catalogue marks these conditions “relief only”: nothing removes them. Severe influenza is filed that way. So are sepsis, ARDS, severe COVID-19, the autoimmune diseases, and Ebola virus disease — for which no licensed vaccine or therapeutic exists against the Bundibugyo species behind the 2026 outbreak in the Democratic Republic of the Congo.

The first assumption to break is inherited from operation 2: that more response is better. It is not.

A Sierra Leone cohort of 112 Ebola patients showed this clearly. Of 246 differentially abundant transcripts, 220 were higher in those who died. Their interferon signalling was stronger. Their acute-phase responses were stronger. Fibrinogen levels were roughly 54-fold, 28-fold, and 19-fold higher in fatal cases. Meanwhile, viral load failed to separate outcome where it mattered. At the extremes, cycle threshold predicted death (Ct 12) or survival (Ct 30). Around the cohort mean of 21.4, outcome was close to even. Host-gene classifiers reached 79–92% accuracy in exactly that range. In this row, the size of the response is not a measure of defence. It is a measure of what is not settling.

Why trials in this row keep returning null

Because the intervention reverses sign inside the enrolled population. A single pooled estimate averages the two halves. RECOVERY showed the reversal inside one randomisation:

Respiratory support at entry	Dexamethasone	Usual care	Rate ratio
None	17.0%	13.2%	1.22 (0.93–1.61) — harm
Oxygen only	21.5%	25.0%	0.80 (0.70–0.92) — benefit
Invasive ventilation	29.0%	40.7%	0.65 (0.51–0.82) — strong benefit

Mortality under usual care rises across the strata: 13.2, 25.0, 40.7. The rate ratio falls: 1.22, 0.80, 0.65. The same drug, at the same dose, in one trial, changes sign.

HARP-2 showed what happens without stratification. It randomised 540 ARDS patients to simvastatin or placebo and found no overall survival difference. A secondary analysis then applied latent class analysis to baseline data alone — no outcomes used. Two classes emerged: 353 patients (65%) hypoinflammatory, 186 (35%) hyperinflammatory. Their baseline 28-day mortality was 39% against 17%. Within the hyperinflammatory class, simvastatin significantly improved 28-day survival (P = 0.008). A trial filed as negative contained a benefit. The stratifier was computable from measurements taken before treatment.

SAVE-MORE showed the converse, by getting the design right from the start. Anakinra was given to hospitalised COVID-19 patients selected by plasma suPAR of 6 ng/mL or more, excluding the most severely hypoxic. The primary outcome improved with an odds ratio of 0.36. 28-day mortality was 3.2% against 6.9%. Entry was decided by a measured marker, not by clinical stage. The boundary is therefore not lateness and not crude severity. It is a state that can be measured. A patient can be past it while still looking mild.

Direction 7: stratify at randomisation on a measured marker, and report the strata separately. Every trial in this row should be designed this way. A single pooled estimate is not conservative here. It is a design that destroys the signal.

6. Operation 4 outside the intensive care unit

Everything above comes from critical care, where death arrives in days. If the fourth row were confined there, it would be a small category. It is not.

CANTOS tested canakinumab, an antibody against interleukin-1-beta, in 10,061 people who had had a myocardial infarction and whose baseline hsCRP was 2 mg/L or above. Entry was already decided by a measured marker, as in SAVE-MORE. Three results matter.

The lipid route was not touched. LDL cholesterol went slightly up on the 150 mg dose (+1.7%) and events went down. Fifty years of cardiology have aimed at that lipid quantity. Whatever changed, it was not that.

Pooled, the effect is real and small. Primary events ran at 3.86 per 100 person-years on 150 mg against 4.50 on placebo. That is a hazard ratio of 0.85 — a 14.2% reduction.

The statistical space was divided across the wrong thing. The trial split its alpha across three doses of one drug. The 150 mg dose passed its prespecified threshold by a margin of 0.0004. The 300 mg dose produced essentially the same effect (HR 0.86) but was declared non-significant, because its own threshold was stricter and its P value nearly three times that threshold. Two doses of one antibody produced the same effect. One counted. One did not.

Then the division that matters: on-treatment hsCRP at three months, below or above 2 mg/L.

Endpoint	hsCRP < 2 mg/L	hsCRP ≥ 2 mg/L
MI, stroke or cardiovascular death	0.75 (0.66–0.85)	0.90 (0.79–1.02)
Cardiovascular death	0.69 (0.56–0.85)	0.99 (0.82–1.21)
Death from any cause	0.69 (0.58–0.81)	1.05 (0.90–1.22)
Number needed to treat over 5 years	16	57

All-cause death falls by nearly a third on one side of the line and crosses 1 on the other. The NNT differs 3.6-fold. The share of patients reaching the lower state rose with dose: 44%, 55%, 65%.

An honest caveat: this division is weaker evidence than HARP-2 or SAVE-MORE. HARP-2 classified on baseline data alone. SAVE-MORE decided entry before randomisation. CANTOS divided on a measurement taken three months after treatment began. The groups are therefore defined partly by how they responded. A group that responds may simply have had more room to improve. What CANTOS establishes is not that this marker identifies a treatable stratum. It is that a pooled estimate can conceal a division large enough to reverse the sign of mortality. The trial that would settle it: enter on the baseline marker, dose, measure at three months, randomise again.

The harms are in the same record. Fatal infection or sepsis ran at 0.31 per 100 person-years against 0.18. Against that, fatal cancer fell (0.45 against 0.64) and the primary endpoint fell by 0.64. This is not a defence of the drug, which was never licensed for this indication. It is a description of the fourth row: nothing is removed, and the cost appears exactly where the capacity to settle an infection is reduced.

The consequence is the most important in the paper. Atherosclerotic cardiovascular disease is the largest single cause of death in the world. If it belongs in the fourth row — and the LDL result is the strongest available argument that it does — then the fourth row is not a residual category of rare and desperate conditions. It is the largest of the four. And it has no standard trial design, no accepted biomarker class, and no regulatory precedent.

Direction 10: divide the statistical space across strata, not across doses. CANTOS spent its alpha on three doses and had none left for the division that reversed mortality. The correction is purely procedural: prespecify fewer doses and more strata. The dose can be found afterwards. The stratum cannot be recovered once the result is filed as negative.

7. The first clean demonstration that operation 4 is real

Autoimmune disease is the purest case in the fourth row. Nothing is missing. Nothing foreign is present. The relation is wrong. And treatment has consisted of holding it down indefinitely.

Eight patients with treatment-resistant systemic lupus erythematosus received a single dose of CD19-directed CAR T cells between March 2021 and June 2023. All eight entered remission, with SLEDAI scores of zero. All came off immunosuppressants, including glucocorticoids. Autoantibodies disappeared — one antibody in one patient excepted — and stayed negative through 12 to 24 months of follow-up. Naive B cells returned within months. The disease did not return with them. Patients retained vaccine responses to measles, mumps, rubella, chickenpox, Epstein-Barr, tetanus, and pneumococcus.

The B cell compartment was not removed permanently. It was depleted. It re-formed. And what came back did not reproduce the disease. The relation reset.

The caveats are real. Eight patients. An ethnically narrow cohort. Cytokine release syndrome and neurotoxicity as risks. No US approval for this indication. A cost of roughly $375,000 to $425,000 per infusion.

But the result is categorical, not incremental. These patients are not better managed. They are off treatment, with no disease activity. Nothing in the first three operations predicts that outcome for lupus.

Direction 8: pursue reset rather than suppression across the fourth row. The question to carry from lupus into sepsis, ARDS, and severe viral disease is not which mediator to block. It is which compartment can be made to re-form in a settled state. Depletion followed by reconstitution is one mechanism. It is unlikely to be the only one.

The cost structure is also where the directions converge. Generating engineered cells inside the body, by targeted nanoparticle delivery rather than external manufacture, is the step that would move a reset from an eight-patient demonstration to a usable intervention. That is the same platform problem as Direction 5. Solving it once serves both rows.

8. What it costs to pick the wrong operation

Alzheimer’s disease has absorbed more investment than any other fourth-row target — under an operation-2 programme: identify the agent, remove the agent. Amyloid was the agent. The removal works.

The clinical result does not follow. In the lecanemab phase 3 trial over 18 months, CDR-SB declined 1.21 points on treatment against 1.66 on placebo. That is a difference of 0.45 points, reported as “27% slowing”. For donanemab over 76 weeks, iADRS declined 10.2 against 13.1 — a difference of 3.4 points, reported as “33%”.

Now set those against the minimal clinically important differences: 1.0 on CDR-SB in mild cognitive impairment; 1.6 in mild dementia; 5 and 9 on iADRS. The lecanemab effect is 0.45 of the smaller threshold. One analysis concludes that changes of about 0.5 on CDR-SB and about 3 on iADRS are imperceptible to patients, families, and physicians.

The harms are not imperceptible. Nearly one in four lecanemab patients developed brain swelling or bleeding. More than one in three on donanemab. Treatment-related adverse events ran at 45% and 89%. Severe bleeding occurred in 5 patients against 1 on placebo for lecanemab, and 7 against 2 for donanemab. The number needed to harm is about 3 for both drugs. The number needed to treat could not be computed from the published data.

This is what applying operation 2 to a fourth-row condition produces. The agent is successfully removed. The relation is unchanged. The benefit sits below the threshold of perception. The harm does not.

The correct reading is not that amyloid is irrelevant. It is that removal was the wrong operation — for the same reason killing the virus is the wrong operation in severe influenza. The research question becomes: which relation in the brain has stopped settling, and what would make it settle? That is a fourth-row question. It has never been the one the field was funded to ask.

Direction 9: re-classify the major chronic diseases by operation before funding another removal programme. Neurodegeneration, metabolic disease, chronic kidney disease, heart failure, and the major psychiatric disorders have all been pursued predominantly as operation-2 or operation-3 problems. Each should be re-examined against the four rows. The trial results that already exist are sufficient to do it.

9. The ten directions, collected

#	Direction	Judged by
1	Build the boundary test, disease by disease	Prospective separation of untreated outcomes
2	Treat delivery as the research problem where a cure exists	Share of the ever-infected treated
3	Stop applying removal where nothing needs removing	Reduction in inappropriate antimicrobial courses
4	For resistant organisms and returning tumours, change operation, not compound	Durability under selection, not initial potency
5	Make the editing platform the product	Cost of the second indication
6	Audit the operation-3 / operation-4 boundary	Programmes correctly redirected
7	Stratify at randomisation on a measured marker	Sign reversal detected rather than averaged away
8	Pursue reset rather than suppression	Drug-free remission, not disease control
9	Re-classify the major chronic diseases by operation	Removal programmes stopped where the row is wrong
10	Divide the statistical space across strata, not across doses	Effects retained instead of filed as negative

Three of these — 2, 3, and 6 — require no new science. They are re-allocations, and they act on larger numbers than anything in the pipeline. Two — 5 and 8 — converge on one engineering problem: generating engineered cells inside the body. And two — 7 and 10 — are free. They are changes in how a trial is written down.

10. What follows for each reader

For clinicians. Before reaching for an intervention, name the operation. Where the process is still settling, suppression is not neutral. In RECOVERY’s no-oxygen group, the point estimate ran 3.8 percentage points against the patient. Where an operation-2 cure exists, the question that decides mortality is whether the patient was ever diagnosed — not which regimen to choose.

For trialists. A null result in the fourth row is not evidence of no effect. It is evidence of an unstratified population, until a stratified analysis says otherwise. Fix the strata before allocation. Choose the stratifier as a measured marker, not a clinical stage. Publish the strata separately. The archive of null immunomodulation trials is a research resource, not a closed file.

For funders. Three directions require no new science and act on the largest numbers. Two more cost nothing but a change in protocol writing. These compete poorly for funding because they produce no molecule. They are where the available mortality reduction is concentrated.

For industry. The fourth row is the largest unserved category in medicine. It is unserved because it has no standard trial design, no accepted biomarker class, and no regulatory precedent for a stratified claim. The company that establishes that pathway — a marker, a stratified design, a label naming a stratum rather than a diagnosis — defines the category. The commercial argument for the fourth row is not that it is virtuous. It is that it is where the volume is, and where no one has a design.

One caution. This framing does not predict that any particular intervention will work. It predicts where to look, which populations to enrol, and which negative results deserve re-examination. Every claim it makes is refutable by the trials it proposes.

11. Verification and limitations

The paper states that every figure was recomputed from the reported counts or percentages before being quoted, and it publishes the checks. Two discrepancies are left standing rather than smoothed: a hepatitis C treatment share that computes to 54% of the diagnosed where the source says about 20%; and a diagnosed share of 35% against a stated 36%, attributed to rounding.

The stated limitations are substantial and worth repeating. The CAR-T lupus result rests on eight patients in a single centre, with 12 to 24 months of follow-up, in an unapproved indication. The HARP-2 subphenotype analysis is secondary. The RECOVERY no-oxygen stratum’s confidence interval includes 1, so harm there is a direction, not a finding. The CANTOS division uses a post-treatment measurement, a weaker design than the other two trials. The FLAURA comparator allowed crossover, and compared against first-generation inhibitors rather than against no removal at all. The classification was checked in detail against one chapter of one reference catalogue, not exhaustively. Cancer’s placement in the fourth row is an argument from the shape of its failures, not a demonstration — no boundary marker for it is proposed. And the four-operation scheme itself derives nothing and predicts no effect size: what is refutable is every row assignment in it, and that is where it should be attacked.

---

Annotated reference list

The sources below are as cited in the original paper. The annotations summarise what each source contributed to the argument.

Stratified and biomarker-guided trials

1. RECOVERY Collaborative Group — dexamethasone in hospitalised COVID-19 (with subgroup outcomes as summarised by The Bottom Line and First10EM). The platform trial that provided the central sign-reversal evidence: rate ratios of 1.22, 0.80, and 0.65 across the no-oxygen, oxygen, and ventilated strata. It grounds the claims that operation determines benefit, that the wrong operation harms, and that a simple boundary marker (need for oxygen) already exists in one disease.
2. Calfee, C. S. et al. — ARDS subphenotypes and differential response to simvastatin (HARP-2 secondary analysis). Latent class analysis of 540 ARDS patients on baseline data alone. It split them into hyperinflammatory and hypoinflammatory classes with 39% versus 17% baseline mortality, and showed a significant simvastatin survival benefit in one class only. It is the paper’s demonstration that a “negative” trial can contain a real, recoverable effect.
3. SAVE-MORE (Kyriazopoulou et al.) — early anakinra guided by plasma suPAR in hospitalised COVID-19. A phase 3 trial that enrolled patients on a measured biomarker threshold (suPAR ≥ 6 ng/mL) rather than clinical stage. Odds ratio 0.36 on the primary outcome; 28-day mortality 3.2% versus 6.9%. It is the prospective proof that biomarker-guided entry makes an immunomodulator work.
4. RECOVERY — tocilizumab (summaries by REBEL EM and the European Respiratory Society); CAPE COD (NEJM 2023, Critical Care Reviews) and CAPE COD versus ESCAPE. Further critical-care immunomodulation trials used to establish the fourth row in the ICU: what worked, what failed, and how design choices — stratification, timing, and endpoint — decided the outcome.

Ebola virus disease

5. Liu et al., Genome Biology 2017 — Ebola patient transcriptomes. The Sierra Leone cohort of 112 patients. Source of the paper’s key conceptual break: 220 of 246 differentially abundant transcripts higher in fatal cases; fibrinogen fold-changes of ~54×, 28×, and 19×; viral load failing to predict outcome near the cohort mean while host-gene classifiers reached 79–92% accuracy. It shows that response magnitude is a measure of non-settling, not defence.
6. ECDC — multi-country Ebola outbreak 2026; LSHTM on Bundibugyo virus. Establishes that no licensed vaccine or therapeutic exists against the Bundibugyo virus species behind the 2026 Democratic Republic of the Congo outbreak — evidence that an entire lethal disease sits in the unserved fourth row.

Cell therapy and autoimmunity

7. American College of Rheumatology — CD19 CAR T-cell therapy in refractory systemic lupus erythematosus. The eight-patient series (treated March 2021–June 2023). All achieved SLEDAI-zero remission, discontinued immunosuppressants, lost autoantibodies, reconstituted naive B cells without relapse, and retained vaccine responses. It is the paper’s first clean demonstration that restoring proportion — reset rather than suppression — can end a fourth-row disease.

Gene editing

8. Vertex Pharmaceuticals — long-term exagamglogene autotemcel (exa-cel) data in sickle cell disease and beta-thalassaemia. The 46 + 56 patient dataset: 92.3% and 94.2% sustained crisis-free and transfusion-free rates at 12+ months, with the longest periods at 54.8 and 59.4 months and stable edited-allele levels. It shows operation 3 is technically solved — and, by its per-patient cost structure, why it is economically blocked.

Antimicrobial resistance

9. GRAM Project (University of Oxford) — global AMR burden forecasts. Source of the resistance ceiling figures: 1.14 million attributable deaths in 2021 rising to 1.91 million in 2050; associated deaths rising from 4.71 to 8.22 million; 39 million cumulative deaths projected 2025–2050; deaths in adults over 70 up more than 80% since 1990. It grounds the argument that operation 2 has a structural, mathematical limit.

Hepatitis C

10. “Working towards hepatitis C elimination” (Contagion Live) — global burden, cascade of care, and WHO 2030 targets. Source of the delivery-gap figures: ~47 million chronically infected, ~68 million ever infected, ~24 million diagnosed, ~13 million treated, ~44 million never diagnosed. It shows that where a cure exists, the binding constraint is diagnosis and logistics — not pharmacology.

Alzheimer’s disease

11. “Lecanemab and donanemab: an illustrated perspective on the data” (eNeuro, 2024). Source of the anti-amyloid analysis: the 0.45-point CDR-SB and 3.4-point iADRS differences, the minimal clinically important difference thresholds (1.0–1.6 and 5–9), the imperceptibility analysis (~0.5 and ~3 points), and the harms — brain swelling or bleeding in ~24% (lecanemab) and >33% (donanemab), adverse-event rates of 45% and 89%, and an NNH of about 3. It is the paper’s worked example of operation 2 applied to a fourth-row disease.

Design and delivery platforms

12. RFdiffusion — de novo design of protein structure and function; and atomically accurate de novo design of antibodies. Evidence that computational protein design can now produce binders and antibodies against specified targets, shortening the development pipeline that Directions 5 and 8 depend on.
13. In vivo CAR T-cell generation to treat cancer (and autoimmune disease) — targeted lipid nanoparticle delivery. The engineering step the paper identifies as the shared unlock for operations 3 and 4: generating engineered cells inside the patient, removing external per-patient manufacture, and collapsing the cost that currently blocks both rare-disease gene therapy and widespread cellular reset.

Atherosclerotic cardiovascular disease

14. CANTOS — canakinumab trial (summaries by WikiJournalClub and the American College of Cardiology). The 10,061-patient IL-1β trial in post-MI patients with hsCRP ≥ 2 mg/L. Source of the entry criteria, the per-dose hazard ratios (0.85 for 150 mg; 0.86 for 300 mg), the prespecified alpha thresholds and the 0.0004 significance margin, the LDL result (+1.7% with fewer events), and the harms (fatal infection/sepsis 0.31 vs 0.18 per 100 person-years). It grounds both the classification of cardiovascular disease in the fourth row and Direction 10’s warning about spending statistical space on doses instead of strata.
15. The CANTOS on-treatment hsCRP analysis (Residual Risk Reduction Initiative; TCTMD). Source of the stratified hazard ratios and the numbers needed to treat by hsCRP state (NNT 16 below 2 mg/L versus 57 above; all-cause mortality HR 0.69 versus 1.05). It shows a division strong enough to reverse the sign of mortality — while the paper itself flags that, being a post-treatment division, it is weaker evidence than HARP-2 or SAVE-MORE.

Oncology

16. FLAURA — first-line osimertinib in EGFR-mutated non-small-cell lung cancer (Cancer Management and Research; Journal of Oncology Navigation and Survivorship). Source of the intra-patient resistance argument: median progression-free survival 18.9 versus 10.2 months (HR 0.46), median overall survival 38.6 versus 31.8 months (HR 0.80), and 28% versus 9% still on regimen at three years — meaning 72% off it. The paper reads this as the AMR curve compressed into one body and eighteen months, placing cancer in the fourth row.

Framework

17. J. Konstapel, The Vacuum.Net Theory, Ninth Edition (Constable Research, Leiden, 27 October 2026) — Foundations 9.2 and Physics 9.0. The author’s underlying theoretical framework from which the four-operation scheme derives.
18. J. Konstapel, Where Illness Comes From, Completed (Constable Research, 20 October 2026). Source of the “cure statuses” (self-limiting; curable by removal; curable by supply; relief only) used to classify conditions into the four rows.
19. J. Konstapel, Depth of Discharge (Constable Research, 29 October 2026). The companion work that sets out the stratification argument in full — the basis for Directions 7 and 10 and for the claim that null fourth-row trials average away real effects.

Primary source

20. J. Konstapel, “Research Directions to Cure All Diseases,” Constable Research, Leiden, 29 October 2026. The paper this essay is based on. It sets out the four-operation classification, the evidence for each row, the ten research directions, and its own verification and limitations section.

J.Konstapel, Leiden,29-10-2026.

Article

Jump tpo the Article here.

J.Konstapel,Leiden,29-10-2026.

Does suppressing an immune response save a life, or cost one?

The answer is not a property of the drug. Nor is it a property of the disease.

One trial measured the difference inside a single randomisation. In COVID-19 patients who needed no oxygen, dexamethasone raised mortality.

In patients on a ventilator, it cut mortality by 11.7 percentage points.

The drug was the same.

What differed was how far the damage had already travelled.

The category where nothing removes it

A reference work files every condition by what can be done with it. It uses four cure statuses.

• SELF-LIMITING. The body completes the discharge itself. Intervention adds nothing.
• CURABLE. A treatment exists that removes the condition.
• CONTROLLABLE. The state can be held. The deviation remains.
• RELIEF ONLY. Nothing removes it.

Run through the infectious chapter, the distribution is stark. Tuberculosis is CURABLE on a four-drug regimen. Malaria is CURABLE on artemisinin combination therapy. Bacterial pneumonia is CURABLE on amoxicillin. HIV is CONTROLLABLE on antiretrovirals. Viral pneumonia and mild influenza are SELF-LIMITING. Severe influenza is RELIEF ONLY.

The pattern is simple. Almost all of infectious medicine sits in one operation: remove one party, and the problem it was half of disappears with it. Where that operation is unavailable, the reference work says RELIEF ONLY and stops.

Ebola sits there. There is no licensed vaccine for Bundibugyo virus. There is no licensed treatment either. That is the species behind the 2026 outbreak in the Democratic Republic of the Congo. The licensed products were built against the glycoprotein of Zaire ebolavirus. Supportive care is the standard of care. Severe influenza, sepsis, ARDS and severe COVID-19 sit in the same category. They sit there for the same structural reason. It is not that each is separately mysterious.

This paper asks what the operation for that category is. It proposes that one trial has already measured it, without naming it.

Four operations

Read through a closure framework, a condition is not an invader and a host. It is a joint between two things. Each holds its form. They do not share an order. Damage is what fails to close at one scale and therefore moves to the next. That gives exactly four ways a condition can end. The four cure statuses map onto them one to one.

Operation	What happens	Cure status	Example
It closes by itself	The discharge completes at its own scale	SELF-LIMITING	Viral pneumonia, mild influenza
One party is removed	There is nothing left to fail to close with	CURABLE	Tuberculosis, malaria, bacterial pneumonia
What is missing is supplied	There was no second closed form at all	CURABLE by replacement	Insulin in type 1 diabetes, vitamin C in scurvy
The proportion is restored	Both parties remain; the joint stays open	unfilled	Severe influenza, sepsis, ARDS, severe COVID-19, Ebola

The fourth row is empty. It is empty for a historical reason, not a biological one. Antibiotics made the second row spectacular. The second row absorbed the field.

Two observations suggest the fourth row is not empty of evidence. It is empty of a name.

First: in severe disease, the response is larger in those who die, not smaller. In a Sierra Leone Ebola cohort, 220 of 246 differentially abundant transcripts were higher in fatal cases. Interferon signalling was stronger. Acute-phase responses were stronger. Magnitude is not a measure of defence there. It is a measure of what is not closing.

Second: the interventions that act on the fourth row already exist. They are the anti-inflammatory and immunomodulatory agents. The literature treats their record as inconsistent.

The gradient inside one trial

The record is not inconsistent. It is graded. The grading is visible inside a single randomisation.

RECOVERY allocated hospitalised COVID-19 patients to dexamethasone or usual care. It reported 28-day mortality by level of respiratory support at randomisation.

Respiratory support at entry	Dexamethasone	Usual care	Absolute difference	Rate ratio
None	17.0%	12.2%	+2.9 pp	1.22
Oxygen only	21.5%	25.0%	-3.5 pp	0.80 (0.70–0.92)
Invasive ventilation	29.0%	40.7%	-11.7 pp	0.65 (0.51–0.82)

Two monotone series run through that table, in the same order. Mortality under usual care rises: 13.2, 25.0, 40.7. The rate ratio falls: 1.22, 0.80, 0.65.

The level of respiratory support is not a severity score bolted on afterwards. It states which scale the damage has reached: none, the lung, the whole organism. The same drug, at the same dose, reverses sign across that range. In the group whose discharge had not yet left the tissue, suppression went the wrong way by 3.8 percentage points. In the group whose discharge had taken the whole body, it saved 11.7.

This is the cleanest evidence available on the question. The comparison is randomised. The agent is identical. The contrast is internal. Nothing differs between the three rows except how far the damage had travelled when the drug was given.

A threshold, not a slope

RECOVERY’s tocilizumab comparison enrolled hospitalised COVID-19 patients with hypoxia and C-reactive protein of at least 75 mg/L. Mortality at 28 days was 31% against 35%. The rate ratio was 0.85 (0.76–0.94). Within that population, the benefit was consistent across levels of respiratory support.

Taken beside the dexamethasone table, that looks contradictory. It is not. The entry criteria had already done the selection. Requiring hypoxia plus CRP ≥ 75 mg/L is a statement that the discharge has left the tissue. Everyone enrolled was above the line. No grading remained to be seen.

That is the shape to expect if closure is the governing quantity. A window closes or it does not. There is no partial closure. The dexamethasone table spans the boundary. It shows the sign change. The tocilizumab trial sits entirely on one side of it. It shows a flat effect.

The practical form of the claim is therefore not “more severe, more benefit”. It is this: below the line, suppression removes a response that was still completing. Above the line, suppression removes damage that has no path left to close.

Where the evidence does not cooperate

The pneumonia trials are the obvious place to look for confirmation. They do not give it cleanly.

CAPE COD randomised 800 patients with severe community-acquired pneumonia, in intensive or intermediate care; 795 were analysed. Twenty-eight-day mortality was 6.2% on hydrocortisone (25 of 400) against 11.9% on placebo (47 of 395). The difference was -5.6 percentage points (95% CI -9.6 to -1.7; P = 0.006). Entry required a Pulmonary Severity Index above 130, or ventilation with PEEP ≥ 5, or high-flow oxygen with PaO₂:FiO₂ below 300, or a non-rebreather mask. Septic shock was an exclusion. That is a depth criterion. The result fits.

ESCAPe gave low-dose methylprednisolone in severe community-acquired pneumonia. It found 60-day mortality of 16% against 18%. The absolute difference was 2 percentage points. The confidence interval ran from -8 to +5. P was 0.61. If depth alone decided the matter, ESCAPe should have been positive. A simple severity reading makes it worse, not better: 33% of ESCAPe patients were ventilated at enrolment, against roughly half of CAPE COD patients who were not.

The documented differences between the two trials are not about depth. ESCAPe enrolled 584 of the 1,406 patients it was powered for. Its population was 96% male, with a mean age of 69. CAPE COD stopped early at its second interim analysis. Trials stopped early tend to overstate benefit. One commentator’s reading: the true effect lies between the two, nearer 2 points than 6.

The honest conclusion: between-trial comparisons cannot carry this argument. Agent, dose, timing, population and power all move together across trials. Depth moves with them. The claim rests on the within-trial gradient, where nothing else moves. That restriction is not a weakness of the evidence. It is a specification of the experiment that would settle it.

Ebola has no oxygen gauge, but it has a coagulation one

If the governing quantity is how far the damage has travelled, every application needs a reading of that depth. Respiratory support is only the reading that happened to be available in COVID-19. Ebola does not present as a lung disease. The COVID gauge does not transfer.

The Sierra Leone transcriptomic cohort supplies another one. Among 112 patients, coagulation and acute-phase transcripts gave the strongest separation between fatal and surviving cases. Fibrinogen alpha, beta and gamma were approximately 54-fold, 28-fold and 19-fold higher in fatal cases.

Coagulation is not a cell-scale quantity. Clotting is defined across tissue and circulation. A coagulation signature therefore reports that the damage is no longer local. In the terms used here, it is the same reading that a ventilator requirement gives in COVID-19. It is taken at a different organ.

The same cohort makes the complementary point. Viral load did not separate outcome where it mattered. Cycle threshold values predicted at the extremes: Ct 12 suggested death, Ct 30 survival. But between Ct 20 and 22 — around the cohort mean of 21.4 — outcome was close to even. Host-gene classifiers reached 79–92% accuracy in that same range. How much virus is present is not the quantity that decides. How far the response has travelled is.

Two further notes belong here. Metabolite markers behaved the same way: L-threonine and vitamin D-binding protein stratified outcome completely at day 8 from symptom onset, several days before death. And serial sampling of fatal cases is structurally unavailable. Patients who die do so within three to four days of admission. In one multi-omics study, eleven survivors gave 29 samples, while nine fatalities gave one each. Any usable depth marker in Ebola must therefore be readable from a single draw.

The trial this specifies

The proposal is narrow. Immunomodulation in Ebola virus disease has never been tested stratified by depth of discharge. On this reading, that is the only way it can work. Given to everyone, a treatment that harms below the line and helps above it averages toward nothing. That is what an unstratified trial would report.

The design follows from the three trials above.

1. Stratify at randomisation, not in analysis. The RECOVERY gradient is credible because the strata were fixed before allocation. A post-hoc split would not carry the same weight.
2. Use a coagulation or acute-phase reading as the stratum boundary. It stands in the role that respiratory support plays in COVID-19. Fibrinogen is the candidate the Sierra Leone data points at. It is measurable at the bedside.
3. Read the marker from a single draw at admission. Serial sampling of patients who die is not obtainable.
4. Report the strata separately and expect opposite signs. A single pooled estimate would conceal the effect this trial is designed to detect.
5. Keep supportive care as the floor in both arms. It is the only intervention currently established in this category. The framework treats it as the measure that holds the discharge at its present scale, not as a placeholder for a real treatment.

What refutes this. If immunomodulation above the coagulation threshold produces no mortality difference in Ebola, the claim fails. If the sign does not reverse below the threshold, the threshold is not real. Only ordinary severity is being measured. If a depth-graded pattern cannot be reproduced in any other condition in the fourth row, the generalisation across sepsis, ARDS and severe influenza does not hold.

What this paper does not claim. No theory derives the result. The framework supplies the question. The trials supply the answer. No mortality figure for Ebola is predicted. Nothing here is a treatment recommendation for an individual patient.

Arithmetic and verification

Every figure quoted here was recomputed from the reported counts or percentages. Every quantitative claim was taken from at least two independent pages.

Check	Recomputed	As published
CAPE COD hydrocortisone	25/400 = 6.25%	6.2%
CAPE COD placebo	47/395 = 11.90%	11.9%
CAPE COD difference	-5.65 pp	-5.6 pp (95% CI -9.6 to -1.7)
RECOVERY dexamethasone, overall	21.6 / 24.6 = 0.878 crude	OR 0.83 (0.74–0.92), age-adjusted
RECOVERY ventilated	29.0 / 40.7 = 0.713 crude	RR 0.65 (0.51–0.82), age-adjusted
RECOVERY oxygen only	21.5 / 25.0 = 0.860 crude	RR 0.80 (0.70–0.92), age-adjusted
RECOVERY no support	17.0 / 13.2 = 1.288 crude	RR 1.22 (0.93–1.61), age-adjusted
RECOVERY tocilizumab	31 / 35 = 0.886 crude	RR 0.85 (0.76–0.94)

The crude ratios sit consistently further from 1 than the published ones. The trial reports age-adjusted rate ratios. The published values are quoted in the text. The crude recomputation is shown only to confirm that the percentages and the ratios describe the same data.

The monotone series were checked directly. Control mortality across the three dexamethasone strata: 13.2, 25.0, 40.7 — strictly increasing. Published rate ratios across the same strata: 1.22, 0.80, 0.65 — strictly decreasing.

Verification of the qualitative claims. The RECOVERY subgroup figures were taken from two independent summaries. They agree on the absolute differences of +3.8, -3.5 and -11.7 percentage points. The CAPE COD figures appear identically in the trial report and in an independent trial summary. The tocilizumab entry criteria and mortality figures were confirmed against two separate sources. The confidence interval for the no-support stratum includes 1. Harm in that group is a direction, not an established finding. The text says so.

Two limitations stated plainly. The classification of conditions by cure status was checked against six entries from one chapter of the reference work, not the whole of it. And no new data were analysed for this paper. The coagulation-stratified Ebola comparison it proposes has not been run.

Annotated References

Dexamethasone in COVID-19

RECOVERY — dexamethasone, subgroup outcomes (The Bottom Line).
Used for: 28-day mortality by level of respiratory support, with rate ratios and confidence intervals. This is the primary source for the gradient table. The subgroup split was fixed at randomisation, which is why the within-trial contrast carries evidential weight.

RECOVERY — dexamethasone (First10EM).
Used for: independent confirmation of the absolute differences (+3.8, -3.5, -11.7 pp) and the overall odds ratio. Serves as the second, independent page for every dexamethasone figure, per the two-source rule stated in the paper.

Tocilizumab in COVID-19

RECOVERY — tocilizumab (REBEL EM).
Used for: the entry criteria (hypoxia plus CRP ≥ 75 mg/L) and the observation that benefit was consistent across levels of respiratory support. Supplies the “flat effect above the line” that contrasts with the dexamethasone gradient.

RECOVERY — tocilizumab (European Respiratory Society).
Used for: second confirmation of CRP ≥ 75 mg/L and hypoxia as entry criteria, and of the 31% vs 35% mortality figures with RR 0.85 (0.76–0.94).

Corticosteroids in severe community-acquired pneumonia

CAPE COD, NEJM 2023.
Used for: mortality counts (25/400 vs 47/395), the -5.6 pp difference (95% CI -9.6 to -1.7; P = 0.006), and the depth-based inclusion criteria (PSI > 130, PEEP ≥ 5, PaO₂:FiO₂ < 300, non-rebreather mask; septic shock excluded). Also noted: the trial stopped early at its second interim analysis, which tends to overstate benefit.

CAPE COD (Critical Care Reviews).
Used for: independent confirmation of the same mortality figures and criteria. Second source for the two-source rule.

Steroids for pneumonia: CAPE COD and ESCAPe (First10EM).
Used for: the ESCAPe result (16% vs 18% at 60 days; difference 2 pp; CI -8 to +5; P = 0.61), the documented between-trial differences (under-enrolment: 584 of 1,406; 96% male; mean age 69), and the commentator’s reading that the true effect lies nearer 2 points than 6.

Ebola virus disease

Liu et al., Genome Biology 2017.
Used for: the Sierra Leone transcriptomic cohort of 112 patients. Key figures: 220 of 246 differentially abundant transcripts higher in fatal cases; fibrinogen alpha/beta/gamma approximately 54-, 28- and 19-fold higher in fatal cases; cycle-threshold behaviour (Ct 12 → death, Ct 30 → survival, near-even outcome around the cohort mean of 21.4); host-gene classifier accuracy of 79–92%. This is the empirical basis for proposing coagulation markers as the depth gauge in Ebola.

Multi-platform ‘omics analysis of human Ebola virus disease pathogenesis (PMC5730472).
Used for: the serial-sampling structure (eleven survivors gave 29 samples; nine fatalities gave one each), the three-to-four-day interval from admission to death in fatal cases, and the metabolite markers L-threonine and vitamin D-binding protein, which stratified outcome completely at day 8. Grounds the requirement that any depth marker be readable from a single draw.

ECDC — Ebola disease: multi-country outbreak 2026.
Used for: the status of the 2026 outbreak in the Democratic Republic of the Congo and the species identification (Bundibugyo virus).

LSHTM — Bundibugyo outbreak: where are the vaccines?
Used for: the absence of any licensed vaccine or therapeutic for Bundibugyo virus, and the fact that licensed products target the Zaire ebolavirus glycoprotein. Establishes Ebola as RELIEF ONLY for the current outbreak.

Framework

J. Konstapel, The Vacuum.Net Theory, Ninth Edition, Constable Research, Leiden, 27 October 2026 — Foundations 9.2 and Physics 9.0.
Used for: the closure framework — a condition as a joint between two things that each hold their form and do not share an order; damage as what fails to close at one scale and moves to the next. Supplies the four operations and the logic behind the “unfilled” fourth row.

J. Konstapel, Where Illness Comes From, Completed, Constable Research, 20 October 2026.
Used for: the four cure statuses (SELF-LIMITING, CURABLE, CONTROLLABLE, RELIEF ONLY) and the layered organisation of conditions. The classification of tuberculosis, malaria, bacterial pneumonia, HIV, influenza and Ebola by cure status was checked against this work. Verified against six entries from one chapter, not the whole work.

DOI: 10.13140/RG.2.2.18291.90402

Jump to the Article here

Related: The history of marhematics

J.Konstapel,Leiden, 29-10-2026.

The Vacuum.Net Theory makes a strong claim. It says that a chain of results follows from one ground. That ground is the trinity -1, 0, +1. From it, and from five axioms, the theory builds a chain of twelve steps. Eight steps belong to Foundations. Four belong to Physics.

A theory that claims such unity must stand somewhere in history. Its proofs use earlier results. Those results have authors, dates and places of publication. The article asks two plain questions. Where does the theory stand in the history of mathematics? And does its own order match the order in which its parts were discovered?

To answer, the theory was laid beside a second text: Mathematics as a Chain of Proven Steps. That book tells the history of all 63 fields of the Mathematics Subject Classification. Each field appears as a chain of results. Each link rests on an earlier proof.

2. What the comparison found

The comparison gives a clear picture. The theory’s chain rests on 45 historical results. Each result was checked against one opened source. Each was assigned to a field. Each was then looked up in the book.

The book holds only 7 of the 45 in full. It holds 3 more in part. The other 35 are missing. The largest gap is knot theory. The theory’s first knot rests on seven results. They run from Gauss to Kauffman. The book contains none of them.

The theory draws on 18 of the 63 fields. Number theory and knot theory carry the most, six results each. Many major fields carry nothing at all. Analysis, algebraic geometry, category theory, differential geometry, fluid mechanics and statistics are absent.

3. Logic, not chronology

The most striking finding concerns order. History did not find these results in the order the theory uses them. The theory’s ground is young. Its first step rests on Spencer-Brown, Kauffman and Rosen, from 1958 to 2005. But one of its ladders rests on the Pythagorean comma, from about 300 BCE. That is the oldest result in the whole chain.

Almost every step spans centuries. The trinity joins Fibonacci (1202) to Knuth (1969). The address joins Euler (1748) to language models of 2024. The first knot joins Hamilton (1843) to Kauffman (1987).

The numbers confirm the eye. The rank correlation between step number and year is 0.16. The p-value is 0.31. That is no relation at all. The theory does not retell history. It reads it. It gathers results from 23 centuries and places them where its logic needs them.

4. Agreement where they meet

Where both texts use the same result, the facts agree. Ten results overlap. Seven appear in the book in full, three in part. The dates, authors and statements match. Only the use differs. The book asks what a result made possible later in its own field. The theory reads the same result as a step in its own chain.

One rule recurs in a telling way. The theory’s window rule says that three trits close when they are all equal or all different. Fox found the same rule for colouring knot crossings in 1956. The theory uses it first, in arithmetic mod 3. Then it uses it again for its first knot. The theory says so itself: the register and the first knot obey one rule.

5. Checks and one open date

The work did not stop at reading. Every citation was traced. Every one was found as given. For six citations the sources added history. Leibniz’s Discours de métaphysique was written in 1686 but only published in 1846. Hamilton first used the \nabla operator in 1846, seven years before the book the theory cites.

All numbers were recomputed. The continued fraction of \log_2 3, the Pythagorean comma, Watson’s return probability and more were recalculated in Python at 60 digits. Every value agrees with the theory.

One date stays open. Lagrange’s theorem on best approximations is often dated 1789. But his Additions to Euler’s Algebra appeared in 1774. The sources disagree. The question remains open.

6. What this shows

The theory stands on a narrow but solid base. Two fields dominate. Number theory supplies digits, Eisenstein integers and the approximation of \log_2 3. Knot theory supplies the trefoil and its invariants. Around them sit groups, quaternions, Euler’s partitions, Pólya’s walks and the logic of distinction. Physics adds Dirac, Maxwell and Onsager.

The chain is also economical. Most steps use one to four historical results. One step uses none at all. The step of the three elementary closures proves its own counting statements. It is the theory’s most original part.

The gaps are just as informative. The book’s Topology chapter follows a line from Euler’s formula to the Kervaire invariant. That line has no knots. Yet knots are exactly what the theory needs. And 45 of the 63 fields carry no result in the chain. Those fields mark where the theory’s reading has still to be tried.

7. Conclusion

The Vacuum.Net Theory does not invent its own history. It stands on 45 verifiable results. Its citations check out. Its numbers recompute. Its logic, not chronology, gives it order. Laid beside a full history of mathematics, it agrees wherever the two meet. Where they do not meet, the comparison shows precisely which chapters of history the theory still has to open.

---

Annotated reference list

Each entry names the source and gives a short note on its role.

The two texts

1. Konstapel, J. (2026a). Mathematics as a Chain of Proven Steps: The 63 Fields of Mathematics, from the First Proofs to 2026. Constable Research, Leiden. Published on ResearchGate.
   Annotation: The history against which the theory is read. It tells all 63 MSC2020 fields as chains of proven results, each checked against opened sources.
2. Konstapel, J. (2026b). The Vacuum.Net Theory, Ninth Edition: Foundations, version 9.2, and Physics, version 9.0. Constable Research, Leiden, 27 October 2026.
   Annotation: The theory under examination. It derives a twelve-step chain from the trinity -1, 0, +1 and five axioms. This essay is based on the article that compared the two texts.

Sources for the historical results

3. Andrews, G. (2007). “Euler’s ‘De Partitio Numerorum’.” Bulletin of the AMS 44.
   Annotation: Anchors balanced ternary in history. Traces the idea to Fibonacci’s weights problem of 1202 and places Euler’s partition identity in the Introductio of 1748.
4. Roth, R. (2001). “A History of Lagrange’s Theorem on Groups.” Mathematics Magazine 74.
   Annotation: Source for the chain of group theory, from Lagrange (1770–71) through Cauchy and Jordan to Cayley’s abstract group of 1854.
5. Kauffman, L. and Varela, F. (1980). “Form dynamics.” Journal of Social and Biological Structures 3, 171–206.
   Annotation: Grounds the theory’s reading of re-entry as an oscillation. The alternation [+1, -1] squares to -1.
6. Kauffman, L. (2005). “EigenForm.” Kybernetes 34, 129–150.
   Annotation: Defines the eigenform condition O(A) = A. The theory reads its closure rule N2 and its fixed rings as eigenforms.
7. Spencer-Brown, G. (1969). Laws of Form. London: Allen & Unwin.
   Annotation: The calculus of distinctions that opens the theory. Its two laws and the notion of re-entry ground the first step of the chain.
8. Rosen, R. (1958; 1991). Relational biology, Bulletin of Mathematical Biophysics 20; Life Itself.
   Annotation: Source of the theory’s open question IX.2: which rewrite dynamics is its own fixed point.
9. Hayes, B. (2001). “Third Base.” American Scientist 89.
   Annotation: On radix economy and balanced ternary. Names the 1950 book High-Speed Computing Devices as the earliest use of the cost measure.
10. Przytycki, J. (2018). On Fox colourings. arXiv 1809.00063.
    Annotation: Establishes that Fox introduced three-colourings in lectures at Haverford College in 1956. The rule returns in the theory’s window theorem T4 and its first knot.
11. Encyclopedia of Mathematics. Entries on the Reidemeister theorem, Kronecker’s theorem, Burnside’s lemma, and the Hopf invariant.
    Annotation: Standard reference works for four results: Reidemeister’s three moves (1927), Kronecker’s density theorem (1884), Burnside’s lemma, and Hopf’s 1931 result \pi_3(S^2) = \mathbb{Z}.
12. Jones, V. (1985). “A polynomial invariant for knots via von Neumann algebras.” Bulletin of the AMS 12, 103–111.
    Annotation: The Jones polynomial. With the Kauffman bracket it shows that the trefoil differs from its mirror image.
13. Kauffman, L. (1987). “State models and the Jones polynomial.” Topology 26, 395–407.
    Annotation: The bracket polynomial as a state sum. The theory computes it for the trefoil over its eight smoothings.
14. Lickorish, W. B. R. and Millett, K. (1986). “Some evaluations of link polynomials.” Commentarii Mathematici Helvetici 61.
    Annotation: Connects three-colourings to the Jones polynomial. Gives the count of 9 colourings for the trefoil.
15. Trinity College Dublin. Letters on the discovery of the quaternions, Broome Bridge, 1843.
    Annotation: Documents Hamilton’s discovery of 16 October 1843. Unit quaternions double-cover the rotations, which grounds the theory’s 4\pi return.
16. Newman, M. H. A. (1942). Journal of the London Mathematical Society 17.
    Annotation: Gives a proof of Dirac’s belt trick. A 2\pi turn of an attached strand cannot be undone; a 4\pi turn can.
17. Dirac, P. A. M. (1928). “The quantum theory of the electron.” Proceedings of the Royal Society A 117, 610–624.
    Annotation: The input of the theory’s step on motion. The theory rewrites it in Rowlands’ nilpotent form, Q^2 = 0 on the mass shell.
18. Rowlands, P. (2007). Zero to Infinity: The Foundations of Physics. World Scientific.
    Annotation: Source of the nilpotent rewriting of the Dirac operator. The theory reads nilpotence as closure.
19. Maxwell, J. C. (1873). A Treatise on Electricity and Magnetism, vol. 2, Articles 591–619.
    Annotation: Holds the grouped quaternion form of the electromagnetic equations. The theory reads them as one closure operation with two readings.
20. MacTutor History of Mathematics. Entry on Oliver Heaviside.
    Annotation: Source for the Gibbs–Heaviside split of the quaternion product into dot and cross, 1881–1885.
21. Onsager, L. (1931). “Reciprocal relations in irreversible processes.” Physical Review 37, 405–426.
    Annotation: The symmetric coefficients of irreversible processes. The theory reads every loss as having an address on the net.
22. Milgrom, M. (1983). “A modification of the Newtonian dynamics.” Astrophysical Journal 270, 365–370.
    Annotation: Introduces the acceleration scale a_0. The theory identifies it with cH_0/2\pi.
23. Berengut, J. et al. (2013). Physical Review Letters 111, 010801.
    Annotation: On the fine-structure constant in white-dwarf spectra. Grounds the theory’s prediction for G191-B2B.
24. Seiberg, N. and Witten, E. (1994). Nuclear Physics B 426, 19–52.
    Annotation: The example in the theory’s references of physics carried by a closed curve.
25. Ma, S. et al. (2024). Ternary-weight language models. arXiv 2402.17764.
    Annotation: Recent machine-learning work on the same three weights -1, 0, +1. It closes the chain’s span of 23 centuries at the modern end.
26. Khinchin, A. (1935–36). Continued Fractions.
    Annotation: The theory’s citation for best approximations and convergents. Behind it lies the one open date in the whole comparison: Lagrange’s work is dated 1774 in one source and 1789 in another.

Your DOI: 10.13140/RG.2.2.28148.51845

Jump to the article here

J.Konstapel,Leiden,29-10-2026.

Biology has a habit it cannot shake. It keeps asking what life is.

It keeps asking where chemistry ends and where life begins.

The question has never produced an answer.

It has only produced a grey zone.

Viruses, giant viruses, reduced endosymbionts, and organelles all sit in that zone.

Each new discovery seems to blur the line further.

The Vacuum.Net Theory refuses the question. It does not ask whether a thing is alive. It asks something else. Every whole, on this view, is a closure. And every closure has two properties that matter. Does it close? And what does it pass on to the scale above it?

The second property is called the carry. This is the pivot of the whole reading. Two wholes can both be closed. Yet they can be wholly unlike in what they do to the next scale. A ring carries. A meeting carries nothing. Read this way, bacteria and viruses do not differ in whether they are alive. They differ in carry.

This essay works through that reading in seven steps. First, the ground of the theory. Second, the three closures and their carries. Third, bacteria and viruses as kinds of closure. Fourth, the grey zone as a mixture, not a boundary. Fifth, seams and what they say about infection. Sixth, the ladder and its arithmetic. Seventh, what the reading yields, and what it does not. A full annotated reference list follows at the end.

2. The Ground: One Strand, Three Values, Closed Windows

The theory begins with a single act. There is one strand. A distinction is the strand returning to a place it has been (N1). Nothing is distinguished before that return. No inside. No outside. No object. What we ordinarily call a thing is a loop the strand makes on itself.

Beneath this lies a trinity of values: −1, 0, and +1. Their sum is zero. Read as the digits of balanced ternary, they add by a digit rule. A sum keeps its residue modulo 3 at its own scale. It passes the rest on as a carry to the next scale (T1).

The central selection rule concerns windows. A window is three consecutive trits. A loop persists if and only if it lies in a closed window. A window closes when its sum vanishes modulo 3. Of the 27 possible windows, exactly 9 close (T2, T3). Whatever does not close passes on as tension. Closure itself is the selection rule. There is no external selector (N2).

Two further axioms do real work in the reading that follows. First, the passages at every node are ordered. The first passage fixes a direction. Later passages are taken relative to it. The net therefore carries its history as its shape (N3). Second, merged blocks obey the same rules again. What does not close at one scale is the material of the next (N5).

That is the entire apparatus. It is worth stating plainly: no notion of life enters it. No notion of organism, function, or purpose enters it either. Everything below is built from strand, window, closure, carry, and scale.

3. Three Closures and What Each Carries

Every closed trit word reduces completely to three elementary closures. There are exactly three, and there is no fourth (T7):

1. A standing — a single 0, of length one.
2. A meeting — a pair x, −x, of length two.
3. A ring — a triple x, x, x, of length three.

No open word reduces at all (T7). The important difference, however, is not the length. It is the carry. A standing closes with carry 0. A meeting closes with carry 0. Nothing passes upward from either. A ring is different. Because 1 + 1 + 1 = 10 and T + T + T = T0 in balanced ternary, a ring closes with carry ±1. A ring therefore closes at its own scale and becomes one passage of the next scale. This is N5 restated as arithmetic (T7.1).

Closure	Length	Carry	Consequence
Standing	1	0	Persists; contributes nothing upward
Meeting	2	0	Persists only as a pair; contributes nothing upward
Ring	3	±1	Persists and becomes one passage of the next scale

That single arithmetic fact carries the whole reading. One more result sharpens it. A word that closes in every window does something special. It repeats one closed window. For every length there are exactly nine such words (T8). These are the fixed points of the register’s rewrite (T10). They are eigenforms: forms that reproduce themselves under their own operation. Reproduction, in this vocabulary, is not an added property of certain matter. It is a fixed point of closure.

4. Bacteria and Viruses as Kinds of Closure

Now the reading itself. A free-living bacterium is a ring. It closes on its own account. It repeats its own closed window. And its carry passes upward. It becomes one passage of a colony. Of a biofilm. Of a mat. In the endosymbiotic case, it becomes a passage of a cell.

A virus is a meeting. It closes only together with the ring of something else. The biological record says the same thing in its own words. A viral genome carries no information for generating metabolic energy. It carries no information for protein synthesis. It depends on the host cell for both. On its own it contributes nothing to the scale above.

Note what these two statements do not say. They do not say that a bacterium is alive and a virus is not. The theory draws no line of life. It draws a distinction of carry.

4.1 The Cases Between

The interesting cases sit between the two, and biology has found plenty of them.

Giant viruses. The mimiviruses were the first shock. Then came the tupanviruses, the most extreme case. A tupanvirus carries up to 70 tRNAs, 20 aminoacyl-tRNA synthetases, and 11 translation factors. This is the largest translational apparatus known in the virosphere. The translation machinery was long held to be exactly what separates cells from viruses. Part of that machinery is now carried inside a virus. But only part. The rest is still borrowed. In the language of the theory: part of the ring is carried internally. The virus is a meeting that has begun to internalise its partner.

Reduced endosymbionts. The same ground is approached from the other side. Candidatus Carsonella ruddii lives inside the psyllid Pachypsylla venusta. Its genome is 159,662 base pairs and encodes 182 genes. Escherichia coli carries roughly 4,000. Carsonella’s gene set is no longer sufficient for DNA replication, transcription, or protein synthesis. The original authors conclude that it can no longer be considered an independent organism. The story continues. In a related psyllid, a symbiont called Psyllophila completes a tryptophan pathway that Carsonella leaves unfinished. Neither symbiont closes alone. Each supplies what the other leaves open.

In the language used here: Carsonella has outsourced most of its ring. It now closes only as a meeting.

5. Not a Boundary but a Mixture

Biology keeps asking where chemistry ends and life begins. On this reading, the question is malformed. There is no property of being alive that a thing either has or lacks. There is only closure, and what that closure passes on.

Nothing in the theory itself is vague. A window closes or it does not (T2). A closure is a standing, a meeting, or a ring, and nothing else (T7). What varies is something different. It is the proportion in which a given whole is built from meetings and from rings. That proportion determines how much carry the whole sends upward. A virus is nearly all meeting. A bacterium is nearly all ring. A tupanvirus is a meeting with a partial ring. Carsonella is a ring that has become a meeting. No line exists between them. A gradient does.

The second source of apparent vagueness is scale. What fails to close at one scale may close at the next (N5). Carsonella does not close alone. Carsonella together with its insect does. Whether something qualifies as a whole depends on the scale at which the question is asked. The theory makes that dependence explicit. It turns an embarrassment into a rule.

The biological record agrees, in its own vocabulary. Carsonella is described as standing at the frontier between cell and organelle. The giant viruses are said to have blurred the roots of the tree of life. Two independent groups of researchers, working with components, described what the carry rule describes as a proportion.

6. Seams: Closure Is Not Enough

When two closed windows meet in one word, the joint between them is a seam. The seam closes if and only if the second window is the same as the first. Count over all 81 ordered pairs. There are 9 closed seams, 36 with one open window, and 36 with two (T11).

The condition is sharper than it first appears. Both parties being closed in themselves does nothing for the joint. Of the nine closed windows available to the second party, exactly one leaves no tension. Order must be shared, not merely possessed.

This reframes symbiosis and infection without any appeal to aggression or defence. Gut flora and mitochondria are not harmless guests that happen to be tolerated. After a long common history, they run in the same order as their host. The seam closes. A novel pathogen is not malicious. It simply brings a different order. The seam opens. The difference lies in the seam, not in a property of either party. A reservoir host and a spillover host meet the same virus. They do not meet the same seam.

The record offers a third way to absorb the tension: not by matching order, but by adding a loop. The virophage mavirus integrates into the genome of the flagellate Cafeteria. When the giant virus CroV infects that host, mavirus is reactivated. It suppresses CroV production. The infected cell still dies. But the surrounding population survives. Closure is not restored at the scale of the cell. It is restored one scale up.

7. The Ladder Is a Count, Not a Size

Merged into blocks, a coarse passage contains exactly 2^a · 3^b passages of the finest scale, after a merges of meetings and b merges of rings. On the ring ladder, each step multiplies by three (T15). The ladder therefore counts passages. It does not measure length. Reading it as a length is the first mistake to avoid.

Take the atom as the finest passage, and average atomic mass as 2 × 10⁻²⁶ kg. The depths then come out as follows. The masses are order-of-magnitude figures, so each depth carries roughly half a step of uncertainty.

	Mass (kg)	Atoms	Ring steps (log₃)
Protein	1 × 10⁻²²	5 × 10³	7.8
Ribosome	4 × 10⁻²¹	2 × 10⁵	11.1
Ebola virion	1 × 10⁻¹⁹	5 × 10⁶	14.0
Mimivirus	1 × 10⁻¹⁷	5 × 10⁸	18.2
E. coli	1 × 10⁻¹⁵	5 × 10¹⁰	22.4
Human cell	1 × 10⁻¹²	5 × 10¹³	28.7
Human body	7 × 10¹	3.5 × 10²⁷	57.7

The choice of the atom as the finest passage is a choice, not a result. It shifts every depth by the same constant, so the absolute numbers say little. The differences do not depend on it:

• virion to bacterium: 8.4 steps
• bacterium to human cell: 6.3 steps
• human cell to human body: 29 steps

The last figure deserves a pause. The distance from a single cell to a whole body is larger than the entire distance from an atom to a cell. On this ladder, most of the work is not in building a cell. It is in stacking them.

One observation is offered as nothing more than an observation. Of the depths in the table, only the ribosome at 11.1 falls near a closure depth (12). The rest do not. That may be coincidence. The order-of-magnitude masses are too coarse to decide it.

8. Two Ladders That Never Meet

Meetings merge by two. Rings merge by three. And the two ladders never coincide: 2ⁿ ≠ 3ᵏ for all integers other than n = k = 0 (T16, proved). A whole built from both kinds of closure therefore closes exactly at no depth whatever. Something is always left over.

The miss is small at particular depths. The closure depths are the denominators of the convergents of log₂ 3: 1, 2, 5, 12, 41, 53, 306, 665, 15601 (T17). At depth 12 the residue is 1.35 per cent — the Pythagorean comma, familiar from music. At 53 it is 0.21 per cent. At 306 it is 0.10 per cent (T18). It never reaches zero.

The consequence for a living whole is direct. Any such whole mixes closures that carry with closures that do not. So there is a permanent residue. It is not produced by a pathogen. Not by a mutation. Not by a failure. It follows from arithmetic. And by N5, that residue is exactly what constitutes the next scale. Without a residue, there is no scale above.

9. Where the Residue Discharges

A carry moves one scale up, never sideways (T7.1). What does not close in a cell appears at the level of the tissue. What does not close there appears at the level of the body. The residue can only rise until it finds a scale at which it closes.

This changes the central question of medicine. Medicine asks what disturbs the equilibrium. On this reading there is no equilibrium to disturb. The residue is permanent. The question is: at what depth does it discharge? Locally, within the tissue? Across the whole organism? Or, as in the virophage case, only at the level of the population, with the individual lost?

The question also stops being about the identity of an invader. Two seams with the same participants can differ entirely. The seam depends on shared order, not on what the parties are.

What the reading does not yield should be said just as plainly. It gives no case fatality rate. No prediction of who recovers. No therapy. It is a different bookkeeping of the same facts. Its value lies in which questions it makes askable.

10. Epistemic Status: What Stands Where

The Foundations document assigns every statement one of five statuses. This essay keeps that discipline.

Element	Standing
The trinity, N1–N5	Assumed
T1, T2, T7.1, T12, T16, T17, T18	Proved — arithmetic
T7, T8, T10, T11, T15	Conditionally proved, under their stated hypotheses
Ring = free-living organism; meeting = virus	Read into the theory here; carries no status in the document
What the places and passages of a living body are	Open (IX.5)

The third row is the load-bearing one. Foundations makes no physical claim. The physical reading of the trit rule is listed there as an open question. The identification of biological kinds with kinds of closure is therefore a reading. It is offered because it is coherent. And because it reorganises the facts. Not because the theory entails it.

Two numerical checks were run for the original article rather than taken from the document. First: in a word that closes everywhere, altering a single trit opens exactly three windows. This holds for all nine such words, at every position. Second: with trits drawn uniformly and independently, a window closes with probability 1/3 (0.3341 over 300,000 draws). Trits derived from a real sequence are correlated and close more often. Both checks agree with T2 and T6.

11. Conclusion

The reading offered here is austere. It replaces one unanswerable question — what is life? — with several answerable ones. Is this whole closed? What does it carry upward? Does its seam with that other whole close? At which scale does the residue discharge?

The biological record supplies the witnesses. Giant viruses internalise part of a ring. Carsonella outsources nearly all of its own. The virophage closes at the population what the cell could not. None of these facts is new. What is new is the bookkeeping. A ring carries. A meeting carries nothing. The ladder counts passages, not lengths. And the residue that no living whole can avoid is not a flaw in it. It is the reason there is a scale above it at all.

---

Annotated Reference List

The Theory

Konstapel, J. — The Vacuum.Net Theory, Ninth Edition. Constable Research, Leiden, 27 October 2026 (Foundations 9.2 and Physics 9.0).
The source of all axioms and theorems cited in this essay (N1–N5, T1–T18). Foundations is the formal document; every statement in it carries one of five statuses, from “assumed” to “open”. The biological reading developed above is not part of this document. It is an interpretation offered on top of it. The document itself makes no physical claim.

Viruses and the Host Cell

Ratti et al., Viruses — reported via the University of Bologna.
The empirical basis for reading a virus as a meeting: viral genomes lack the information for ATP generation and for protein synthesis. Both capacities must come from the host. This is the biological counterpart of “carry 0”: the viral closure contributes nothing upward on its own account.

“General Properties of Viruses”, lecture notes, King Saud University.
A standard summary of viral properties: obligate intracellular parasitism and the absence of ribosomes. Used to corroborate the same point from an independent educational source.

Giant Viruses

Abrahão, J. et al. (2018), Nature Communications — tupanviruses.
Describes the most translation-rich viruses known: up to 70 tRNAs, 20 aminoacyl-tRNA synthetases, and 11 translation factors. The key witness for a meeting that has internalised part of a ring. In the essay’s terms, tupanvirus is a borderline case in which the proportion of ring has shifted — but not far enough to close alone.

Silva, L. C. F. et al. (2017), PLoS Genetics.
Documents translation-related genes in mimiviruses and argues that giant viruses blur the roots of the tree of life. Supplies the biological wording (“blurred”) for what the theory describes as a proportion.

Reduced Endosymbionts

Tamames, J. et al. (2007), BMC Evolutionary Biology.
The genome analysis of Candidatus Carsonella ruddii: 159,662 base pairs, 182 open reading frames, against roughly 4,000 genes in E. coli. The authors place it at the frontier between cell and organelle. In the reading, this is a ring that has outsourced itself into a meeting.

Nakabachi, A. et al. (2006), Science — reported via ScienceDaily.
The original sequencing of Carsonella, then the smallest known cellular genome. Provides the primary measurement on which Tamames et al. build.

mSystems (2023) — the Psyllophila / tryptophan pathway.
Shows complementary genome reduction in psyllid symbionts: one partner completes the tryptophan pathway the other leaves unfinished. Neither symbiont closes alone. Evidence for the essay’s claim that closure — and organismhood — can be a property of a pair at a higher scale.

Virophages

Fischer, M. G. & Hackl, T. — Max Planck Institute for Medical Research.
The original work on mavirus, the virophage that integrates into the genome of the flagellate Cafeteria and is reactivated by infection with the giant virus CroV. Mavirus suppresses CroV production. The infected cell dies; the population survives. The essay’s central example of closure restored one scale up, at the cost of the individual.

Racaniello, V. — Virology Blog.
An independent account of the same mavirus research. Used to corroborate the primary result rather than as a source of new facts.

Closure to Efficient Causation

Siekmann, H.-G. (2017), Ecological Complexity — Rosen’s (M,R) systems.
Robert Rosen defined an organism as a system closed to efficient causation. Siekmann argues that this closure is a postulate to be tested, not a definition to be assumed. This is the closest prior use of “closure” as a criterion of life. The Vacuum.Net reading differs from Rosen’s in one respect: it distinguishes closures that carry from closures that do not. On Rosen’s criterion, a virus and a bacterium would both count or neither would. On the carry criterion, they come apart.

Your DOI: 10.13140/RG.2.2.13887.88481

The 63 fields of mathematics, from the first proofs to 2026

J.Konstapel,Leiden,29-11-2026.

Some forty-four thousand years ago, in a cave near what is now the border of South Africa and Eswatini, someone cut a series of notches into a piece of baboon bone. We do not know their name. We do not know what they counted — days, kills, members of a band, phases of a moon. We know only this: they made a mark, and the mark stood for something else.

That bone, the Lebombo bone, may be the oldest surviving artifact of mathematical thought. It contains no formula, no theorem, no proof. What it contains is something arguably more important: the first evidence of the human mind grasping that the world can be counted — that a flock, a season, a debt, a life can be matched, one-to-one, against a row of marks. Every equation ever written since, from Babylonian clay to machine-verified proof assistants, rests on that single revolutionary act of abstraction. A number is not a thing; it is what things have in common.

This essay tells the story of mathematics as that story must be told: as one long chain. Not a chain of geniuses, though geniuses appear. Not a chain of cultures, though cultures matter enormously. A chain of proven steps — each discovery resting on the ones before it, each proof opening the door to the next. To read mathematics this way is to see something both humbling and inspiring: that the tallest structure of certainty our species has ever built was assembled link by link, by ordinary hands, over fifty millennia — and that the chain is still being forged, right now, by people who were born after you were.

Come, then, and walk the chain.

---

II. The First Act: Number and Shape (c. 44,000 BCE – 500 CE)

The Babylonians and the place-value miracle

Around 2000 BCE, in the river valleys of Mesopotamia, scribes did something with numbers that no surviving culture had done before: they made a digit’s position carry meaning. In their base-sixty system, the same wedge shapes meant 1, or 60, or 3600, depending on where they stood. This invention — place value — is so familiar to us that we forget it is an invention at all. It is the reason a child can write a trillion with twelve strokes, and the reason a Babylonian tablet, YBC 7289, could record √2 to six decimal places three and a half millennia before anyone in Greece cared whether the number even existed.

And there is the first crack in the edifice. Place value needed a sign for nothing — for the empty position that distinguishes 3601 from 61. For over a thousand years the Babylonians simply left a gap. Around 400 BCE they began to press two wedges into the clay as a placeholder. It was not zero as a number. It was zero as a space. But the chain had its first recorded repair: a gap in the notation, and a symbol to fill it.

The Greek catastrophe: the discovery that numbers are not enough

Then came the crisis that made mathematics mathematics.

Sometime before 400 BCE, a Greek — tradition says a Pythagorean — proved something devastating. Suppose √2 can be written as a ratio of whole numbers, a/b, in lowest terms. Then a² = 2b², so a is even; write a = 2c and you find b² = 2c², so b is even too. Both even — but we said lowest terms. Contradiction. Ten lines of argument, and a worldview lay in ruins.

The Pythagoreans had taught that everything is number — that the whole cosmos, music included, was built from whole numbers and their ratios. The proof that √2 is irrational — incommensurable, literally “unmeasurable together” — destroyed that. Legend says the discoverer was drowned for it. Whether or not the drowning happened, something more important did: the Greeks responded to the crisis by inventing rigor. This is the moment mathematics became a chain of proven steps rather than a collection of recipes. A contradiction was found; and instead of abandoning the subject, the Greeks built a new foundation deep enough to hold it.

Eudoxus of Cnidus, in the fourth century BCE, did exactly that. His theory of proportion — preserved in Book V of Euclid’s Elements — defined when two ratios are equal without ever mentioning irrational numbers: a:b = c:d when, for all whole numbers m and n, ma > nb exactly when mc > nd. It is a definition of almost unbearable subtlety, and it would take twenty-three centuries — until Dedekind in 1872 — for humanity to say precisely what Eudoxus had been gesturing at. The chain sometimes waits a very long time between links.

Euclid and the standard of proof

Around 300 BCE, Euclid of Alexandria wrote down what a chain of proven steps is. The Elements takes a handful of definitions, postulates, and common notions — and derives 465 propositions from them, each resting only on what came before. The book is not beautiful because of its content; much of it was older than Euclid. It is beautiful because of its form. For the first time, knowledge was organized as a single connected structure, every stone bearing weight, nothing floating.

For two thousand years, to say something was “proved more geometrico” — in the manner of the geometers — was the highest compliment knowledge could receive. Euclid gave our species the very idea of a watertight argument, and that idea escaped mathematics entirely: it became the backbone of theology, of law, of physics, of every discipline that claims to know anything. Spinoza wrote his Ethics in the form of the Elements. Newton wrote his Principia the same way. The American Declaration of Independence is a chain of proven steps — “we hold these truths” — leading to a conclusion. When humans argue well, they are still doing what Euclid did.

Alexandria and Archimedes

In the third century BCE, Archimedes of Syracuse — perhaps the greatest mathematician who ever lived — found the area of a curved region by cutting it into infinitely many slices, in effect integrating two thousand years before calculus. His “Method” reads like a letter from the future. And in Alexandria, Eratosthenes measured the circumference of the Earth with a stick, a shadow, and a piece of pure reasoning — and got it right to within a few percent. Mathematics, by this point, was no longer about counting sheep. It was about measuring the planet.

India and the number zero

In 628 CE, Brahmagupta of Bhaskara’s Ujjain wrote the Brahmasphuṭasiddhānta and did something no one had ever done: he treated zero as a number in its own right, defining it as a − a, and gave rules for arithmetic with it — and with “debts and fortunes,” the negatives. “Fortune times debt is debt,” he wrote; “fortunes cut to nothing” — zero — “become nothing, yet a debt cut to nothing remains nothing too.” Some of his rules were wrong (he set 0/0 = 0), but the idea was seismic. A symbol for the absence of things had become a thing.

There is no evidence — and honest history must say so — that this Indian achievement flowed directly to the Islamic world and then to Europe as a single stream. The chain has long stretches where the links are missing, where cultures worked in parallel, each repairing the number system in their own tongue. What matters is that humanity got there: by the time the chain reached medieval Europe, it carried place value, zero, and the negative numbers — the complete decimal system a Dutch engineer would soon make universal.

---

III. The Second Act: Algebra, the Art of the Unknown (700–1600)

Al-Khwarizmi and the house of wisdom

In ninth-century Baghdad, in the intellectual ferment of the Abbasid caliphate, Muhammad ibn Musa al-Khwarizmi wrote a book with a deceptively humble title: al-Kitab al-mukhtasar fi hisab al-jabr wa-l-muqabala — “The Compendious Book on Calculation by Completion and Balancing.” The word al-jabr — restoration, the moving of a damaged quantity to the other side of an equation — passed into every European language as algebra. His own name, Latinized, passed into them too: algorithm.

Al-Khwarizmi’s equations had no symbols. He solved them in prose, with squares and roots and numbers balancing like goods on a scale. But the idea of the unknown — of posing a question about a quantity you do not yet have and reasoning your way to it — is one of the great liberations of the human mind. The baby, the inheritance, the field, the trajectory of a cannonball: all can be written as x, and what is true of x is true of them all at once.

The cubic, the duel, and the birth of the impossible

The story of the cubic formula is the wildest tale in mathematics. In the 1530s, in Bologna, Scipione del Ferro solved the cubic and told almost nobody — professorships were won by duels of problem-solving in those days, and a secret method was a secret weapon. He confided it to his student Fior on his deathbed. Fior then challenged the Venetian arithmetician Niccolò Tartaglia to a contest; Tartaglia, hearing rumors that cubics could be solved, worked out the method himself, and demolished him. Gerolamo Cardano, the greatest physician in Europe and a mathematician of unstoppable curiosity, begged Tartaglia for the secret — swore an oath — extracted it in an anagram, and then discovered in del Ferro’s surviving papers that the solution predated Tartaglia. Considering his oath thereby void, Cardano published it in his 1545 Ars Magna, with full credit to all three and, in an appendix, a solution of the quartic by his student Ludovico Ferrari that no one had touched since. Tartaglia never forgave him. One of mathematics’ greatest results entered the public record because of a priority quarrel.

Then came the miracle inside the miracle. Apply Cardano’s formula to a cubic with three real roots, and it forces you through the square roots of negative numbers. Cardano, brave to the point of recklessness, wrote to Tartaglia-like colleagues about “imaginary” quantities and asked what they could possibly mean. In 1572, Rafael Bombelli, an engineer who had drained the marshes of the Pontine plains, published his L’Algebra and did the bravest thing in the history of the subject: he laid down rules of arithmetic for the impossible — defining +i and −i, and showing that if you just keep calculating with these meaningless objects, the imaginary parts cancel at the end and out pop the real, correct roots. √−1, an absurdity by every existing definition, turned out to be the necessary path to the true. The lesson — that mathematics is free to reason about what it does not yet understand, and that meaning can follow existence rather than precede it — has never been more sharply stated.

Stevin and the democratization of number

In 1585, Simon Stevin of Bruges — engineer, fortress-builder, author of a famous treatise on decimal fractions — published De Thiende and argued, with a polemical intensity that still crackles off the page, that decimal fractions would free arithmetic from the tyranny of awkward fractions for everyone, merchant and astronomer alike. And in his Arithmétique he made a claim of enormous philosophical weight: that numbers are not merely counting-beads, that irrational and negative quantities are numbers like any other, “aucun nombres qui soient absurdes ou irréguliers” — no absurd numbers exist. The Greek wall between magnitude (the realm of continuous geometry) and number (the realm of the countable) had stood for two thousand years. A practical engineer from the Low Countries knocked it over, and every measurement made since — every price tag, every decimal expansion on every calculator screen — walks through the gap he opened.

---

IV. The Third Act: The Invention of the Infinite (1600–1700)

Descartes: the bridge

In 1637, René Descartes, in an appendix to a book on optics — the Discours de la méthode, and specifically its third appendix, La Géométrie — fused algebra and geometry forever. Assign every point of the plane a pair of numbers, every curve an equation, and suddenly the entire ancient apparatus of geometry — of Euclid, Apollonius, Pappus — becomes computable. The Greeks’ deepest results about conic sections become exercises in completing the square. And more than that: the bridge runs both ways. Equations acquire pictures; intuition acquires proof. Every graph you have ever seen, every animation on every screen, is Descartes’ coordinate system redrawing the world.

Newton and Leibniz: the machinery of change

And then, in the 1660s and 1670s, the two greatest intellects of their century, working in complete ignorance of each other, built calculus.

In his plague years at Woolsthorpe, Isaac Newton — twenty-three, sent home from Cambridge as the Great Plague emptied it — invented “fluxions”: rates of change, quantities that flow, and with them the laws of motion and universal gravitation that bind the solar system to a falling apple. In Paris, a decade later, Gottfried Wilhelm Leibniz — diplomat, philosopher, historian of the Guelfs, the last man who genuinely knew everything known to his century — invented the same machinery, and invented, crucially, the notation we still use: the ∫, the dx, the d/dx. Leibniz’s symbolism was so good that it thought for you — generations of physicists would report that their discoveries felt like following the notation where it wanted to go.

The bitter priority dispute between them poisoned the mathematical cultures of England and the Continent for a century. But the mathematics did not care. Calculus worked — and worked so well that people stopped asking what, exactly, an “infinitely small” quantity was. This was the gap in the chain that would take two hundred years to repair. In the meantime, humanity used the unexamined tool to build celestial mechanics, to predict the return of Halley’s comet, to navigate the oceans, and — in time — to fly. Mathematics has often been like that: engine first, foundations later.

Probability: the mathematics of the unknown

The same century invented something equally audacious: the mathematics of uncertainty. In 1654, Blaise Pascal and Pierre de Fermat exchanged letters over a gambling problem — how to divide the stakes of an unfinished game — and in those letters, probability theory was born. Then in 1713, Jacob Bernoulli’s posthumous Ars Conjectandi proved the law of large numbers: that in the long run, frequencies stabilize toward an underlying chance, so that chance itself becomes knowable. It is one of the strangest and most wonderful facts in this whole story: the mathematics of randomness is deterministic. Insurance, epidemiology, polling, quantum theory, machine learning — every statistical inference ever made descends from a gambler’s interrupted game.

---

V. The Fourth Act: The Tower Grows (1700–1900)

Euler: the master of them all

Leonhard Euler, the most prolific mathematician in history, wrote mathematics the way rivers flow — over eight hundred papers and books, half of them composed after he went blind in 1766, dictating to assistants from memory. He gave mathematics its modern language, standardizing the notation of functions, of e, of π, of i, of Σ. His Introductio in analysin infinitorum (1748) is arguably the most influential textbook ever written. His identity — e^iπ + 1 = 0, binding together the five most important constants of mathematics into a single sentence — is often called the most beautiful equation in existence. Euler’s true gift was not one theorem but a style: the fearless use of the infinite, the general, and the symbolic to reveal patterns no one had imagined were there.

Gauss: mathematics’ rigid standard

Carl Friedrich Gauss, “the prince of mathematicians,” was harder to impress. His 1799 doctoral thesis — the first valid proof of the fundamental theorem of algebra, repairing the gaps that d’Alembert and Euler had left — set a new standard of rigor. His 1801 Disquisitiones Arithmeticae, published at twenty-four, founded modern number theory as an organized discipline. That same year, astronomers lost the asteroid Ceres and Gauss, inventing a new method of error estimation on the spot (the method of least squares), told them where to point their telescopes. It was there. The story made Gauss famous across Europe, and it made something else famous too: the idea that mathematics is not merely description — it is prediction.

The century of abstraction begins

The nineteenth century saw the tower of mathematics climb higher and stranger than anyone had imagined, and it began — characteristically — with a crisis. For two thousand years, everyone had known Euclid’s parallel postulate was true, merely awkward, and generations had tried to prove it from the rest. In the 1820s–1830s, Nikolai Lobachevsky in Kazan and János Bolyai in Transylvania (whose father, writing to him the most heartbroken sentence in the history of mathematics, warned: “For God’s sake, please give it up… it may take all your time…”) independently built consistent geometries in which the postulate is false. Bernhard Riemann, in a single 1854 habilitation lecture, generalized the idea beyond all imagining: geometry can describe any curved space of any number of dimensions. The implications did not arrive for sixty years, and when they did, they arrived as general relativity — the discovery that gravity is geometry, that we live inside a curved four-dimensional spacetime, and that our cosmos obeys the mathematics of a lecture given by a shy German before a nearly-empty hall in Göttingen.

Meanwhile, in Cambridge, George Boole (1847, 1854) showed that logic itself — the laws of valid thought, from Aristotle on — could be written as algebra: variables that take only the values 0 and 1, AND, OR, NOT. He hoped to codify the mind of God; what he built, ninety years later, was every logic gate in every microchip on Earth. And in Göttingen, Bernhard Riemann’s friend Richard Dedekind and others began to ask what, exactly, a number is.

---

VI. The Fifth Act: Foundations — the Deepest Repairs (1872–1940)

What is a real number? What is a set?

In 1872, Richard Dedekind and Georg Cantor, independently, built the real numbers out of nothing but the rationals — Dedekind with his “cuts,” Cantor with convergent sequences of fractions. The infinite decimals of Stevin finally had an exact foundation. It is a perfect example of the chain of repairs: Babylon’s approximation of √2 → the Greek discovery of the gap → Eudoxus’s laborious proportion → Stevin’s practical decimals → and now, at last, a definition of what the continuum is, precise enough to prove things about.

Cantor then went where no one had gone before: he counted the infinite. Two sets have the same size if their members can be paired one-to-one — a notion as old as the Lebombo bone’s notches — and by this standard the even numbers are exactly as many as all numbers. Fine. But in 1874, Cantor proved that the real numbers are uncountable: no list can contain them all. His 1891 diagonal argument is perhaps the most copied proof-idea in mathematics. There are different sizes of infinity — and, he showed in arithmetic of “transfinite” cardinals ℵ₀, ℵ₁, …, a whole infinite tower of them. “I see it,” Cantor wrote, “but I don’t believe it.”

The crisis of the paradoxes

In 1901, Bertrand Russell applied Cantor’s own diagonal trick to the grand edifice of set theory itself, defining the class R of all classes that do not contain themselves — which then must both contain itself and not contain itself. Russell’s paradox cut down not just Cantor’s system but Gottlob Frege’s Grundgesetze, the labor of a decade, at the very moment its second volume was in press. (“Your discovery is a very remarkable one,” Frege wrote; “hardly anything more unfortunate could befall it.”) The tower of mathematics, at its foundation, had been found to rest on air.

The response defines what mathematics is. In 1908, Ernst Zermelo proposed the axiom of separation — you may not form sets of “everything whatsoever,” only carve them out of existing sets — and over the following decades Abraham Fraenkel, Thoralf Skolem and others refined the system into what is now ZFC: a short list of axioms that has, a century on, proved equal to essentially all the mathematics humanity has ever done. The wound was cauterized, the foundation relaid, the chain repaired — with almost no loss above.

Gödel: the limits of the provable

And then, in 1931, a twenty-five-year-old Viennese logician named Kurt Gödel struck the tower itself. In any consistent formal system rich enough to contain arithmetic: (1) there are true statements it cannot prove; (2) it cannot prove its own consistency. Gödel’s incompleteness theorems are, in a precise sense, the most important theorems ever proved about proof. They ended a dream — Hilbert’s dream, the dream of a final, complete, self-certifying foundation for all mathematics — and replaced it with something more honest and more profound: mathematics is inexhaustible. No finite system of rules captures all truth. There will always be a next question.

This is sometimes told as a tragedy. It should be told as a liberation. It means the mathematical frontier will never close; it means the human mind can always reach past any rulebook it writes for itself. And it means that the chain of proven steps has no final link — not because the chain is flawed, but because truth is larger than any single system built to hold it.

Turing: proof becomes machine

In 1936, Alan Turing, working on a problem of pure logic (Hilbert’s Entscheidungsproblem: is there a mechanical procedure for deciding every mathematical question?), defined a machine — the Turing machine — that captures precisely what “mechanical” means. He proved no such universal procedure exists: some problems are undecidable. In doing so, he invented the computer. Not built it — defined it, years before any existed. The abstract machine he drew on paper, running left and right along an infinite tape of squares, is the exact mathematical ancestor of every processor ever manufactured. And every laptop, phone, and server on Earth executes a “universal machine” that Turing proved could exist, imitating any other, if fed the right program. The mathematics of what cannot be computed built the entire digital world.

---

VII. The Sixth Act: The Abstract Century (1900–2000)

The twentieth century took the great engines of the nineteenth and generalized them beyond recognition. The year 1900 is a good marker: at the International Congress of Mathematicians in Paris, David Hilbert listed twenty-three problems “for the centuries to come,” and mathematics spent the twentieth century answering them — and often, in answering, transforming them.

• Group theory — the mathematics of symmetry, from Galois’s 1832 deathbed work on why the quintic has no formula — grew to dominate algebra and then physics: Emmy Noether’s 1918 theorem (every symmetry of a physical law corresponds to a conserved quantity — every conservation law is a symmetry) is arguably the deepest single principle in physics. Energy conservation is time-translation symmetry. Angular momentum conservation is rotational symmetry. Noether, a woman barred from a professorship for most of her life, quietly taught the universe’s deepest grammar.
• Topology — Henri Poincaré’s “analysis situs” (1895) turned the vague notion of shape into a rigorous one: which properties survive stretching and bending? The century spent itself classifying knots, surfaces, and higher-dimensional manifolds, culminating in the century’s hardest triumph: the 2003 proof of the Poincaré conjecture by Grigori Perelman, who then declined the Fields Medal and the million-dollar Clay prize and returned to private life in St. Petersburg. It remains the only solved Millennium Prize Problem.
• Functional analysis and quantum theory — David Hilbert and John von Neumann built the mathematics of infinite-dimensional spaces (Hilbert spaces) that turned out to be the exact language of quantum mechanics. The Spectral Theorem, proved as pure mathematics, became the operating manual of the atom.
• Probability theory — Andrei Kolmogorov, in 1933, rebuilt all of probability on measure theory in a 60-page monograph, giving Fermat’s gambling letters the precision of Euclid.
• Computation and information — Claude Shannon’s 1948 “A Mathematical Theory of Communication” founded information theory: the bit, the laws of compression, the absolute limits of reliable communication through noise. And Noam Chomsky, at the same MIT, applied formal grammar to human language. The century ended with three billion people carrying devices that live entirely in the mathematics of this list.
• The great classifications — In a staggering coordinated effort (some 100 authors, ~10,000 pages), the classification of finite simple groups was completed in the early 1980s: every finite building block of symmetry is now known — a map of a whole universe. And through the Langlands program, begun by Robert Langlands in a 1967 letter to André Weil, number theory, representation theory, and geometry revealed themselves as one subject in three costumes — a grand unified theory of mathematics still being built.

The Chain meets physics

In 1915, Einstein’s general relativity took Riemann’s 1854 geometry as the literal structure of spacetime. In 1925–26, Heisenberg and Schrödinger took Hamiltonians, matrices, and PDEs as the literal structure of matter. In 1964, Peter Higgs and colleagues used group theory to predict a particle that took 48 years and the largest machine ever built (the LHC, a 27-km ring) to find. And in 2015, gravitational waves — ripples in the curved geometry of Riemann’s space, predicted by Einstein’s equations 100 years earlier — were detected. Time and again in this century, mathematics built for its own beauty turned out to have been tailoring the universe’s own clothes. Eugene Wigner called it “the unreasonable effectiveness of mathematics in the natural sciences,” and no one has satisfactorily explained it. The chain of proven steps keeps turning out to be, apparently, the same chain as the chain of being.

---

VIII. The Seventh Act: The Present (2000–2026)

The century so far has seen results that belong with any of the above.

• Perelman and Poincaré (2003) — via Richard Hamilton’s Ricci flow program, the century’s greatest topological problem fell, using differential equations to melt away the irregularities of 3-dimensional space itself.
• Fermat’s Last Theorem (Andrew Wiles, 1995, announced 1993–94) — a marginal note of 1637 (“I have a truly marvellous proof… too large for the margin”) took 358 years, and fell not by direct assault but as a corollary of the Taniyama–Shimura conjecture connecting elliptic curves and modular forms — a vastly deeper truth of which Fermat’s riddle was a shadow. The proof drew on the full machinery of the Langlands program, and Wiles worked seven years in his attic in secrecy. This is the chain at its most dramatic: the final link of a 358-year-old gap forged from metal smelted in the 1950s–80s.
• Machine verification — building on decades of work in proof assistants (Coq, Isabelle, HOL Light, and especially Lean), the mathematics community has begun to formally verify major theorems: Peter Scholze’s 2019 challenge to verify the “liquid vector space” cornerstone of his condensed-mathematics program was met by an Isabelle team in 2022; major parts of the classification of finite simple groups have been formalized; and recent results (2024–2025) — the Green–Tao-to-quadratic-system verifications, and the formalization of entire research-level arguments by large Lean collaborations — mark a genuine phase change. Mathematics is no longer solely an activity of the unaided mind; the chain of proven steps is now being checked, link by link, by machines that cannot be fooled.
• Terence Tao’s era of “new tools” — Tao’s 2024 result with Gowers and others on the conjectured structure of sets with no three-term arithmetic progressions uses new kinds of algebraic decompositions; Tao’s liquid tensor experiments and his polymath projects show another modern mode: mathematics done in public, in real time, by many hands. The chain now has many forges.
• The Asperó–Schindler theorem (2021) — in the set theory of the continuum, showed that two of the strongest competing axioms for resolving Cantor’s first great unanswered question (MM⁺⁺ implies Woodin’s axiom (∗)) both settle the size of the continuum at ℵ₂. Two schools of thought about the infinite turned out, astoundingly, to converge on the same answer.
• AI enters the forge — the 2024 Fields medal-level work on problems like bounds in combinatorics and 2025-era large collaborations pairing human mathematicians with AI conjecture-generation and proof-search tools are early signs of a new epoch: not computers replacing mathematicians, but the chain of proven steps growing new instruments. Wiles needed an attic; the next Wiles may need a GPU cluster.

And the open problems remain magnificent: the Riemann hypothesis (the music of the primes, Hilbert’s eighth problem, Clay’s second Millennium problem) — where a strange deep ocean remains unplumbed; the Hodge conjecture; P vs NP (is finding a proof really harder than checking it? — the question that guards all of cryptography and much of complexity theory); the Navier–Stokes regularity problem; the Birch and Swinnerton-Dyer conjecture. Every one of these will fall, or be transformed, or reveal a new Gödel-like limit. Each will deserve an essay of its own.

---

IX. The Future: Where the Chain Leads

To predict mathematics is famously dangerous — Hilbert himself declared in 1930 that “in mathematics there is no ignorabimus,” we will not forever not-know. Gödel answered him within the year. But some directions are less guesses than the visible momentum of the chain itself.

Foundations will pluralize. Homotopy type theory (2013) — univalent foundations, in which equality itself is rebuilt — offers an alternative bedrock to ZFC, with machine-checkability built in from the start. Mathematicians of 2100 may stand on several foundations at once, moving between them the way we move between geometries today.

Verification will become the norm. Within a generation, it is reasonable to expect that major results will routinely ship with a machine-checkable certificate, the way software ships with tests. The sociological transformation — a proof anyone can check — may be as big as the invention of the journal.

AI as collaborator. Systems that conjecture, search for proofs, and find counterexamples — already finding novel results in combinatorics and extremal problems — will change the craft of mathematics as profoundly as the telescope changed astronomy. The chain of proven steps will not change: each link will still be a result and its proof. But the hands that forge the links will include ones we built.

The physics boundary will keep dissolving. The dialogue between mathematics and fundamental physics — mirror symmetry, the geometric Langlands program and its 2024 proof of the fundamental lemma in the function-field case, quantum field theory as a source of new mathematics — suggests that the deepest mathematics of the next century is partly already present in physical ideas we cannot yet formalize. Richard Feynman said “I can’t do the math” of quantum electrodynamics; the next century’s task is to do the math that the universe has been doing all along.

The old questions will not die. The Riemann hypothesis is 167 years old. Goldbach’s conjecture is 284. Fermat waited 358. Some readers of this essay will see problems from this list fall. When the Riemann hypothesis falls, the primes will be revealed not as scattered noise but as the hidden music — and the structure of every secure code ever written will simultaneously be newly illuminated and newly fragile.

And that is the future of the chain: not an ending, but a wider forge.

---

X. Epilogue: What the Chain Teaches

Stand back from the whole chain and it says something simple and staggering at once.

Every proof that exists was checked by a human being — or, lately, by a machine built by human beings — and none of it fell from the sky. The theory of everything curved began as a dispute about a Greek square root. The digital age began as a logician’s question about whether every question has an answer. The device on which you may be reading this essay runs on Boolean algebra — the thought of a nineteenth-century schoolteacher in Cork who wanted to write down the laws of reasoning itself.

The chain also teaches honesty. It is a chain of proven steps, which means every link had to be verified — by someone who had no reason to be gentle, and who often had every reason to be hostile. Tartaglia hated Cardano’s proof, and checked it anyway. Hilbert’s own program was demolished by Gödel, and Hilbert himself accepted it. The deepest social technology humanity has ever invented is not the market, not the state, not the court. It is the proof: the practice of settling questions in a way that any competent person, of any nation, era, or temperament, can check for themselves. It is the one arena in which the human race has ever built genuinely permanent knowledge, where “wrong” simply does not survive.

And it teaches patience. The Lebombo bone’s notches waited forty-four thousand years to be recognized as mathematics. Eudoxus waited twenty-three centuries for Dedekind. Fermat’s margin waited three and a half centuries for Wiles. The chain does not hurry, but it also does not break.

Somewhere tonight, a twenty-three-year-old is writing a proof of something no one has ever understood. In a century we will know whether it was right. In three centuries we may know what it meant.

The chain is still being forged. That is the best news in the history of the world.

---

Annotated References

1. Bell, E. T. Men of Mathematics (1937). The classic romantic history of mathematics — wonderfully written, inspirational to generations of mathematicians (including the young Freeman Dyson and countless others), but factually loose by modern standards: its anecdotes (e.g., the deaths of Hypatia’s era, Galois’s legend) should be checked against primary sources. I cite it as literature, not as record; the essay above deliberately avoids its more theatrical claims.
2. Boyer, Carl B., and Uta C. Merzbach. A History of Mathematics, 3rd ed. (Wiley, 2011). The standard one-volume scholarly history; the best single reference for the sweep of this essay, from Babylonian place value through the twentieth century. Its treatment of the Greek crisis of irrationality and of the medieval transmission is authoritative.
3. Kline, Morris. Mathematical Thought from Ancient to Modern Times (Oxford, 1972). Massive, opinionated, and unsurpassed for the nineteenth century — especially the growth of rigor (Cauchy, Weierstrass, Dedekind) and the parallel postulate saga (Lobachevsky, Bolyai, Riemann). Kline’s emphasis on mathematics as driven by problems rather than pure ideas complements the chain-of-repairs narrative of this essay.
4. Katz, Victor J. A History of Mathematics: An Introduction, 3rd ed. (Addison-Wesley, 2009). The most balanced modern textbook history, particularly strong on Islamic, Indian, and Chinese mathematics — correcting the Eurocentric bias of older accounts — and on the manuscript transmission that links al-Khwarizmi to Fibonacci to Renaissance Italy.
5. Neugebauer, Otto. The Exact Sciences in Antiquity (Princeton, 1952; Dover reprint). The foundational modern study of Babylonian mathematics, including the analysis of YBC 7289 (√2 to six sexagesimal places) that this essay’s second section draws on. Neugebauer changed forever the view that Greek geometry was the start of mathematics.
6. Heath, Thomas L. A History of Greek Mathematics (Oxford, 1921; Dover reprint). Still indispensable for the Greek chapters: the irrationality of √2, Eudoxus’s theory of proportion (with Heath’s own translation of Euclid Book V), and the Euclidean standard of proof. Heath’s claim that Dedekind’s cut is “Eudoxus made exact” remains disputed — see Gardes (1988, in French) for the skeptical view — and the essay follows the disputed-status reading.
7. Joseph, George Gheverghese. The Crest of the Peacock: Non-European Roots of Mathematics, 3rd ed. (Princeton, 2011). The essential corrective history for the Indian, Chinese, Islamic, and African chapters: the Lebombo and Border Cave notched bones, Brahmagupta’s rules for zero and negative numbers (628 CE), and the priority disputes around them. Joseph’s insistence on parallel developments rather than single linear transmission directly shaped this essay’s honesty about missing links in the chain.
8. MacTutor History of Mathematics Archive (University of St Andrews). The best free online biographical resource, with careful dating and full bibliographies for essentially every mathematician named above (Eudoxus, Brahmagupta, Bombelli, Stevin, Cantor, Noether, and so on). Its “history of zero” and “history of the real numbers” articles were used repeatedly. Cross-checked against Britannica where dates conflicted.
9. Stanford Encyclopedia of Philosophy (SEP). Peer-reviewed, authoritative entries on the foundations chapters: “The Early Development of Set Theory” (Cantor 1874/1891, the paradoxes), “Dedekind’s Contributions to the Foundations of Mathematics” (1872 cuts), “Gödel’s Incompleteness Theorems,” “Skolem’s Paradox,” and “Zermelo’s Axiomatization of Set Theory.” The gold standard for the 1872–1940 sections.
10. Zermelo, Ernst. “Über Grenzzahlen und Mengenbereiche” (1930) and Zermelo (1908). Primary sources for the axiomatization that became ZFC. Paired with Fraenkel, Abraham. Einleitung in die Mengenlehre (1919) — the textbook that carried Replacement into the standard system. Read in translation in From Frege to Gödel: A Source Book in Mathematical Logic (van Heijenoort, ed., Harvard, 1967), which is itself the essential primary-source collection for this entire period.
11. Gödel, Kurt. “Über formal unentscheidbare Sätze…” (1931). The incompleteness paper, in van Heijenoort’s translation. Also Gödel’s 1940 monograph The Consistency of the Continuum Hypothesis (Princeton, Annals of Mathematics Studies), the record of the 1938 PNAS announcement. Cohen’s independence is Cohen, Paul J. “The Independence of the Continuum Hypothesis,” PNAS 50/51 (1963–64). Together these are the decisive modern results on the limits of axioms.
12. Turing, Alan M. “On Computable Numbers…” (1936). The founding document of computer science, in the Proceedings of the London Mathematical Society — still perfectly readable, and the origin of the universal machine. Paired with Davis, Martin, ed. The Undecidable (1965) for the surrounding decidability literature.
13. Shannon, Claude E. “A Mathematical Theory of Communication” (Bell System Technical Journal, 1948). The founding paper of information theory; the source for the claim that the digital age has a single mathematical charter. Also Shannon’s 1937 master’s thesis (Boolean algebra applied to switching circuits) — cited here as the moment Boole became hardware.
14. Noether, Emmy. “Invariante Variationsprobleme” (1918). The Noether symmetry-conservation paper; for an accessible modern account see Byers, Nina. “E. Noether’s Discovery of the Deep Connection Between Symmetries and Conservation Laws” (1998). The essay’s account of conservation-as-symmetry rests on these.
15. Wiles, Andrew. “Modular Elliptic Curves and Fermat’s Last Theorem” (Annals of Mathematics 141, 1995). The proof itself; the definitive popular account is Singh, Simon. Fermat’s Enigma (1997), used for the historical narrative (the 1637 margin, the 1986 Ribet link, the seven secret attic years). Verified against Darmon–Diamond for the mathematical structure.
16. Perelman, Grigori. “The Entropy Formula for the Ricci Flow…” (arXiv:math/0211159, 2002) and its two sequels (2003). The Poincaré conjecture proof, posted and never journal-published; the mathematical completion of the program is documented in Cao–Zhu and Morgan–Tian book-length expositions. The essay’s account of the surrounding history follows the Clay Mathematics Institute’s official record.
17. Wigner, Eugene P. “The Unreasonable Effectiveness of Mathematics in the Natural Sciences” (Communications on Pure and Applied Mathematics, 1960). The canonical statement of the mystery this essay returns to repeatedly — mathematics built for itself turning out to fit the universe. Still unanswered; the best modern replies are in Hamming (1980) and the “effectiveness” literature since.
18. Univalent Foundations Program. Homotopy Type Theory: Univalent Foundations of Mathematics (Institute for Advanced Study, 2013). The collective manifesto of the alternative-foundation program described in the future section; written in a single semester by a remarkable group collaboration, and itself an argument for the polymath style of mathematical production.
19. The mathlib Community. “The Lean Mathematical Library” (2020–present), at leanprover-community.github.io/mathlib. The record of the machine-verification epoch: the formalization statistics, the liquid tensor experiment (2021–22, Scholze’s challenge), and the growing corpus of formally verified research mathematics. Used for the 2019–2026 formalization claims.
20. Asperó, David, and Schindler, Ralf. “\mathsf{MM}^{++} implies Woodin’s axiom (*)” (Annals of Mathematics 195, 2021). The source for the continuum-convergence result in the present-tense section — with the surrounding set-theoretic context from Woodin, W. Hugh. “The Continuum Hypothesis” (2001) for the two competing axiom programs that it unexpectedly reconciled.
21. Clay Mathematics Institute. “The Millennium Prize Problems” (2000), at claymath.org. The official statements of the seven problems (Riemann, Hodge, P vs NP, Navier–Stokes, BSD, Yang–Mills, Poincaré), one of which is now solved. Used for the open-problems section; the Riemann hypothesis exposition follows Conrey, J. Brian. “The Riemann Hypothesis” (Notices of the AMS, 2003).
22. Konstapel, J. Mathematics as a Chain of Proven Steps: The 63 fields of mathematics, from the first proofs to 2026 (2026). The book that occasioned this essay — a history of mathematics organized, uniquely, by dependencies between proofs rather than by people or periods, across all 63 top-level fields of the MSC2020 classification, with each link sourced to opened scholarship and each quick check recomputed. The chain-of-repairs framing of this essay, the honest notation of independent developments, and several specific details (the Lebombo/Border Cave dates and their disputes, the two-wedge zero placeholder, the Brahmagupta-to-Bombelli missing link, the Eudoxus-to-Dedekind dispute, the Cohen 1963 dating, and the Asperó–Schindler result) are drawn from its chapters and notes.

Jump to the english text here

Visit the website here.

lees de handleiding hier.

J.Konstapel,Leiden,28-10-2026.

De Maze Transformer is een instrument dat wetenschappelijke artikelen leest en er alleen de kwantitatieve claims uit haalt: in de natuurkunde zijn dat machtswetten ((y \sim x^e)), in de geneeskunde effectgroottes zoals odds ratio’s met betrouwbaarheidsintervallen. Die claims worden in één ruimte geplaatst. Twee claims die een grootheid delen, kunnen aan elkaar geketend worden tot een pad. Zo’n pad voorspelt een verband dat geen enkel afzonderlijk artikel formuleert. Als een derde artikel dat verband rechtstreeks meet, wordt de voorspelling getest.

Werkwijze in vier stappen

1. Extraheren – alleen kwantitatieve claims (wetten en effecten) uit elk artikel.
2. Plaatsen – elke grootheid krijgt een plek; artikelen die dezelfde grootheid gebruiken, ontmoeten elkaar daar.
3. Ketenen – twee of drie claims uit verschillende artikelen worden gecombineerd tot een pad (exponenten worden vermenigvuldigd; richtingen ook).
4. Testen – een directe meting bevestigt, weerlegt of laat het pad onbeslist.

Theoretische basis

Het instrument steunt op de Vacuum.Net-theorie (Fundamental Paper, 8e ed., v8.1). Daarin is een wet tussen twee grootheden een rechte lijn in logaritmische positieruimte, en is een richting een trit (+1, 0 of −1). De theorie levert de woordenschat (plaats, passage, trit, pad, sluiting); de Transformer test niet de fysische claims van de theorie, maar of die woordenschat gepubliceerde claims nuttig ordent.

De twee MAZEs

• Oude MAZE – een zelfadresserend kennisnet van ~8,2 miljoen items (“residents”), o.a. Wikipedia-artikelen en patenten. Adressen komen uit inhoud en worden nooit herschreven.
• Nieuwe MAZE – een afgeleide bibliotheek van 635 nodes in 71 “windings”, opgebouwd van niets met exacte checks per stap (SymPy/integer-rekenen).

De Transformer is op 28 oktober 2026 aan beide MAZEs gekoppeld: grootheden zijn aan Wikipedia-artikelen gekoppeld (oude MAZE) en aan nodes (nieuwe MAZE); de 645 arXiv-artikelen en 1.303 PubMed-records zijn zelf residents geworden.

Resultaten

Map	Bron	Inhoud
Melkwegstelsels	150 arXiv-papers (astro-ph.GA)	190 plaatsen, 210 wetten, 2.648 paden
Levenswetenschappen	495 arXiv-papers (q-bio, physics.bio-ph)	202 wetten na filter, 2.212 paden
Geneeskunde	1.303 PubMed-records (parkinson)	224 plaatsen, 314 edges, 41 testbare paden

• Geneeskunde: 36 van 41 testbare paden bevestigd door een onafhankelijke meting ((p = 0.0015)); onder een strengere onafhankelijkheidsregel 24 van 26 ((p = 0.002)). Dit toont richting, niet causaliteit.
• Natuurkunde: in geen van beide corpora meet een artikel een geketend paar rechtstreeks; de 4.860 paden blijven dus ongeteste voorspellingen (een eigenschap van de corpus, niet van de methode).
• Blinde checks: 36 van 40 MAZE-links correct, geen fout; 45 van 50 namen correct; na de taalmodel-filter 52 van 60 “wetten” echt (87%).
• Beperkingen: de assessor is een taalmodel; taalmodellen zitten in de leeslus; L13–L19 zijn op dezelfde data afgeleid; geneeskundepaden zijn niet causaal; naam-only links kunnen een veldverkeerde betekenis pakken (bv. “noise” in de biofysica).

Geannoteerde referentielijst

1. Konstapel, J. (2026). The Vacuum.Net Theory: Foundational Paper, 8e ed., v8.1. Constable Research, Leiden. — De canonieke uiteenzetting van de theorie: axioma’s N1–N5, stellingen T1–T23, correspondenties, voorspellingen en open punten, met een register van alle checks. Lees dit voor de status van elke claim.
2. Konstapel, J. (2026). Curves in the Vacuum.Net, v1.0. — Leidt af wat een pad in het net is; levert Stelling 8 (twee wetten ontmoeten elkaar in precies één punt) waarop de Transformer rust.
3. Spencer-Brown, G. (1969). Laws of Form. London: Allen & Unwin. — Het calculus van de onderscheiding waaruit de theorie vertrekt: één daad van onderscheiding die zichzelf herintreedt, genereert een logica.
4. Kauffman, L. H. (1987). “Self-reference and recursive forms.” Journal of Social and Biological Structures, 10(1), 53–72. — Verbindt re-entry, eigenvormen en imaginaire waarden; onderbouwt de stap van onderscheiding naar oscillatie.
5. Rowlands, P. (2007). Zero to Infinity: The Foundations of Physics. Singapore: World Scientific. — De nilpotente vorm van de Dirac-vergelijking die het Fundamental Paper overneemt; laat zien waar de theorie Dirac als input neemt.
6. Larson, D. B. (1959). The Structure of the Physical Universe. Portland: North Pacific. — Historische, onorthodoxe bron voor de ruimte-tijddimensies van de eenheden; kritisch lezen.
7. Knuth, D. E. (1997). The Art of Computer Programming, Vol. 2, 3e ed., §4.1. — Standaardwerk over gebalanceerd ternair rekenen, het getallensysteem achter elk MAZE-adres.
8. Hayes, B. (2001). “Third base.” American Scientist, 89(6), 490–494. — Leesbaar betoog waarom basis drie het zuinigste talstelsel is; verklaart de trit als technische keuze.
9. Barenblatt, G. I. (1996). Scaling, Self-Similarity, and Intermediate Asymptotics. Cambridge UP. — Waarom machtswetten de natuurlijke vorm van kwantitatieve claims in de fysica zijn; achtergrond voor Layer A.
10. West, G. B., Brown, J. H., & Enquist, B. J. (1997). “A general model for the origin of allometric scaling laws in biology.” Science, 276(5309), 122–126. — Het klassieke voorbeeld van machtswetten in de biologie; verklaart waarom de levenswetenschappen-map bestaat.
11. Swanson, D. R. (1986). “Fish oil, Raynaud’s syndrome, and undiscovered public knowledge.” Perspectives in Biology and Medicine, 30(1), 7–18. — Oorsprong van literatuurgebaseerde ontdekking: A→B in de ene literatuur en B→C in de andere impliceren A→C. De paden van de Transformer zijn hier een kwantitatieve versie van.
12. Smalheiser, N. R., & Swanson, D. R. (1998). “Using ARROWSMITH…” Computer Methods and Programs in Biomedicine, 57(3), 149–153. — De software-opvolger van Swansons methode; vergelijk de woordgebaseerde links met de exponent- en effectgebaseerde links van de Transformer.
13. Tshitoyan, V., et al. (2019). “Unsupervised word embeddings capture latent knowledge from materials science literature.” Nature, 571, 95–98. — Toont dat verborgen relaties over veel artikelen latere bevindingen kunnen voorspellen; moderne tegenhanger die met embeddings werkt in plaats van expliciete claims.
14. Fortunato, S., et al. (2018). “Science of science.” Science, 359(6379), eaao0185. — Overzicht van de kwantitatieve studie van de wetenschappelijke literatuur; plaatst de Transformer in dat veld.
15. DerSimonian, R., & Laird, N. (1986). “Meta-analysis in clinical trials.” Controlled Clinical Trials, 7(3), 177–188. — De random-effects-methode voor het poolen van effectgroottes; achtergrond voor hoe Layer B effecten op één edge combineert.
16. Higgins, J. P. T., & Thompson, S. G. (2002). “Quantifying heterogeneity in a meta-analysis.” Statistics in Medicine, 21(11), 1539–1558. — Definieert (I^2), dat de geneeskunde-map toont voor edges met meer dan één effect.
17. Davey Smith, G., & Ebrahim, S. (2003). “‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease?” International Journal of Epidemiology, 32(1), 1–22. — Het design dat geneeskunde-edges een causale richting kan geven; de geplande volgende stap voor Layer B.
18. Lipscomb, C. E. (2000). “Medical Subject Headings (MeSH).” Bulletin of the Medical Library Association, 88(3), 265–266. — Korte geschiedenis van de woordenschat die de plaatsen van Layer B levert.
19. Lehmann, J., et al. (2015). “DBpedia: a large-scale, multilingual knowledge base extracted from Wikipedia.” Semantic Web, 6(2), 167–195. — Bron van de artikelsamenvattingen waarmee de oude MAZE residents naar betekenis rangschikt.
20. arXiv (2026). arXiv API User’s Manual and bulk source access; NCBI (2026). Entrez Programming Utilities. — De twee publieke interfaces waarlangs alle artikelen zijn opgehaald.

Code, plannen en data: SWARP-repository, maze/transformer. De kaarten: maze.swarp.nl/transformer (ook /levenswetenschappen en /geneeskunde). Dit essay als PDF: maze.swarp.nl/transformer/essay.pdf.

Article

A research field states its knowledge across many papers. Each paper makes a few hard, quantitative claims. Few papers state how their claims connect to those of other papers.

The Maze Transformer extracts such claims and places them in one space. In the physical sciences a claim is a power law, y ∼ x^e. In medicine it is an effect size, such as an odds ratio with its interval.

Two claims that share a quantity can be chained. A chain predicts a relation that no single paper states. When a third paper measures that relation directly, the prediction is tested.

The instrument rests on the Vacuum.Net theory. In that theory a law between two magnitudes is a straight line in logarithmic position space. A direction is a trit: +1, 0 or −1.

Three maps are live: galaxies (150 arXiv papers), the life sciences (495 arXiv papers) and Parkinson’s disease (1,303 PubMed records). In medicine, 36 of 41 testable chains were confirmed by an independent measurement (p = 0.0015; 24 of 26 under a stricter independence rule, p = 0.002).

In the two physics maps no paper measures a chained pair directly. Their 4,860 chains remain untested predictions.

Where a sure match exists, each quantity is linked to the existing knowledge net (the old MAZE) and to the derived library (the new MAZE). The papers themselves have become residents of the net. A blind check found 36 of 40 links correct and none wrong.

---

1. Purpose

The Maze Transformer has one aim. It should show what a research field claims, where those claims meet, and where a claim is still missing.

A field publishes thousands of papers a year. A reader sees them one at a time. Connections between papers live in the heads of experts and in review articles. Those reviews are slow and selective. The Transformer makes the connections explicit and countable.

The work proceeds in four steps:

1. Extract. It reads each paper and keeps only its quantitative claims: laws and effects.
2. Place. It gives every quantity a place. Papers that use the same quantity meet at that place.
3. Chain. It combines two or three claims from different papers into a path. A path predicts a relation that none of the papers states.
4. Test. When a further paper measures the predicted relation directly, the path is confirmed, undecided or contradicted.

The result is a visual map of a field. Dots are quantities. Lines are claims. The list of untested paths is a list of concrete research questions. Each question names the two quantities, the predicted exponent or direction, and the papers it rests on.

The design rule is simple. Count what can be counted. Say “untested” when something is untested. Every number on the map can be traced to a sentence or a formula in a named paper.

---

2. The Setting: SWARP and the Two MAZEs

2.1 SWARP

SWARP is a family of web applications built by Constable Research in Leiden. Its stated goal is bottom-up: citizens should know what happens in their own neighbourhood and find each other there.

Around that goal sit several standalone services. Among them are a daily guide, a local-environment monitor, a financial weather instrument and a set of research tools.

The research tools share one knowledge net, called the MAZE, at maze.swarp.nl. The Transformer is one of these tools. It runs on the same server as the MAZE and uses its database.

2.2 The Old MAZE: a Self-Addressing Net

The MAZE is a knowledge net in which every item has an address computed from its content. It holds about 8.2 million items, called residents.

The largest groups are 3.6 million English Wikipedia articles and 4.6 million European patent applications. Smaller groups include mathematical theorems, works of art, music, blog essays and open questions.

An address is a word of 24 trits. Two generations of addressing exist side by side.

The first computes the address from the words of a description. The second computes it from meaning. A text embedding is projected on fixed, variance-ordered axes. A ternary tree then adapts to the residents below each prefix. Items with similar meaning therefore share a long address prefix.

Five invariants govern the net. The most important are that an address follows from content alone, and that an address, once written, is never rewritten.

2.3 The New MAZE: a Derived Library

The new MAZE is built the other way round. It does not copy any existing classification. It unwinds from nothing, one operation at a time.

Each node follows from earlier nodes by one stated operation. Each node carries an exact check, computed with integer arithmetic or SymPy. If one check fails, nothing is written.

The library now holds 635 nodes in 71 windings. It runs from the first distinction through quaternions, the Dirac algebra and the fermion spectrum. It continues to the hydrogen atom, the periodic table, chemical binding, the genetic code and further layers.

Every node has one of five statuses: adopted, derived, corresponding, tested or open. Its names come afterwards, from the literal first sentence of a Wikipedia article with its revision number.

On 19 October 2026 the old MAZE was hung under the new one as a branch. Every resident can now be reached from the first step.

---

3. The Vacuum.Net Theory in Brief

The Transformer takes its concepts from the Vacuum.Net theory. The canonical statement is the Foundational Paper, eighth edition, version 8.1 (17 October 2026). The essay Curves in the Vacuum.Net (24 October 2026) derives what a path in the net is.

This section summarises both, as far as the Transformer needs them. It reports the theory’s own claims. Their status is the status the theory itself assigns.

3.1 Five Axioms

The theory starts from one object: a strand, an ordered course through a set of places. A loop is a return of the strand to a place it has visited. A place visited twice or more is a knot.

Five axioms follow:

• N1, strand. There is exactly one strand. Every distinction is a loop.
• N2, closure. A loop persists only if it lies in a closed window. What does not close passes on as tension.
• N3, memory. At every knot the passages are ordered. The first passage fixes the direction; later passages are read relative to it.
• N4, context. Every constant is a function of the strand.
• N5, scale. Merged into blocks, the coarse strand obeys N1 to N4 again. What does not close at one scale is material for the next.

3.2 Trits, Closure and Depth

A direction must differ from its reverse. The smallest cyclic group in which that holds is Z₃.

A passage through a knot therefore has one of three values relative to the first passage:

• +1: it leaves the same way.
• −1: it leaves the way the first came in.
• 0: another way.

These values are trits.

Three trits close when their sum is 0 modulo 3. Of the 27 possible windows, 9 close. Every closed word breaks down into three elementary closures: a standing (0), a meeting (x, −x) and a ring (x, x, x).

Every integer has exactly one balanced-ternary word. Its length is its depth in scales. Meetings double and rings triple.

Because 2^a = 3^b has no solution except a = b = 0, the net never closes completely. It comes closest at depths 1, 2, 5, 12, 41, 53, 306 and further. These are the convergents of log₂3, the source of the Pythagorean comma.

3.3 What the Foundational Paper Adds

Edition 8.1 writes the theory as one chain. It starts from the ground of distinction (Spencer-Brown, Kauffman). It continues through the ternary register: unique addresses, the window register and the two ladders.

It then states a knot, a double cover and a conserved Hopf charge. It gives a lever for a free path, γ = −(1 + P)/P. It writes Dirac’s equation in Rowlands’ nilpotent form; there Dirac is input, not output. It reads Maxwell’s quaternions as two readings of one closure, as a correspondence. Units follow Larson’s dimensions.

The paper separates proved theorems (T1–T23), correspondences, identifications, five predictions with prior commitments and seven open points. It also keeps a register of every check.

3.4 The Part the Transformer Uses

The Transformer uses one bridge from this theory to ordinary science. A measured magnitude q is placed by its position p = log₃ q, its depth in scales.

Theorem 8 (Curves, Two Laws). A power law q₂ = C · q₁^e is a straight line in the plane of p₁ and p₂ with slope e. Two laws between the same two magnitudes are identical, parallel, or meet in exactly one point.

A law is therefore a slope. The sign of the slope is a trit. A chain of laws through shared magnitudes is a path in the sense of the theory. Its exponent is the product of the exponents along the way.

A path either closes against a direct measurement or carries tension. In the new MAZE this reading is fixed in winding 69. Node lus_helling holds the slope trit, with Kepler’s third law over eight planets as its calibration (slope 1.4978 ± 0.0007 against the law’s 3/2). Node lus_overal holds the rule for which paths close everywhere.

---

4. How the Transformer Works

The Transformer has two layers. Layer A reads formulas. Layer B reads effect sizes. Both build the same structure: places, passages, paths and a gap list.

4.1 Places, Passages and Paths

A place is a quantity, such as stellar mass, luminosity or the risk of dementia.

A passage is one claim in one paper, from one place to another. In Layer A it carries an exponent e and, when given, an uncertainty u. In Layer B it carries a ratio with a confidence interval and a direction.

A path is a chain of two or three passages from different papers through shared places. Its exponent is the product of the exponents. In Layer B its direction is the product of the directions.

A path is tested when some paper measures its two end places directly. It is confirmed when that measurement agrees within the combined uncertainty. It is contradicted when its sign differs, or when it lies more than three combined uncertainties away. Anything between is undecided.

A path without a direct measurement is a gap: a prediction and a research question.

4.2 Layer A: Formulas from arXiv

For each arXiv paper the Transformer fetches the LaTeX source. Fixed, deterministic rules read the formulas. They recognise forms such as:

• y ∝ x^e, y ∼ x^e, y = C x^e;
• products of several factors;
• a power on the left-hand side;
• logarithmic forms, exponentials and cycles.

They read the exponent, an uncertainty in ± or parenthesis notation, and the definition of each symbol from the surrounding text.

The rules were extended step by step as blind checks exposed errors. Nineteen reading rules now exist (L1–L19). Examples: a value with units is not a law; nothing is read from figures and tables; “for t ∼ t” is a condition, not a law.

Ten path rules (K1–K10) fix how laws chain. A law may be taken in reverse, with exponent 1/e. A path has two or three links from different papers. A path may not loop back on itself.

A language model (GPT-4o, temperature 0, cached) does two things only. First, it names a symbol when the text does not define it (rule L10). Second, in the life sciences, it filters out “laws” that are really definitions, notations or parameter values (rule L20). Both steps were tested blind (S6).

4.3 Layer B: Effect Sizes from PubMed

For medicine the claims are effects, not formulas. The input is a PubMed record in MEDLINE format: identifier, year, MeSH terms and abstract.

A deterministic reader finds every odds ratio, hazard ratio or relative risk with its interval in the abstract. Places are the MeSH descriptors of the record.

A language model then reads, per effect, which exposure and which outcome are meant, and in which direction. It chooses from the MeSH list of that record.

Effects on the same pair are pooled with a random-effects model (DerSimonian–Laird). Each edge gets a pooled estimate, an interval, the identifiers of its papers and a direction trit.

A path A → M → B predicts the direction of A → B as the product of its two directions.

4.4 “Have It Examined”

Any visitor can submit an arXiv identifier. The server fetches the paper, reads its laws with the same rules and lays it against the map.

It reports which quantities the paper shares with the map, which new paths it opens, and whether it measures a gap directly. A paper that measures a gap turns a prediction into a test.

Limits keep the cost bounded: at most 40 name questions per paper, 300 per day in total, five papers per visitor per day. A paper already examined costs nothing.

---

5. What Is in It, and What Can Come In

5.1 Contents Today

Map	Source	Contents
Galaxies	150 arXiv papers, astro-ph.GA, February 2025, each with at least one ∝	190 places · 210 laws · 2,648 paths
Life sciences	495 arXiv papers, q-bio and physics.bio-ph, March–November 2025	924 laws read · 202 after filter · 2,212 paths
Medicine	1,303 PubMed records on Parkinson’s disease, retrieved 25 October 2026	2,184 statements · 1,791 edges · 51,057 paths

The medicine map shows the strongest part of the net: 224 places, 314 edges and the 41 testable paths.

All three maps are browsable at maze.swarp.nl/transformer, /transformer/levenswetenschappen and /transformer/geneeskunde.

5.2 What Can Come In

• Any arXiv paper with LaTeX. The reading rules are field-independent. The naming step currently describes the paper as astrophysics, so names in other fields are less reliable until that is generalised.
• Any disease or exposure in PubMed. Layer B needs only a PubMed query. Parkinson’s disease was the first domain.
• Publisher XML. The specification allows JATS XML with MathML for papers outside arXiv. This route is not yet built.
• Larger corpora. A single subfield with about 1,000 papers is the next step for physics. More overlap means more chance that one paper measures what two others imply.
• Measured values. Catalogues of measured data, such as stellar or galaxy catalogues, could test physics paths directly instead of waiting for a paper.
• Directed edges. In medicine, edges from randomised trials, cohorts and Mendelian randomisation can carry direction of cause. The publication type is already stored.

---

6. Results

Every measurement below was run under a plan whose hash was committed before counting. Failures are reported, not repaired after the fact. Blind checks were done by a separate assessor, a language-model agent, that saw only the items. Planted errors served as a control that the assessor actually checks.

6.1 Does the Space See Topics?

The first pilot (25 October) still used words as places, as in the first specification, which was later replaced by quantities. It used 30 papers on two unrelated topics.

Papers on the same topic shared 14.0 knots on average, against 5.2 across topics (p = 0.0001). The share of +1 trits was 0.089 within a topic against 0.0023 across, a factor of 38.

The first run was invalid because of an error in the control. The second, with a corrected control, passed.

A caution remains. A +1 at the word level mostly marks a recurring two-word term. This pilot shows that the space separates topics. It does not yet show that trits add anything beyond word pairs.

6.2 Reading Laws and Naming Quantities

Naming was the bottleneck. With the original definition patterns (“where X is…”), 97% of symbols in 50 papers stayed unnamed. Authors rarely write definitions that way. Two patterns and a language-model step were added.

Names, blind. On the galaxy map 45 of 50 names were judged correct: 39 of 44 from the language model, 6 of 6 from fixed rules. Four of five planted errors were caught.

Laws in the life sciences. The first blind check found only 25 of 60 “laws” real (42%). A third were definitions or notation, such as curvature = 1/radius or X ∼ Exp(λ). After the language-model filter, a fresh blind sample scored 52 of 60 (87%), with all planted errors caught.

6.3 Paths in Physics: Untested

The galaxy map has 2,648 paths and the life-sciences map 2,212. In neither corpus does any paper measure the two end quantities of a path directly. The paths are therefore predictions only.

This is a property of the corpus, not of the method. Papers report laws, not paired measurements of quantities that other papers link. A larger subfield or a catalogue of measured values is needed.

6.4 Effects in Medicine: Tested

Binding exposure and outcome. Fixed rules bound the right exposure and outcome in only about 25% of effects. An independent second implementation of the same specification agreed with the first on 77% of sentences, but both were mostly wrong on the disputed ones. With the language model reading each effect, the second prompt version scored 54 of 60 (90%) in a fresh blind sample.

Paths. Of 41 testable paths, 36 were confirmed by a direct measurement (88%). A permutation test gave p = 0.0015.

Stricter count, after the fact. In 13 of the 41 paths the direct measurement came from the same papers as the chain. One paper confirmed itself. Counting only paths whose direct edge shares no paper with its links gave 24 of 26 (92%), p = 0.002. This stricter rule was added afterwards; from now on independence is required in every plan beforehand.

What this shows: when two independent papers report A → M and M → B, the product of their directions predicts the directly measured direction of A → B in about nine of ten cases.

What it does not show: cause. Many paths mix populations. “Male sex → Parkinson’s → mortality” links the risk of getting the disease to the risk of dying with it. A correct direction is not a mechanism.

---

7. Relations Established with the MAZE

On 28 October 2026 the Transformer was linked to both MAZEs. Before that it ran on the same server but stood apart in content.

Two plans governed the work. Plan 1 (hash 89aaba04…) failed its blind check. Plan 2 (hash 92b3f7b0…) passed.

7.1 Quantities to the Old MAZE

Each quantity on the three maps is matched to an English Wikipedia article. Two rules apply:

• The article must exist under exactly the quantity’s name, with only the first letter capitalised.
• Or Wikipedia must redirect to it while every content word of the name is kept.

A disambiguation page never counts. The matched title is then looked up as a resident of the old MAZE. No statistics and no language model take part.

Plan 1 failed. Without these safeguards, 31 of 40 sampled links were correct (77.5%, threshold 85%). The errors had clear causes: disambiguation pages (“survival”, “higher”), titles of works (“football players” led to a painting) and redirects to broader concepts (“time interval” to Time, “body mass” to Human body weight).

Plan 2 passed. In a fresh blind sample, 36 of 40 links were correct (90%). Four were judged too broad and none wrong; all three planted errors were caught. The four broad cases share one limit of a name-only match: the article fits the name but not the field’s sense, as with “noise” or “effective temperature” in biophysics.

Coverage. Galaxies 42 of 208 quantities (20%), life sciences 114 of 765 (15%), medicine 106 of 224 (47%). Most unmatched names are compound quantities, such as “dark matter halo mass function”, for which no single article exists. Each unmatched quantity shows its reason on the map.

7.2 Quantities, Laws and Paths to the New MAZE

• By title. A quantity whose Wikipedia title appears in a node’s title list is placed under that node. This is the same title identity that links the new MAZE to the old. Nine quantities meet nodes this way: time, space, the Schwarzschild radius and the number of neurons are examples.
• By dimension. A rule exists to place a quantity by the unit stated in the paper, under the nodes for mass, space, time and charge (winding 4) and on the ladder (winding 12). The maps carry no units, so this rule places nothing yet. It is not bypassed with a dictionary of concepts.
• By form. Every law sits under lus_helling, because a law is a slope. Every path sits under lus_overal, because a path either closes or carries tension.

7.3 Papers as Residents

The papers themselves become residents of the old MAZE: the 645 arXiv papers and the 1,303 PubMed records. They pass the same gate as any other item.

The address comes from the content. The receipt names the arXiv identifier or PMID. Nothing is overwritten. Before any writing, 50 random papers were addressed without storing; all 50 received distinct addresses (threshold 90%).

On the maps, α behind a paper opens it at its address in the net, among its neighbours.

The result is a two-way thread. From a dot on a map the reader reaches the Wikipedia resident, its neighbourhood in meaning, and the derived node where one exists. From a paper the reader reaches its place in the net. The Transformer has become the entrance through which a research field enters both MAZEs.

---

8. Limitations

• The assessor is a model. All blind checks were done by a language-model agent. Planted errors and spot checks control for this only in part. A human expert in each field would be a stronger judge.
• Language models in the loop. Naming, law filtering and effect binding use a language model. Each step is cached, run at temperature 0 and tested blind. The model can still share blind spots with the assessor.
• Rules derived on seen data. The reading rules L13–L19 were derived from the same papers they now read. Only the test samples were fresh.
• Physics paths untested. No paper in the current corpora measures a chained pair directly.
• Medicine paths are not causal. They predict directions, not mechanisms, and they mix populations.
• Name-only links. A Wikipedia title can fit a name but not the field’s meaning. Coverage is limited for compound quantities.
• The theory. The Vacuum.Net theory supplies the vocabulary: place, passage, trit, path, closure. The Transformer’s results do not test the theory’s physical claims. They test whether that vocabulary organises published claims in a useful way.

---

9. Reproducibility

All code, plans, blind samples, keys and outcomes are in the SWARP repository under maze/transformer.

Main files:

• Layer A: v1.1/lees.py and kern.py
• Layer B: LayerB-v1.4/Layerb.js and ai/
• MAZE links: koppeling.mjs and artikelen-opname.mjs

Each plan has a .sha256 file committed before its run. Each outcome has an uitkomst-*.md file.

The live API is documented at maze.swarp.nl/handleidingen/transformer. Its main routes are /api/transformer/kaart, /api/transformer/geneeskunde and /api/transformer/koppeling?kaart=.

---

Annotated References

1. Konstapel, J. (2026). The Vacuum.Net Theory: Foundational Paper, 8th ed., v8.1. Constable Research, Leiden. constable.blog.
   The canonical statement of the theory: axioms N1–N5, theorems T1–T23, correspondences, predictions and open points, with a register of checks. Read it for the status of each claim.
2. Konstapel, J. (2026). Curves in the Vacuum.Net, v1.0. Constable Research, Leiden.
   Derives what a path in the net is. It supplies the reading of a law as a slope and of a path as a chain of slopes.
3. Spencer-Brown, G. (1969). Laws of Form. London: Allen & Unwin.
   The calculus of distinction from which the theory starts. Read it for the idea that one act of distinction, re-entering itself, generates a logic.
4. Kauffman, L. H. (1987). Self-reference and recursive forms. Journal of Social and Biological Structures, 10(1), 53–72.
   Connects re-entry, eigenforms and imaginary values. It underlies the theory’s step from distinction to oscillation.
5. Rowlands, P. (2007). Zero to Infinity: The Foundations of Physics. Singapore: World Scientific.
   The nilpotent form of the Dirac equation that the Foundational Paper adopts. Read it to see where the theory takes Dirac as input.
6. Larson, D. B. (1959). The Structure of the Physical Universe. Portland: North Pacific.
   The space-time dimensions used for the theory’s units. A historical, unorthodox source; read it critically.
7. Knuth, D. E. (1997). The Art of Computer Programming, Vol. 2: Seminumerical Algorithms, 3rd ed., §4.1. Reading, MA: Addison-Wesley.
   The standard treatment of balanced ternary, the number system behind every MAZE address.
8. Hayes, B. (2001). Third base. American Scientist, 89(6), 490–494.
   A readable account of why base three is the most economical radix. It explains the trit as an engineering choice.
9. Barenblatt, G. I. (1996). Scaling, Self-Similarity, and Intermediate Asymptotics. Cambridge University Press.
   Why power laws are the natural form of quantitative claims in physics. Background for Layer A.
10. West, G. B., Brown, J. H., & Enquist, B. J. (1997). A general model for the origin of allometric scaling laws in biology. Science, 276(5309), 122–126.
    The classic example of power laws in biology. It explains why the life-sciences map exists at all.
11. Swanson, D. R. (1986). Fish oil, Raynaud’s syndrome, and undiscovered public knowledge. Perspectives in Biology and Medicine, 30(1), 7–18.
    The origin of literature-based discovery: A → B in one literature and B → C in another imply A → C. The Transformer’s paths are a quantitative version of this idea.
12. Smalheiser, N. R., & Swanson, D. R. (1998). Using ARROWSMITH: a computer-assisted approach to formulating and assessing scientific hypotheses. Computer Methods and Programs in Biomedicine, 57(3), 149–153.
    The software successor to Swanson’s method. Compare its word-based links with the Transformer’s exponent- and effect-based links.
13. Tshitoyan, V., et al. (2019). Unsupervised word embeddings capture latent knowledge from materials science literature. Nature, 571, 95–98.
    Shows that relations hidden across many papers can predict later findings. A modern counterpart that works with embeddings instead of explicit claims.
14. Fortunato, S., et al. (2018). Science of science. Science, 359(6379), eaao0185.
    An overview of quantitative study of the scientific literature. It places the Transformer within that field.
15. DerSimonian, R., & Laird, N. (1986). Meta-analysis in clinical trials. Controlled Clinical Trials, 7(3), 177–188.
    The random-effects method for pooling effect sizes. Background for how Layer B combines effects on one edge.
16. Higgins, J. P. T., & Thompson, S. G. (2002). Quantifying heterogeneity in a meta-analysis. Statistics in Medicine, 21(11), 1539–1558.
    Defines I², which the medicine map shows for edges with more than one effect.
17. Davey Smith, G., & Ebrahim, S. (2003). ‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease? International Journal of Epidemiology, 32(1), 1–22.
    The design that could give medicine edges a causal direction. The planned next step for Layer B.
18. Lipscomb, C. E. (2000). Medical Subject Headings (MeSH). Bulletin of the Medical Library Association, 88(3), 265–266.
    A short history of the vocabulary that supplies Layer B’s places.
19. Lehmann, J., et al. (2015). DBpedia: a large-scale, multilingual knowledge base extracted from Wikipedia. Semantic Web, 6(2), 167–195.
    The source of the article summaries that the old MAZE uses to rank residents by meaning.
20. arXiv (2026). arXiv API User’s Manual and bulk source access (arxiv.org/e-print). NCBI (2026). Entrez Programming Utilities.
    The two public interfaces through which all papers in the Transformer were retrieved.

---

Code, plans and data: SWARP repository, maze/transformer. The maps: maze.swarp.nl/transformer. This essay as PDF: maze.swarp.nl/transformer/essay.pdf.

Jump to the article here. DOI: 10.13140/RG.2.2.20919.89765

J.Konstapel,Leiden,27-10-2026

Cause: Leidse onderzoeksprojecten ontvangen Open Competitie financiering voor vernieuwend onderzoek

Tuberculosis remains the deadliest infectious disease
attributable to a single pathogen.

Our only licensed vaccine, BCG, is a century old.

It protects infants against severe, disseminated forms of childhood disease.

It does not reliably protect adolescents and adults
against pulmonary tuberculosis —the very form that keeps the epidemic alive.

The field’s own literature states the obstacle plainly.

Vaccine development is held backbecause we do not know what protection against
Mycobacterium tuberculosis (Mtb) actually requires
(Dijkman, Coler and Joosten, 2022).

That is a remarkable admission from a mature discipline.
Modern immunology can enumerate immune cells,
sequence their receptors, measure their cytokines,
and track them over time with spectral flow cytometry.
What it cannot do is say
which of these measurements means “protected”.

The problem is not a shortage of numbers.
The problem is that the numbers do not converge
on a quantity that behaves like protection.

A second anomaly sits beside the first.
Roughly ninety-five per cent of people infected with Mtb
never develop the disease.
About five per cent do — sometimes decades later.
The protected majority are not cured;
the organism persists.
They are not ill either.
They occupy a state that is stable for a lifetime
and yet is not resolution.
The field names this state awkwardly: “latency”.
The clinical boundary between latent infection and active disease
is notoriously difficult to draw.
A system that holds an unresolved condition stable for forty years
is doing something
that the language of presence and absence
does not describe well.

2. The empirical material: the Leiden programme

The source text builds its argument on a concrete body of work:
the research programme on human anti-mycobacterial immunity
at the Leiden University Center for Infectious Diseases,
associated in particular with Simone A. Joosten.
Four features of that programme matter.
None of them is unusual.
Each is a move away from component thinking,
made for empirical reasons,
inside a field still organised around components.

2.1 A reading channel that is the same for everyone

Classical antigen presentation runs through
HLA class Ia and class II molecules —
among the most polymorphic loci in the human genome.
Any vaccine that depends on them
depends on a lock that differs from person to person.

HLA-E is different.
It is virtually monomorphic.
It presents both self-derived and pathogen-derived peptides.
It is not downregulated in HIV co-infection.
T cells that recognise HLA-E-presented peptides
are therefore called donor-unrestricted:
the restricting element barely varies across the population
(Ruibal et al., 2021; Voogd et al., 2024b).

The significance is architectural, not merely immunological.
Beside the individual-specific reading layer
runs a second reading layer invariant across the species.
Vaccine design aimed at HLA-E
is design aimed at the invariant layer.
Design aimed at classical HLA
is design aimed at the variable one.

2.2 Many paths, one connection point

When CD8+ T cells recognising HLA-E tetramers
folded with Mtb-derived peptides
were sorted from twenty-one individuals
and their receptors sequenced,
the repertoire proved highly diverse
(Voogd et al., 2024a).
One nearly invariant presenting molecule;
an open-ended variety of receptors arriving at it.

Earlier work had already shown
that these cells’ functional phenotype
is unorthodox relative to classical class I-restricted T cells:
they can secrete cytokines
associated with both conventional helper programmes.

Read as a diagram rather than a mechanism,
this is a connection point
with many curves terminating on it.
The point is fixed; the approaches are not.
The field calls this a repertoire-characterisation problem.
It can equally be described as the ordinary situation
when a system is organised around closure:
what has to be the same is the junction,
not the route taken to reach it.

2.3 Function rather than quantity

In tuberculosis, antibody titres
have long been a poor guide to anything.
A recent consensus review from the same group
reframes the question around functional capacity —
what antibodies do rather than how much of them is present —
with glycosylation and effector function
carrying the discriminating information
(van Meijgaarden et al., 2026;
see also Rijnink, Ottenhoff and Joosten, 2021).

Joosten’s own statement of method is explicit:
evaluate the sum of immune responses
rather than isolated components,
and assess that sum from a functional perspective.
This is the methodological core of the whole argument,
stated by a practitioner on empirical grounds.
A quantity measured at a point
is being replaced by a property of a completed relation.
The move is forced by the data:
the component measurements did not behave.
The functional ones do.

2.4 The granuloma as a held state

Within the granuloma, the local architecture of cellular interaction
tracks with outcome:
interactions between natural killer cells and macrophages
correlate with limited pathology
(Niewold et al., 2025).

The granuloma is not a storage container for bacteria.
It is a configuration under tension
that neither resolves into clearance
nor collapses into disease.
It is maintained by continuous activity,
and it can persist for decades
before doing one or the other.

3. The reading: protection as closure

The source text borrows a small, positional vocabulary
from the Vacuum.Net framework:
points in a position space,
curves running between them,
and connection points where curves meet
and a path can close.
No physical claim is required.
The vocabulary is used as a geometry of relation.

In that vocabulary:

• A component measurement is the measurement of a point.
• A functional assay is the measurement of a segment of curve.
• Protection — if the reading is right — is neither.
  It is the property that a particular path closes.

Closure is not located anywhere on the path.
It is not the sum of the properties of the points,
and it cannot be recovered
by measuring the points more precisely,
because precision at a point
says nothing about whether the circuit returns.

Four consequences, all observed

If that is what protection is,
four things follow immediately.
All four are observed.

First, no single correlate of protection will be found,
because the thing being correlated
is not a quantity carried by any element.
A century of negative results
is then not a record of insufficient resolution
but a correct empirical answer to a malformed question.

Second, the same intervention
will protect in one host state and not in another.
Closure is a property of the path as a whole,
and the path includes the state of the system
at the moment of the attempt.
BCG in an infant and BCG in an adolescent
are not the same path with the same agent.
They are different paths.
The field’s central vaccine anomaly
becomes, on this reading, not an anomaly at all.
The related observation —
that repeated exposure to Mtb and to environmental mycobacteria
shapes subsequent host responses
and may shift vaccine efficacy in either direction —
belongs to the same point:
the history of the path is part of the path.

Third, invariant presentation elements
will be unusually productive targets,
because they are the connection points
that every individual shares.
This is the design logic of the HLA-E programme,
arrived at independently.

Fourth, latency is neither a weak form of disease
nor a cured state.
It is a path held open under tension that has not discharged.
Disease is the discharge.
This removes the need to treat
the latent-active boundary as a diagnostic defect.
The two are not two states
but two positions on one curve —
which is why no test cleanly separates them.

4. What this is, and what it is not

Nothing in the Leiden work refers to any of this.
No part of the argument is evidence for the Vacuum.Net framework.
What is shown is narrower,
and, in the framework’s own terms,
is the only thing that was being asked:
a well-developed experimental programme
can be restated without remainder
in the language of points, curves and connection points;
and when it is restated,
its two standing anomalies —
the missing correlate and the age-dependence of BCG —
stop being anomalies and become consequences.

Restatement is cheap if it only relabels.
The test of whether it does more
is whether it forbids anything.
This one does.

5. A consequence that can fail

The reading issues a prohibition
rather than a prediction of the usual kind.

It holds that no component measured at a single point —
no titre, no cell frequency,
no transcriptional signature evaluated as a quantity —
will ever function as a general correlate of protection against Mtb
across host states.
Every apparent success of this kind
will turn out, on extension,
to be a correlate of a particular path
in a particular population.

Conversely, measures constructed as closure measures —
whether a given effector relation completes,
assayed functionally,
across the whole response rather than within one arm of it —
will generalise where the component measures do not.

This is falsifiable in the ordinary way.
A single robustly transferable component correlate,
holding across ages, co-infections,
prior exposure histories and populations,
would refute it.
The field has searched for exactly that
for a hundred years without finding it —
weak evidence, admittedly.
The argument is that the absence is structural
and should be expected to continue.

There is a constructive side to the same statement.
If protection is closure,
then the design question for a vaccine
is not which antigen to present
but which junction to make reachable
from the greatest number of starting states.
An invariant, population-wide presentation element
with a diverse set of routes converging on it
is precisely such a junction.
That is an argument for the HLA-E direction
that does not depend on any mechanism specific to HLA-E —
and it would carry over
to any other invariant element that is found.

6. Summary for the practitioner

Three sentences carry the essay’s practical weight.

1. A century of failed searching for a correlate of protection
   is better read as a correct answer
   to a malformed question
   than as a failure of measurement resolution.
2. Functional, relation-level assays
   should be expected to generalise
   where point-level quantities do not;
   the field’s own methodology is already moving this way.
3. Vaccine design should target invariant junctions
   reachable from many starting states —
   a design principle, not a mechanism claim.

---

Annotated references

The following list reproduces the sources cited in the original text,
with a short annotation on each.
Where a source underwrites a specific claim above,
the annotation says which one.

Dijkman, K., Coler, R. N., and Joosten, S. A. (2022).
Editorial: Beyond Th1 — novel concepts in tuberculosis vaccine immunology.
Frontiers in Immunology, 13, 1059011.
An editorial statement of the field’s central admission:
progress in tuberculosis vaccine development
is constrained by the absence of knowledge
about what protection actually requires.
It supplies the opening premise of the argument —
that the gap is conceptual, not merely technical.

van Meijgaarden, K. E., Grace, P. S., Freund, N. T., Achkar, J. M.,
Lindenstrøm, T., Tan, J., Chan, J., King, C. G., Ottenhoff, T. H. M.,
and Joosten, S. A., on behalf of the CTVD B cell group (2026).
Antibodies in tuberculosis: functional capacity as key determinant.
Infection and Immunity, 94, e00394-25.
A consensus review reframing humoral immunity in tuberculosis
around what antibodies do rather than how much is present.
It is the clearest published instance
of quantity being replaced by function —
the methodological move the essay generalises.

Niewold, P., IJsselsteijn, M. E., Dijkman, K., de Miranda, N.,
Ottenhoff, T. H. M., Verreck, F. A. W., and Joosten, S. A. (2025).
NK cell–macrophage interactions in granulomas correlate with limited
tuberculosis pathology. bioRxiv, 2025.02.14.638234.
Evidence that the granuloma’s cellular architecture —
specifically NK–macrophage interaction —
tracks with clinical outcome.
It grounds the reading of the granuloma
as a held configuration under tension
rather than a passive containment vessel.

Rijnink, W. F., Ottenhoff, T. H. M., and Joosten, S. A. (2021).
B-cells and antibodies as contributors to effector immune responses
in tuberculosis. Frontiers in Immunology, 12, 640168.
An earlier statement of the position
that antibodies contribute functionally to tuberculosis immunity,
predating and supporting the 2026 consensus review.

Ruibal, P., Voogd, L., Joosten, S. A., and Ottenhoff, T. H. M. (2021).
The role of donor-unrestricted T-cells, innate lymphoid cells,
and NK cells in anti-mycobacterial immunity.
Immunological Reviews, 301(1), 30–47.
The programme’s overview of donor-unrestricted T-cell biology.
It establishes the biological facts behind
the “invariant reading layer”:
HLA-E’s near-monomorphism and its stability in HIV co-infection.

Voogd, L., Drittij, A. M. H. F., Dingenouts, C. K. E.,
Franken, K. L. M. C., van Unen, V., van Meijgaarden, K. E., Ruibal, P.,
Hagedoorn, R. S., Leitner, J. A., Heemskerk, M. H. M., Davis, M. M.,
Ottenhoff, T. H. M., and Joosten, S. A. (2024a).
Mtb HLA-E-tetramer-sorted CD8+ T cells have a diverse TCR repertoire.
iScience.
The tetramer-sorting and sequencing study
showing one nearly invariant presenting molecule
served by a highly diverse receptor repertoire.
The empirical basis for the
“many paths, one connection point” observation.

Voogd, L., Riou, C., Scriba, T. J., van Wolfswinkel, M.,
Ottenhoff, T. H. M., and Joosten, S. A. (2024b).
HLA-E/Mtb specific CD4+ and CD8+ T cells have a memory phenotype
in individuals with TB infection.
Frontiers in Immunology, 15, 1505329.
Further characterisation of HLA-E-restricted, Mtb-specific T cells,
including their memory phenotype in infected individuals.
Supports the invariant-channel reading
together with Ruibal et al. (2021).

Konstapel, J. (2026). Curves in the Vacuum.Net.
constable.blog, 24 October 2026.
The author’s prior exposition of the framework vocabulary
of points, curves and connection points
used as the essay’s geometry of relation.

Konstapel, J. (2026). Where Illness Comes From:
The Net from First Chemistry to the Human. Eighth edition.
The longer work whose invariant-versus-variable distinction
the essay notes was reached from an entirely different direction
than immunology —
and which the Leiden programme’s architecture
independently mirrors.

---

Article

Jump to the Article with DOI: 10.13140/RG.2.2.17105.70249 here

J.Konstapel,Leiden,27-10-2026.

Cause: Leidse onderzoeksprojecten ontvangen Open Competitie financiering voor vernieuwend onderzoek

1. Introduction: the question of the round trip

Mathematics and physics keep returning to a single, deceptively simple question: if you know everything about how something comes back to where it started, can you reconstruct the thing itself? A closed curve — a loop, a period, an orbit — records its own history in the values it accumulates as it travels. The mathematical problem of inverse periods asks whether that record is complete: whether the period matrix of an algebraic curve, the array of integrals of its holomorphic differentials over each of its closed loops, determines the curve up to isomorphism.

This question is not idle. It sits at the junction of arithmetic geometry (where periods have been computed and classified since the nineteenth century), of mathematical physics (where the periods of special curves encode the masses and couplings of quantum field theories), and, more speculatively, of any worldview that treats persistence itself — the fact that a thing keeps coming back — as the fundamental property worth explaining.

The occasion for this essay is a concrete one: the mathematician Marco Streng of Leiden University has received NWO Open Competitie funding for a project named ReSCue, which asks exactly this reconstruction question for superelliptic curves — curves of the form y^n = f(x). This essay reads Streng’s body of work as a coherent programme about closure, connects it to a speculative framework called the Vacuum.Net, in which everything that persists is a closed path, and presents one new computational observation: the framework’s ternary unit of address, the trit, turns out to arise naturally in the Eisenstein integers — the very lattice that governs Streng’s Picard curves.

2. The central mathematical problem

2.1 Periods and Torelli’s theorem

Start with the classical situation. A smooth complex algebraic curve C of genus g is, analytically, a surface with g holes. It has 2g independent closed loops, and g linearly independent holomorphic differentials \omega_1, \dots, \omega_g. Integrating each differential over each loop produces a g \times 2g matrix of complex numbers — the period matrix. Torelli’s theorem states that this matrix, suitably normalised, determines the curve completely. In other words: the record of round trips already contains the path.

Torelli’s theorem, however, is an existence statement, not an algorithm. Given an approximate period matrix — measured, say, to a finite precision — how does one recover the curve, with its exactly-defined algebraic coefficients? This is the inverse problem: practical, quantitative, and wide open in general. For elliptic curves (genus 1) the answer is classical and constructive, running through the theory of complex multiplication. For genus 2 and 3, Streng and his collaborators have spent a decade making it concrete. ReSCue extends the question to superelliptic curves of the form y^n = f(x), whose extra rotational symmetry both constrains and enriches the period lattice.

2.2 Complex multiplication: symmetry of the lattice

A complex elliptic curve is a torus: the complex plane modulo a lattice \Lambda. Generically, the only symmetries of such a torus are multiplication by \pm 1. But for special lattices — the hexagonal one, the square one, and countably many others — the torus admits extra rotations: endomorphisms beyond \mathbb{Z}. This is complex multiplication (CM), and it is the arithmetic key to explicit reconstruction. A CM elliptic curve is not described by a transcendental period but by an algebraic invariant, the j-invariant, which is an algebraic integer satisfying a class polynomial whose degree is the class number of an imaginary quadratic field.

The moral is the one that runs through everything discussed here: symmetry turns analysis into arithmetic. The period matrix of a generic curve is an unruly transcendental object; the period matrix of a CM curve is governed by a lattice with rotational symmetry, and the reconstruction problem collapses into a finite, computable algebraic task. The same principle powers genus 2 (where the relevant invariants are Igusa class polynomials), genus 3 (Streng’s Picard and quartic curves), and much of modern computational number theory.

2.3 The table of Streng’s programme

Read as a whole, Streng’s seventeen listed publications fall into five operations on closed paths — a taxonomy worth making explicit:

Operation on a closed path	Representative papers	Interpretation
Giving a closed path a basis-free address	Igusa class polynomials (2014); generalized class polynomials (2022, with Houben); Schertz-style class invariants for higher-degree CM fields (2026, with Enge)	The j-invariant and its higher-genus analogues name a torus independently of any chosen basis — an address without a starting point
Counting paths that share one rotational symmetry	Polarised class groups of quartic CM orders (2017, with Bisson); CM class number one for genus 2 (2023, with Kılıçer)	Complex multiplication is a turn that maps the lattice into itself; the class number counts how many distinct lattices share that turn; class number one means a unique closure
Finding the step at which a point returns	Divisibility sequences for CM elliptic curves (2008); algebraic divisibility sequences over function fields (2012); primitive divisors with constant j-invariant (2020, with Naskręcki); modular units for \Gamma_1(N) (2023)	A primitive divisor of the n-th term of a divisibility sequence is a prime modulo which a point first closes exactly at step n; \Gamma_1(N) describes the structure of points that close after N steps
Finding where a loop pinches	Primes dividing invariants of CM Picard curves (2019); a bound on bad-reduction primes for genus-3 CM curves (2020); plane quartics over \mathbb{Q} with CM (2018)	Bad reduction is a connection point: modulo that prime, two strands of the path fall together and the loop degenerates
Using closure as a working tool	CM construction with p-rank 1 (2011); prescribed embedding degree (2008); curves in characteristic 2 with 2-torsion (2014); (\ell,\ell)-endomorphisms (2013); CM curves over the reflex field (2015)	Closure arithmetic already runs as practical machinery — the cryptographic papers are the engineering wing of the same programme

The row that matters most for what follows is the fourth and fifth: the family of Picard curves y^3 = f(x) with \deg f = 4. These genus-3 curves carry a built-in threefold rotational symmetry, (x, y) \mapsto (x, \omega y), where \omega is a primitive cube root of unity. Their period lattices are modules over the Eisenstein integers \mathbb{Z}[\omega] — and it is precisely there that a ternary arithmetic of closed paths lives.

3. The Vacuum.Net: a framework of closure

To explain why this matters beyond arithmetic geometry, a short detour into the speculative part.

The Vacuum.Net (as developed by J. Konstapel) is a proposed reading of physics in which everything that persists is a closed curve: a path that returns to where it started. Points are locations in a position space; curves run through them; where curves meet there is a connection point. The slogan is: a thing is a turn that keeps coming back. On this reading, what we call a particle is not a small object but a stable recurrence — an address that loops.

Two structural consequences follow, and only these two are needed here.

First, the address of a closed path has no starting point. A loop is a rotation class: reading its description starting from any point on the loop gives the same path. Therefore the number of distinct addresses of length n over a three-letter alphabet is not 3^n but the necklace countg(n) = \frac{1}{n} \sum_{d \mid n} \varphi\!\left(\frac{n}{d}\right) 3^{d},

the number of circular words up to rotation — for instance g(3) = 11, not 27.

Second, the unit of the address is the trit: a digit from \{-1, 0, +1\}, read off from balanced ternary rather than assigned by convention. The question this essay adds to the framework is: where does the trit come from? The answer, developed in the next section, is that it is not a convention at all. It is the natural residue in the Eisenstein integers — the lattice behind the Picard curves.

4. The trit in \mathbb{Z}[\omega]: a computational result

4.1 The residue

Let \omega = e^{2\pi i/3} be a primitive cube root of unity and let \beta = 1 – \omega, an element of norm N(\beta) = 3. The quotient \mathbb{Z}[\omega]/(\beta) is the field with three elements, and since \omega \equiv 1 \pmod{\beta}, the class of a + b\omega is simply a + b \pmod 3. The natural representatives of these three classes are -1, 0, +1 — exactly the balanced-ternary digits. In the Eisenstein integers, the trit is not a design choice; it is the residue modulo the prime above 3.

4.2 The digit map and its dynamics

Define the balanced residue d(z) \in \{-1, 0, +1\} and the digit mapT(z) = \frac{z – d(z)}{\beta}.

Iterating T produces a balanced expansion in base \beta:z = d_0 + d_1\beta + d_2\beta^2 + d_3\beta^3 + \dots, \qquad d_k \in \{-1, 0, +1\}.

A computational experiment — running T on every a + b\omega with |a|, |b| \le 100 (40,401 elements) — shows that every orbit ends in one of exactly three places: at 0, or at one of the two fixed points \pm(1+\omega). Moreover, the proportion ending at a fixed point rather than at zero grows with the size of the box: 27% for |a|, |b| \le 40 (1,788 of 6,561) and 41% for |a|,|b| \le 100 (16,546 of 40,401). Whether these shares converge to a limit is open.

The result holds for all of \mathbb{Z}[\omega], and the proof is short. The norm drops at every step:N(T(z)) = \frac{N(z – d)}{3} \le \frac{(\sqrt{N(z)} + 1)^2}{3},

which is smaller than N(z) whenever N(z) \ge 2. So every orbit falls into the elements of norm 0 or 1: the zero and the six units. Checking those seven by hand: 0, \pm 1, \pm\omega run to 0, while \pm(1+\omega) map to themselves.

4.3 The meaning of the fixed point

The fixed point has a clean description: 1 + \omega = -\omega^2 = e^{i\pi/3}, a sixth root of unity — a 60° turn. Its balanced expansion is the digit -1 repeated forever, since -\omega^2 = -1 + \beta \cdot (-\omega^2). The ordinary number 2 already exhibits the dichotomy of the dynamics: 2 \to 2 + \omega \to 1 + \omega, and from there the turn never stops. The number 1, by contrast, comes to rest in a single step.

The reading in the framework is this: the ternary ground, placed inside the Eisenstein lattice, reproduces the Vacuum.Net’s basic dichotomy unprompted. A finite trit word comes to rest at zero — a closure. An eventually periodic word does not rest: it ends in a unit that keeps turning, a turn that keeps coming back. Rest and return, the two states the framework posits, fall out of the arithmetic of \mathbb{Z}[\omega] itself. (Taking -\beta = \omega – 1 as the base instead merges the two fixed points into a single 2-cycle — the same turn seen from the other side.)

5. ReSCue as the bridge

5.1 From the framework to the project

For n = 3, a superelliptic curve y^3 = f(x) is a Picard curve, and (x, y) \mapsto (x, \omega y) is an automorphism of order 3. This automorphism acts both on the loops of the curve and on its differentials, so the period lattice is not merely a \mathbb{Z}-module but a module over \mathbb{Z}[\omega]. Since \mathbb{Z}[\omega] is a principal ideal domain, that lattice has a basis over \mathbb{Z}[\omega], and the digit map of Section 4 applies to the coordinates of every period in that basis: each coordinate has a balanced-trit expansion in base 1 – \omega, either finite or ending in the 60° turn. Whether this ternary address is a useful coordinate for reconstruction — in lattice reduction, or in deciding to what precision periods must be known before reconstruction succeeds — is a concrete and testable question.

5.2 From the project to the framework

The dependence runs both ways. The Vacuum.Net needs exactly the procedure ReSCue is building: given only measured closures, rebuild the curve. A text, a dataset, a physical measurement — each arrives as numbers, not paths. If periods determine the curve, then closures determine the path, and the framework’s position space can be reached from data rather than postulated. Reconstruction is the route from the world of records to the world of paths.

5.3 The physics already on the bridge

The bridge to physics is not hypothetical; three results anchor it.

• Seiberg–Witten theory (1994). A four-dimensional N=2 supersymmetric Yang–Mills theory is encoded by an algebraic curve, and its periods are the physical observables: the masses of stable particles and the coupling constants. The exact mass formulas of the abstract follow from the periods of the Seiberg–Witten curve. Here the period matrix is not a bookkeeping device; it is the theory’s measurable content.
• Moore’s “Arithmetic and Attractors” (1998). For N=4 and N=8 string compactifications, the attractor varieties of supersymmetric black holes are built from elliptic curves with complex multiplication, and for N=4 the number of inequivalent backgrounds with the same horizon area is precisely the class number of an imaginary quadratic field — the quantity Streng’s class-number papers compute. Supersymmetric black holes, in other words, are counted by the same arithmetic that counts closures.
• Gukov–Vafa (2002). The rational conformal field theories with a two-torus target are exactly those whose torus has complex multiplication. The consistency of a physical theory on a torus is equivalent to an arithmetic condition on the torus’s lattice.

In each case the physics sits on closed curves, accessed through their periods or their complex multiplication — the two objects that ReSCue and Streng’s earlier work handle. ReSCue already stands on a bridge to physics; it has simply not been named as such. The Vacuum.Net offers a third bank: a ternary address for the same periods, and a reading in which rest and return are the only two states.

6. Three open questions

Each question below can be checked in an afternoon with Sage or PARI/GP, and each answer, either way, is informative.

1. A trit address for periods. For Picard curves, does the balanced expansion in base 1 – \omega give a useful coordinate on the period lattice — for instance in lattice reduction, or in deciding to what precision periods must be known before reconstruction succeeds? And which period vectors end in the 60° turn rather than at rest?
2. First closures and necklace counts. In the divisibility sequences attached to a CM curve, the index at which a prime first appears is the step at which a point first closes. Do these indices, gathered over many primes, follow a distribution comparable to the necklace count g(n) = \frac{1}{n}\sum \varphi(n/d)\,3^d — especially for curves with CM by \mathbb{Z}[\omega]?
3. Reconstruction as a general procedure. If ReSCue yields an algorithm from period matrix to curve, could the same algorithm take any set of measured closures — not only periods of an algebraic curve — and return the simplest closed path that produces them?

7. Conclusion

The wager of this essay is that three things usually kept apart — a computational number theory programme (class polynomials, divisibility sequences, bad reduction), a reconstruction project (ReSCue), and a speculative physics of closure (the Vacuum.Net) — are one subject seen from three distances. The arithmetic gives the framework its digits and its dichotomy of rest and return; the framework gives the arithmetic a structural vocabulary; and ReSCue, if it succeeds, gives both a procedure: from the record of round trips back to the path that made them. What a closed curve is, in the end, may be exactly what its periods can reconstruct.

Annotated references

Mathematical sources

• Streng, T.C., publication list, Universiteit Leiden. The seventeen papers cited in the taxonomy table above (2008–2026), spanning class polynomial computation for genus 2 and 3, CM class number problems, divisibility sequences, bad-reduction bounds for Picard curves, and cryptographic CM constructions. Read together they form a coherent programme on the invariants, first-returns and degenerations of closed curves with rotational symmetry; the present essay’s taxonomy of “five operations on closed paths” is derived from this list.
• Universiteit Leiden, “Leidse onderzoeksprojecten ontvangen Open Competitie financiering”, 26 October 2026. The announcement that establishes ReSCue as an NWO-funded project on recovering superelliptic curves y^n = f(x) from their period matrices. The immediate occasion for this essay; the reconstruction problem of Section 2 and the bridge of Section 5 are framed around its stated aims.

Framework sources

• Konstapel, J., “Curves in the Universe” (PDF title “Curves in the Vacuum.Net”, v1.0), constable.blog, 24 October 2026, updated 26 October 2026. The longer statement of the Vacuum.Net framework: points in position space, curves, knots and seams, and the introduction of trits and balanced ternary as the unit of address. Sections 3 and 4 of this essay assume only its two consequences: the rotation-class address and the trit.
• Konstapel, J., “The Positron and the Cogwheel”, constable.blog, 26 October 2026. The companion piece working out the address of a closed path as a rotation class (a necklace), including the worked example g(3) = 11. Source for the necklace count g(n) used in Sections 3 and 6.

Physics sources (each checked against its arXiv record on 27 October 2026)

• Seiberg, N. and Witten, E., “Electric-magnetic duality, monopole condensation, and confinement in N=2 supersymmetric Yang–Mills theory”, Nuclear Physics B 426 (1994) 19–52; arXiv:hep-th/9407087. The founding paper of Seiberg–Witten theory: an N=2 field theory is encoded in an algebraic curve whose periods give exact particle masses and couplings. Cited in Section 5.3 as the paradigm case in which periods are directly physical observables; the exact mass formulas appear in the abstract, their derivation from the periods in the body.
• Moore, G., “Arithmetic and Attractors”, arXiv:hep-th/9807087 (1998). Shows that attractor varieties of supersymmetric black holes in N=4 and N=8 compactifications are built from CM elliptic curves, and that for N=4 the degeneracy of inequivalent backgrounds sharing one horizon area equals the class number of an imaginary quadratic field. Cited in Section 5.3 as the bridge between black-hole physics and the class-number arithmetic of Streng’s genus-2 programme.
• Gukov, S. and Vafa, C., “Rational Conformal Field Theories and Complex Multiplication”, arXiv:hep-th/0203213 (2002). Establishes that rational CFTs with two-torus target are precisely those whose torus has complex multiplication. Cited in Section 5.3 as the cleanest known equivalence between a physical consistency condition and an arithmetic property of a lattice.

Computational source

• Computation of Section 4: a script over all a + b\omega with |a|, |b| \le 100, run 27 October 2026. Exhaustive iteration of the digit map T(z) = (z – d(z))/\beta over 40,401 Eisenstein integers, recording the terminal state (zero or the fixed points \pm(1+\omega)) of each orbit, with growing fixed-point share (27% at box 40, 41% at box 100). The norm-descent argument of Section 4.2 makes the two end states general for all of \mathbb{Z}[\omega]; the convergence of the shares remains open.

Article

J.Konstapel,Leiden,27-10-2026.

The book asks a single question — what is a city for? — and follows the answers across 7,500 years, from a mud-brick room on a sand dune in Eridu to a Rome shared with participants that have no bodies. Its ambition is not encyclopaedic but argumentative: twenty chapters, each with the same five-part anatomy (the idea, the builders, the art, the measured curve, the lines forward), converging on a comparative synthesis (chapter 19) and a proposal (chapter 20).

I. The return, not the foundation

The book’s first and most quietly radical claim is that the original act of urbanism was not building but returning. Eridu, founded around 5400 BCE on a rise where river, sea and groundwater met, was for two millennia a sequence of eighteen temples, each raised on the ruins of the last. The first city, in Konstapel’s reading, was not a wall, a market or a king, but “a place that is returned to” — the house of Enki, god of the deep fresh water, built and rebuilt by generations who left no names. The measured evidence is thin and honestly presented as such: nine floor plans growing from a room the size of a human body (7.8 m²) to a building the size of a house (~200 m²), read as a “word of steps” — rises, standings, falls — that the author treats as data rather than decoration.

From this beginning the book draws a line that sustains its entire architecture: the idea of the city travels. In the Sumerian myth of Inanna and Enki, the gifts of civilisation are carried off from Eridu to Uruk; in historical fact, Uruk then grows into the world’s first megacity and invents writing to administer itself. Chapter 2’s most elegant observation is that during Uruk’s demographic pause (c. 3300–3000 BCE) the number of proto-cuneiform tablets multiplied sixteen-fold. “The standing was not a rest: it was the time the city built its memory.” The plateau, elsewhere a demographic statistic, becomes here a category of urban creativity.

II. The curve as method

The book’s signature instrument is the “measured curve.” For each city, Konstapel assembles a population or size series (often from the standard compilers, Chandler and Modelski, sometimes from excavation reports) and reduces it to a word of steps — + + 0 +, 0 0 – ×8 0 0 0 0 + ×8 for Rome — using a fixed rule: a change of less than roughly 11 per cent in the relevant measure counts as “standing.” Chapter 19 lays the curves side by side and extracts five regularities, of which three deserve an intellectual reader’s attention.

First, cities that serve one centre turn on the spot: Thebes, Chang’an, Versailles and Baghdad rise and fall with the arrival or departure of a court, a dynasty, a caliphate. Second, cities that last turn only through plateaus: Rome’s population curve moves in century-long standings, and the standing city — not the growing one — is what kept the place alive for a thousand years. Third — the finding with the flavour of a law, though the author carefully calls it an observation — a long rise ends in a standing in the cities that lasted, and in a direct fall in the cities that broke. A recurring figure of about 30,000 (Athens’ citizenry, Amarna, medieval Rome, Howard’s garden city of 32,000, Letchworth’s 34,000) surfaces as the size at which cities have repeatedly stood.

The methodological candour of the “How firm is this?” section is the book’s intellectual credit. Konstapel concedes that most pre-1800 series are estimates from two compilers who disagree, that the 11 per cent rule is a choice (Rome’s word holds only between 7 and 20 per cent), and that the cities were chosen for the history of an idea, not as a sample. A real test, he notes, would fix the rule in advance and apply it to every city with a measured history — an open question that the book itself lists as the beginning of the next one.

III. Art as evidence

A third current, unusual in an analytical history, is the use of art not as illustration but as argument. The plastered skulls of Jericho show ancestors kept in the house; the Babylonian Map of the World shows a city with the ocean ringing it like an amphitheatre; Bruegel’s Tower of Babel converts a ziggurat into the permanent image of a city that cannot finish; Lorenzetti’s Good Government (1338–39) is called, without irony, “still the best picture of what any city, with or without bodies, is for.” The art sections enact the book’s thesis that a city is a shared idea before it is a built object — each chapter’s images show what the city meant to those who drew it.

IV. The oldest new idea

It is on the strength of these curves that chapter 20 makes its proposal, and it is here that the general reader should attend most closely. The city of the future, Konstapel argues, is not a new city on empty ground — every placed city in the book (Amarna, the Round City of Baghdad, Palmanova, Brasília, Masdar, The Line) either emptied within a generation or grew somewhere other than planned. Instead it is an old city that has learned to stand: written over an existing one (the book chooses Rome, the only city to have passed through rise, fall and standing and remained), divided into many centres rather than one court (81 units of roughly 34,000 each — the standing size), designed for standing as its normal state, with water infrastructure closed first, and deliberately never complete.

The sixth proposal is the book’s boldest and, by its own argument, its oldest: participants without bodies. Against the obvious reading that this means software agents, Konstapel shows that the first city was already built for a bodiless resident — Enki in the waters beneath the temple — and that Rome, China and Japan each institutionalised guardian participants at the scales of house, crossing and city, complete (in the Chinese case) with accountability: a city god who failed to bring rain could be punished. The step the future city takes is “from stand-in to participant,” with Rome’s three scales supplying the form and Shannon Mattern’s A City Is Not a Computer (2021) supplying the counter-voice the book expressly endorses: these participants are not a control room.

The final move converts demographic decline from tragedy into design. Rome’s population curve now reads + + + + 0, and the last three steps “close only with a fall” on the axis of bodies. The proposal is to let the window close on several axes at once: fewer people and less matter on one axis, more memory, more crossings and more bodiless participants on the others — “the standing that follows is a standing the city chose.”

V. Judgment

The book’s weaknesses are those of its method, and it states most of them itself. The curves rest on estimate-heavy compilations; the 11 per cent rule is untested outside Rome; the sample is argumentative, not statistical. One may add that the five-part chapter anatomy, for all its clarity, flattens some cities into their idea more briskly than a historian would allow, and that the twenty-chapter arc from Eridu to AI agents occasionally asks antiquity to carry more futuristic freight than it can bear.

But these are the honest limits of a book that does something rare: it makes the history of the city testable. Its findings are stated so that they could be wrong, its sources so that they can be checked, and its proposal so that it can be drawn — the closing line forward asks whether the 81 units can in fact be laid on Rome’s real ground, along its water. For the intellectual reader, The History of the City offers not a pageant but a hypothesis: that the city is above all a place that returns to itself, that its pauses are its memory, and that its future — as in its beginning — will be shared with residents who have no bodies. It is an argument of unusual discipline and considerable ambition, and it deserves to be argued with.

---

Annotated reference

Konstapel, J. (2026). The History of the City: From Eridu to the City of the Future. Leiden: Constable Research. 47 pp., 20 chapters (0–20), manuscript dated 19–27 October 2026.

Genre and audience. Neither a conventional chronicle nor a planning manual: a structured, source-annotated argument in the history of ideas, aimed at a general intellectual readership, with an apparatus (per-chapter source lists, a “How firm is this?” section) that invites verification.

Structure. Every chapter repeats a five-part anatomy — (1) the city and its idea, (2) the builders, (3) the art, (4) the measured curve, (5) the line forward — which makes the twenty cases directly comparable and drives the synthesis of chapter 19 and the proposal of chapter 20.

Core thesis. A city is an answer to the question “what is a city for?” The answers form a single travelling idea (Eridu → Uruk → … → Rome), and their measured curves reveal that lasting cities turn only through plateaus, while cities built around a single centre turn on the spot and break.

Method. Comparative curve analysis: demographic and size series reduced to “words of steps” (+ / 0 / –) under a fixed standing rule (~11 per cent); sources are named per chapter (Wikipedia, Livius, ETCSL, excavation reports, Chandler and Modelski, census data after 1800, UN projections), with the author’s own Curves in the Vacuum.Net and The Path of Rome (2026) supplying the method and comparison baseline.

Key findings. (i) The plateau is not stagnation but memory-building (Uruk’s writing boom during its demographic pause); (ii) a recurring standing size of c. 30,000 people across very different eras; (iii) no placed city filled as drawn — Amarna, Baghdad’s Round City, Palmanova, Brasília, Masdar, The Line all emptied or grew off-plan; (iv) urban falls come from water infrastructure or from the departure of a single centre, never from smallness.

Distinctive claim. Chapter 20’s “city of crossings” with “participants with and without bodies,” argued as the return of the oldest urban idea (Enki at Eridu; Roman lares at house, crossroads and city; Chinese chenghuang; Japanese ujigami) rather than a novelty of the AI age — with Mattern’s A City Is Not a Computer (2021) as the endorsed counter-voice against the computational city.

Reliability. The author flags his own limits: pre-1800 series are estimates from disagreeing compilers; the 11 per cent rule is calibrated on Rome alone; the city sample is chosen for the history of the idea, not as a statistical sample. A pre-registered test on all cities with measured series is proposed as future work.

Use. Best read as a hypothesis-generating synthesis of urban history: valuable to historians of the city for its comparative tables, to planners for its anti-places-city evidence, and to anyone thinking about AI in urban life for its long prehistory of bodiless urban participants.

J.Konstapel,Leiden,26-10-2026.

Waar de Nederlandse staat kan terugtreden

Staat en bedrijf sinds 1940

Kan de overheid terugtreden, tot op nul? Die vraag klinkt eenvoudig. Ze is dat niet. De Nederlandse staat is namelijk drie dingen tegelijk. Hij is werkgever, hij is uitvoerder en hij is financier. Wie alle drie doorschakelt, gooit drie antwoorden op één hoop.

Dit essay volgt de analyse van Konstapel over de relatie tussen staat en bedrijf sinds 1940. Zijn centrale bevinding is scherp geformuleerd. Van de drie rollen is er maar één die ooit langdurig is gekrompen: de staat als werkgever en uitvoerder. Tussen 1985 en 2017 daalde het aandeel van het eigen apparaat van 8,4% naar 5,4% van alle werkenden. Dat gebeurde door privatisering en bezuinigingen. De staat als financiert kromp daarentegen nooit. Collectieve zorg steeg van 0,8% naar 9,3% van het bbp, zonder één blijvende dip. En de last van belastingen en premies staat nog altijd rond 39% van het bbp, tegen 14% in 1938.

“De overheid op nul” valt daarmee uiteen in drie vragen met drie verschillende antwoorden. Ambtenaren kunnen veel verder omlaag dan nu; daarvoor bestaat precedent. De uitvoerende laag tussen staat en burger — zelfstandige bestuursorganen, zorg- en onderwijsinstellingen — is waar het meeste collectieve werk feitelijk zit. Terugtrekken daar betekende tot nu toe verplaatsen van werk, niet verwijderen. En de financieringsrol is de kern van de naoorlogse Nederlandse verhouding. Die is nooit teruggedraaid.

De methode van het onderliggende onderzoek verdient vermelding. Konstapel ontleent die aan twee eerdere studies: The Great American Job Shift en The Fundamental Fractal. Elke beroep krijgt het Holland-type waarop O*NET 31.0 het hoogst scoort. Via een crosswalk naar internationale beroepsgroepen (ISCO-08) en Nederlandse tellingen — volkstellingen van 1960 en 1971, Eurostat-arbeidsmarktcijfers tot 2025 — ontstaat een beeld van de Nederlandse beroepsstructuur in zes types. Die structuur wordt als één gewogen punt op de hexagoon van Holland geplaatst. De afstand tot het middelpunt zegt hoe eenzijdig het werk is; de hoek zegt in welke richting het werk helt.

2. De rollen van de staat, 1940–2025

Sinds 1940 groeide de staat vooral als betaler, veel minder als werkgever. Het eigen apparaat kwam nooit boven de circa 8% van alle werkenden uit. De overheidsuitgaven passeerden daarentegen de helft van het bbp.

De kerncijfers uit het onderzoek:

Rol	Indicator	Rond 1950	Piek	Laatst
Werkgever	Collectieve domein in totale werkgelegenheid	19%	27% (1985)	26% (2020)
Werkgever	Overheidsadministratie, defensie en politie in werkgelegenheid	7,2% (1969)	8,4% (1985)	6,6% (2025)
Producent	Toegevoegde waarde publieke sector, % bbp	8,6%	13,6% (1980)	12,5% (2020)
Financiert	Collectieve uitgaven, % bbp	27,6%	58,6% (1983)	42,5% (2018)
Financiert	Sociale zekerheid, % bbp	4,8%	19,7% (1983)	11,8% (2018)
Financiert	Collectieve zorg, % bbp	0,8%	9,8% (2012)	9,3% (2018)
Belastinginner	Belastingen en premies, % bbp	25,5%	41,5% (1987, 1993)	38,8% (2018)
Ondernemingssteun	Overdrachten aan bedrijven, % bbp	4,1%	5,4% (1985)	1,4% (2015)
Eigenaar	Staatsbedrijven (PTT, DSM, Postbank, KLM-deelname, NS)	groot	tot 1986	verkocht of verzelfstandigd 1986–1998

Uit deze tabel volgen twee dingen voor de vraag naar terugtreden. Ten eerste: de staat als werkgever is klein en is eerder gekrompen. Ten tweede: de staat als betaler is groot en nooit structureel gekrompen, met uitzondering van de sociale zekerheid na 1983. Het meeste wat de staat betaalt, wordt bovendien uitgevoerd door mensen die geen ambtenaar zijn. In de zorg werken 1,7 miljoen mensen en in het gesubsidieerd onderwijs 400.000, tegen 221.000 bij het Rijk.

3. Waar de staat al terugtrad

Nederland heeft één groot experiment in terugdringing gedraaid: van 1983 tot 2017. Het werkte voor de staat als eigenaar en werkgever, maar niet voor de staat als betaler.

Verkopen en verzelfstandigen, 1986–1998. Postbank, DSM, de PTT (later KPN en PostNL), de KLM-deelname en de spoorwegen verlieten de staat of werden zelfstandige bedrijven. De PTT alleen al haalde bijna 100.000 mensen uit het ambtenarenbestand. In dertig jaar daalde het aantal ambtenaren van 549.200 naar 428.100. Onderwijs, academische ziekenhuizen en politie groeiden in dezelfde periode samen met 173.000 functies.

Bezuinigen, 2011–2017. Na de kredietcrisis verdwenen 71.000 voltijdbanen bij de overheid.

Opnieuw groeien, 2018–2024. Daarna steeg het aantal met 154.000 naar 1,1 miljoen voltijdbanen. Het Rijk +30%, gemeenten +18%, onderwijs +15%. Alles wat na de crisis was gekapt, kwam dubbel terug.

Verplaatsen in plaats van verwijderen. Veel taken gingen naar zelfstandige bestuursorganen (zbo’s). Daar werken meer dan 100.000 mensen, met gezamenlijke budgetten die het Rijksbudget overstijgen. Ze tellen niet meer mee als ambtenaar, maar de staat betaalt nog steeds.

De parlementaire enquête naar de privatisering (commissie-Kuiper, Verbinding verbroken?, 2012) vond de hoofdfout. De redenen per stap waren onduidelijk. Ministeries stuurden achteraf zwak. En niemand toetste of de taak zelf nog nodig was. Terugtreden zonder besluit over de taak verklaart de hergroei. Het werk verdween uit de boeken, de functie bleef, en het werk kwam terug.

4. Vijfenzestig jaar in zes types

In 1960 deed zes op tien Nederlandse werkenden Realistisch werk: maken, bouwen, besturen van machines. In 2025 is dat één op drie. Sociaal werk meer dan verdubbelde, van 5% naar 13%. Investigatief werk groeide zesvoudig. Conventioneel werk — administratie, registratie — bleef de hele periode rond een derde hangen.

De vorm matcht het Amerikaanse verhaal. Een maakconomie wordt een zorg- en kennisconomie, terwijl de administratie blijft bestaan. De cijfers voor 1960 en 1971 zijn middens van een marge van ongeveer vier punten, omdat die volkstellingen managers niet als aparte groep kennen.

Het zwaartepunt van het Nederlandse werk bewoog van 0,70 naar 0,43 van de weg naar een hoek van de hexagoon, en draaide daarbij circa 15 graden. Die draai vond vrijwel allemaal vóór 1992 plaats. Het is een naar binnen lopende spiraal: gestopt met draaien, maar niet met krimpen. De draai loopt van Realistisch naar Conventioneel en Ondernemend — dezelfde richting als in de VS. Amerikaans werk staat in 2025, op dezelfde manier gemeten, ongeveer 26 graden verder gedraaid en iets dichter bij het centrum.

Wie doet welk werk

Het collectieve domein (overheid, onderwijs, zorg) groeide van 15% van de werkgelegenheid in 1930 naar 26% in 2020. Het droeg de opkomst van het Sociale werk. Binnen dat domein veranderde het werk meer dan de omvang.

Sector, 2011 → 2025	Werkenden (×1.000)	Professionals	Administratief	Technici en associaten
Overheid	523 → 736	32% → 45%	16% → 8%	19% → 26%
Onderwijs	554 → 758	75% → 72%	6% → 5%	3% → 7%
Zorg	1.345 → 1.712	25% → 37%	7% → 5%	29% → 27%
Bedrijfsleven	4.878 → 6.415	16% → 27%	11% → 10%	14% → 15%

Drie bewegingen springen eruit.

1. De ambtenaar als klerk verdwijnt. Het aantal administratieve functies bij de overheid daalde van 84.000 naar 60.000. Het aantal professionals verdubbelde tot 332.000. De staat groeit, maar in analyse, advies en toezicht — niet in registratie.
2. Zorg verschuift van verzorgen naar behandelen. Persoonlijke verzorging daalde van 29% naar 23% van het zorgwerk; professionals stegen van 25% naar 37%.
3. Kenniswerk verhuist naar het bedrijfsleven. Professionals in bedrijven meer dan verdubbelden, van 756.000 naar 1,7 miljoen. In 2011 zat bijna de helft van al het Investigatieve en Artistieke werk in het collectieve domein; in 2025 zit meer dan de helft bij bedrijven.

Handwerk en ondernemerschap blijven waar ze waren. Ongeveer driekwart van het Realistische werk en twee derde van het Ondernemende en Conventionele werk gebeurt in beide jaren bij bedrijven.

5. De geldkant

De collectieve uitgaven stegen van 28% van het bbp in 1950 naar een piek van 58,6% in 1983, en zakten naar 42,5% in 2018. Collectieve zorg is de enige lijn die nooit omsloeg: van 0,8% naar 9,3%.

De last bewoog niet mee met de uitgaven. Belastingen en premies bleven tot een decennium na de uitgavenpiek stijgen en staan nu rond 39%, tegen 14% in 1938. Wat veranderde, is wie betaalt. Lasten op inkomen en arbeid stegen van 12% van het bbp in 1950 naar 27% in 1983 en staan nu op 22,5%. Lasten op kapitaal en winst daalden van 5,5% naar 4,4%. Directe steun aan bedrijven kromp van circa 5% van het bbp midden jaren tachtig naar 1,4% in 2015 — vóór de coronasteun van 2020.

6. Ondernemers en zelfstandigen

Het aandeel zelfstandigen is bijna gelijk aan vijftig jaar geleden, maar het zijn andere mensen. In de volkstelling van 1960 was 15,8% van de beroepsbevolking zelfstandig en 4,8% meewerkend in een familiebedrijf: boeren, winkeliers, ambachtslieden. De nationale rekeningen laten 18,2% zelfstandigen zien in 1969, een dieptepunt van 15,3% rond 2000–2005 en 16,4% (1,7 miljoen) in 2025.

De boeren en winkeliers vertrokken; de zzp’er zonder personeel arriveerde. Zijn aandeel in alle zelfstandigen steeg van 60% in 2002 tot meer dan twee derde in 2011. Het aantal bereikte bijna 1,3 miljoen in 2024. In 2025 sloeg de lijn voor het eerst om: 62.000 zzp’ers minder, vooral jongeren, nadat de fiscus begon met handhaving van de regels tegen schijnzelfstandigheid.

Op de hexagoon weegt dit minder dan het debat suggereert. Zelfstandigheid is een contractvorm, geen werksoort. Een zzp-verpleegkundige doet Sociaal werk; een zzp-loodgieter doet Realistisch werk. Het Ondernemende aandeel van het Nederlandse werk daalde juist licht, van circa 10% naar 8%, terwijl het aantal zelfstandigen steeg.

7. De kaart: waar de staat naar nul zou kunnen

Van de 1,1 miljoen voltijdbanen bij de overheid in 2024 heeft één deel het duidelijkste precedent voor terugtreden: regelgebaseerde uitvoering. Denkbare kandidaten zijn het uitbetalen van uitkeringen, het innen van standaardbelastingen en het afgeven van vergunningen. Daar is de klerk al gehalveerd. De kern van geweld en rechtspraak heeft geen enkel precedent voor terugtreden. En de financiering van zorg en onderwijs is een politieke keuze, geen bezettingsvraag.

Onderdeel van de staat, 2024	Voltijdbanen	Vooral bezig met	Historisch precedent	Lezing voor “naar nul”
Sociale verzekeringsfondsen (UWV, SVB)	25.500	Uitkeringen volgens vaste regels	Socialezekerheidsuitgaven daalden na 1983 van 19,7% naar 11,8% bbp	Regelgebaseerd; sterkste kandidaat voor vervanging door processen
Rijk (incl. fiscus, defensie)	221.000	Beleid, belastinginning, defensie, justitie	−19.300 in 2011–2017, daarna +50.600	Gesplitst: routinewerk kan ver krimpen; defensie en justitie zijn de kern van het staatsmonopolie
Gemeenten	183.700	Vergunningen, registraties, sociaal domein sinds 2015	−25.100, daarna +28.100	Registraties en vergunningen krimpen; sociaal domein groeide door decentralisatie
Gemeenschappelijke regelingen	92.100	Veiligheidsregio’s, GGD’s, gedeelde diensten	Grotendeels ontstaan door taakverplaatsing	Gedeelde uitvoering; kandidaat voor contract of proces
Overige organen, waaronder politie	177.800	Politie, zbo’s en andere agentschappen	Zbo’s: taken verplaatst, niet verwijderd	Politie is kern; zbo-taken vergen een besluit per taak
Provincies en waterschappen	23.800	Ruimtelijke ordening, waterbeheer	Klein en stabiel	Waterveiligheid is infrastructuur, geen administratie
Gesubsidieerd onderwijs	402.300	Onderwijsgeven	Nooit gekrompen als aandeel van het werk	De staat financiert; scholen voeren uit; nul betekent stoppen met publieke onderwijsfinanciering
Collectief gefinancierde zorg	circa 1,7 mln werkenden	Behandeling en zorg	Alleen maar gegroeid sinds 1950	De staat financiert en reguleert; nul betekent stoppen met collectieve financiering

Uit het experiment van 1983–2017 volgen twee lessen voor elke nieuwe poging. Ten eerste: beslis per taak of ze nog nodig is, vóórdat je beslist wie ze uitvoert. De commissie-Kuiper constateerde dat deze stap werd overgeslagen — en het werk kwam terug. Ten tweede: het eigen werk van de staat is al verschoven van registreren naar toezien. Dat laag terugdringen betekent dus: minder regels om op toe te zien, niet alleen snellere processen.

8. Conclusies

Vijfentachtig jaar Nederlandse data geven een duidelijke volgorde van waar terugtreden werkte en waar het nooit gebeurde.

1. De staat als eigenaar kan naar nul, en is dat vrijwel. De grote staatsbedrijven werden tussen 1986 en 1998 verkocht of verzelfstandigd.
2. De staat als werkgever kan ver krimpen; dat is eerder gebeurd. Het apparaat daalde van 8,4% naar 5,4% van alle werk tussen 1985 en 2016. De klerk verdwijnt al: van 16% naar 8% van het overheidspersoneel sinds 2011. Regelgebaseerde uitvoering is de eerste kandidaat voor vervanging door processen.
3. Terugtreden zonder taakbesluit komt terug. De kap van 71.000 banen in 2011–2017 werd gevolgd door groei van 154.000 in 2018–2024. Naar zbo’s verplaatste taken bleven op het publieke budget. De volgorde is: eerst vragen of een taak nodig is, dan wie ze uitvoert.
4. De staat als financiert is nooit gekrompen, behalve de sociale zekerheid. Collectieve zorg ging van 0,8% naar circa 9% van het bbp; onderwijs blijft rond 5%. De staat daaruit halen betekent de collectieve financiering beëindigen — een keuze over de naoorlogse verhouding, niet over ambtenaren.
5. De staat als toezichter is de groeiende laag. Overheidsprofessionals verdubbelden sinds 2011. Minder regels is de enige hendel die deze laag doet krimpen.
6. Het bedrijfsleven draagt het nieuwe werk al. Sinds 2011 vindt de groei van kenniswerk vooral plaats bij bedrijven. Overdrachten van staat aan bedrijven daalden van 5,4% van het bbp (1985) naar 1,4% (2015). Werk heeft de staat niet nodig om te groeien; de last op arbeid (22,5% van het bbp) is wat het bedrijfsleven nog voor de staat draagt.

Wat de Hollandsche data toevoegen: de Nederlandse spiraal trekt gestaag naar een evenwichtige mix van werk, met of zonder groeiende staat. De slinger van de staat (uitgavenpiek 1983, lastenpiek 1993) en de draai van de werkspiraal (gestopt rond 1992) zijn afzonderlijke ritmes. Het krimpen van de staat hoeft de spiraal dus niet te stoppen.

Open eindpunten zijn er ook. De volkstelling van 1947 — alleen in gedrukte CBS-delen — zou een echte uitgangspositie voor 1940 geven. Verder ontbreekt een taaksgewijze lijst van functies van Rijk en gemeenten, en de bijbehorende processen per kandidaat-taak. Dat is het onderwerp van een vervolgrapport over het terugbrengen van het aantal ambtenaren naar nul.

---

Geannoteerde referentielijst

Primaire bron

Konstapel, J. (2026). Where the Dutch State Can Step Back: State and Business since 1940. Leiden, 26 oktober 2026.
Het onderliggende essay. Analyseert 85 jaar Nederlandse data over de rollen van de staat (werkgever, producent, financiert, eigenaar) in vier lagen: rollen, eerdere terugtreding, werksoorten en geldstromen. Methode: O*NET 31.0-typing van beroepen, geprojecteerd op de Nederlandse beroepsbevolking via IPUMS-volkstellingen (1960, 1971) en Eurostat (1992–2025). Dit essay volgt de structuur en de cijfers van deze tekst.

Gebruikte databronnen (zoals vermeld in de primaire bron)

O*NET 31.0 database — Career Interest Types.
Amerikaanse beroependatabase die 798 beroepen scoort op de zes Holland-types (Realistisch, Investigatief, Artistiek, Sociaal, Ondernemend, Conventioneel). Vormt de basis van de beroepsclassificatie. Beperking: Amerikaanse beroepen, vertaald via een crosswalk.

O*NET crosswalks — ESCO/ISCO naar ONET-SOC.
Verbindt 434 internationale ISCO-08-beroepen aan de O*NET-scores. Zo krijgt elk Nederlands beroepsgroep-aandeel een type. Elk ISCO-beroep krijgt het type met de hoogste gemiddelde score.

IPUMS-International — beroep, volkstellingen 1960 en 1971.
Nederlandse volkstellingsmonsters met beroepsinformatie. Kent slechts zes à zeven hoofdgroepen en geen aparte managementscategorie; daarom zijn de cijfers voor die jaren een midden van een range van circa vier punten.

Eurostat lfsa_egais — werkgelegenheid naar beroep, 1992–2025.
Jaarlijkse arbeidskrachtenenquête naar ISCO-hoofdgroepen. Dragende reeks voor de ontwikkeling van de beroepsstructuur sinds 1992. De overstap van ISCO-88 naar ISCO-08 breekt de reeks rond 2010/2011.

Eurostat lfsa_egai2d — werkgelegenheid naar 2-cijferige beroepsgroep, 2011–2025.
Meer detailniveau voor de recente periode. Het detailniveau verschuift Realistisch en Conventioneel circa vijf punten; de trends bewegen niet.

Eurostat lfsa_eisn2 — beroep naar economische activiteit, 2008–2025.
Basiswais voor de tabel “wie doet welk werk” per sector (overheid, onderwijs, zorg, bedrijfsleven).

CPB — De Nederlandse economie in historisch perspectief (2023), data hoofdstuk 5.
Historische reeksen 1950–2018 voor collectieve uitgaven, sociale zekerheid, collectieve zorg en de lastendruk. Bron voor de geldkant van het betoog en de constatering dat collectieve zorg nooit daalde.

CBS StatLine — arbeidsvolume naar bedrijfstak, 1969–2016.
Nederlandse reeks voor de ontwikkeling van het collectieve domein en het overheidsapparaat als werkgever, inclusief de krimp van het ambtenarenaandeel van 8,4% (1985) naar 5,4% (2016).

BLS — National Employment Matrix 2025.
Amerikaanse tegenhanger voor de vergelijking van de “spiraal”: Amerikaans werk staat in 2025 ongeveer 26 graden verder gedraaid op de hexagoon dan Nederlands werk.

CBS — daling aantal zzp’ers met 62.000 in 2025.
Eerste keer dat het aantal zzp’ers daalde, vooral jongeren, na fiscale handhaving van de regels tegen schijnzelfstandigheid. Bevestigt dat zelfstandigheid een contractvorm is, geen werksoort.

ESB (1983) — beroepsstructuur 1960, 1971, 1977.
Vroegste uitsplitsing van de Nederlandse beroepsstructuur, gebruikt als brug tussen de volkstellingen en de moderne Eurostat-reeksen.

J. Konstapel — The Great American Job Shift.
Eerdere studie van dezelfde auteur over de verschuiving van Amerikaans werk over de Holland-hexagoon. Leverde de methodische blauwdruk en de vergelijkingsmaatstaf voor de Nederlandse spiraal; de Amerikaanse projectie uit dat rapport komt niet terug uit de actuele data.

Beleids- en onderzoeksbronnen (zoals aangehaald in de primaire bron)

CBS (2025) — Meer mensen in dienst bij de overheid.
Voltijdbanen per onderdeel van de overheid, 2011–2024. Bron voor de tabel in paragraaf 7 en voor de hergroei na 2018 (+154.000 voltijdbanen tot 2024).

Binnenlands Bestuur — Precies genoeg ambtenaren.
Documenteert de daling van het aantal ambtenaren van 549.200 naar 428.100 in dertig jaar, tegen groei van onderwijs, academische ziekenhuizen en politie met 173.000 functies. Kerndata voor het hoofdstuk over privatisering en bezuiniging.

BZK (2019) — De grote onbekende: zbo’s.
Overzicht van de zelfstandige bestuursorganen: meer dan 100.000 werknemers, gezamenlijke budgetten boven het Rijksbudget. Belangrijk omdat het laat zien dat verplaatsing van taken geen vermindering van collectieve middelen betekent.

*Eerste Kamer der Staten-Generaal (2012) — Commissie-Kuiper, Verbinding verbroken? Parlementaire enquête privatisering en verzelfstandiging overheidstaken 1990–2010.
Parlementaire evaluatie van twintig jaar terugdringing. Kernbevinding: onduidelijke motieven per stap, zwakke sturing achteraf en geen toets op de noodzaak van de taak zelf. Verklaart waarom het gekapte werk terugkwam. Bepalend voor les 3 in de conclusies.

---

Dit essay is een Nederlandse uitwerking van de primaire bron. De cijfers, tabellen en conclusies zijn overgenomen uit die tekst; de annotaties beschrijven de rol van elke bron binnen het betoog.

The Vacuum.Net is a Cellular Automaton.

Gerard ’t Hoofts cellulaire automaton, waarin de werkelijkheid een verzameling deterministische ‘tandwielen’ is, en de Vacuum.Net-theorie, waarin alles een gesloten lus is die bestaat door terug te keren zijn gelijk aan elkaar.

J.Konstapel,Leiden,25-10-2026.

based on Wired : Scientists Trapped an Antimatter Particle for Months to Study Its Secrets

In October 2026, a research group at Northwestern University reported a remarkable achievement. They trapped a single positron in a Penning trap and held it there for months. They watched it climb a ladder of quantum states, one step at a time. They watched it fall back down. Every rung of that ladder had exactly the same height.

This experiment is a triumph of precision physics. But it also raises a deeper question. Quantum theory tells us the ladder exists. It does not tell us what the ladder is. This essay sets two answers side by side. The first comes from Nobel laureate Gerard ‘t Hooft. The second comes from the Vacuum.Net theory. The central claim is simple: the two answers are the same answer, stated in two different languages.

1. The experiment: a positron on a ladder

The experiment by Myers, Soucy, Sukra, Sinha and Gabrielse appeared in Physical Review Letters on 21 October 2026. The procedure is exacting. A positron emerges from a sodium-22 nucleus with 546 keV of energy. It is slowed down and caught in a Penning trap. The trap sits in a 5.6 tesla magnetic field and is cooled to 300 millikelvin.

Inside the trap, the positron circles the magnetic field lines at about 156 GHz. It settles into the lowest cyclotron state. Apply a microwave drive, and it jumps one rung up. Wait, and it falls back. Every rung has the same height. The energy levels follow the familiar formula:

E = ħω_c(n + ½), with n = 0, 1, 2, …

On top of this sits a second ladder for the spin. The spin precesses at a slightly different rate. The two frequencies differ by about one part in a thousand. Their difference is called the anomaly frequency.

Several details of the apparatus deserve attention.

• Loading. The positron does not enter the trap alone. It leaves a tungsten moderator pulling an electron from the surface. The two travel together as a neutral positronium atom through a 640 micrometre hole. The trap’s electric field cannot see the neutral pair. Only inside the trap is the field strong enough to pull them apart. The electron is lost. The positron is caught.
• The cavity. Without shielding, an excited cyclotron state would decay in about 83 milliseconds. The trap electrodes form a microwave cavity that suppresses this decay by up to a factor of two hundred. The lifetime becomes about four seconds. That is long enough to see each rung.
• The spin. The spin is far more isolated. An upward spin takes about 2.1 years to decay. The lowest spin-down state does not decay at all.
• Reading. The states are not read directly. Nickel rings create a small magnetic bottle. This shifts the positron’s slow axial oscillation by about 3.8 Hz per quantum. One loop is read through the frequency of another loop. The measurement does not disturb what it reads.

The purpose is precision. An earlier one-electron experiment measured the electron’s magnetic moment to about one part in 10¹³. The positron version should match that precision. Comparing the two will test the symmetry between matter and antimatter some thirty times more sharply than before.

Quantum electrodynamics explains the ladder perfectly. It says the ladder is there. It says nothing about what the ladder is. That is where our two answers begin.

2. The first answer: ‘t Hooft’s cogwheels

Gerard ‘t Hooft won the Nobel Prize in 1999. He has argued for over two decades that quantum mechanics is not the floor of reality. His position is set out in The Cellular Automaton Interpretation of Quantum Mechanics (2016). In his view, quantum theory is a superb calculating device. Underneath it lies a world that is deterministic and discrete.

His construction has three steps.

Ontological states. There is a fixed set of states the world can actually be in. At any moment, exactly one of them is realised. In quantum notation, the only “real” wave function takes the value 1 on one basis state and 0 on all others.

Evolution. At each tick of a clock, the world moves from one state to the next. The step matrix contains only zeros and ones. It is a permutation. It must be invertible, so no two states ever merge.

The cogwheel. The simplest such system has N states visited in a circle: 1 → 2 → … → N → 1. ‘t Hooft calls this a cogwheel with N teeth. He then proves a key structural fact. Take any finite, deterministic, time-reversible system. Start anywhere and follow the evolution. You will return to your starting point after a finite number of steps. Start from an unvisited state and you trace another closed cycle. Continue until every state is used. The result: the most general such system is a collection of cogwheels, each with its own number of teeth, all turning on the same tick.

From cogwheel to ladder. Apply a discrete Fourier transform to the N states of one cogwheel. The evolution operator becomes diagonal. It defines a Hamiltonian with evenly spaced eigenvalues, E_k = 2πk/N. In the continuum limit, with a small constant shift, this becomes (2π/T)(k + ½) — the spectrum of the harmonic oscillator. ‘t Hooft notes that the Zeeman splitting of an atom with angular momentum J is exactly the spectrum of a cogwheel with 2J+1 teeth. A Zeeman atom, seen in the right basis, is a cogwheel.

Superposition as template. What, then, is a superposition? For ‘t Hooft, it is a template: a state we write down when we do not know which ontological state the system is in. Templates obey the Schrödinger equation and are indispensable for calculation. But they are not states the world can be in. The dead cat is an ontological state. So is the live one. The superposed cat solves the equations but never occurs. Wave-function collapse is not a physical process. It is an update of knowledge.

The observer. Bell’s theorem seems to forbid any local deterministic account of entanglement. ‘t Hooft’s reply: Bell’s argument assumes experimenters can choose their settings independently of everything else. In a deterministic world, the experimenters and their instruments obey the same laws as the particles. Their choices are not outside the system. This position is called superdeterminism.

The open gap. ‘t Hooft is candid about what remains unsolved. Free cogwheels are easy. Interacting particles are hard. Models with interactions often fail to be both local and bounded from below. He does not yet have a cellular automaton for the Standard Model. He suggests gravity may be needed to close the gap.

3. The second answer: everything is a closed line

The Vacuum.Net theory starts from a single axiom. Everything is a closed line. A thing exists because it returns. Outside the path there is nothing. The observer is part of the path.

Three elements fill this out.

Trits. The net is written in balanced ternary. Each position carries one of three values: −1, 0, +1. Balanced ternary needs no minus sign and no rounding rule. Negation is simply the exchange of + and −. A trit is not an extra assumption. It is how the ternary numbers read a position on the loop.

Addresses. An address is a word of trits written on a ring. A ring has no first position. So an address is not a single word but a rotation class: the word plus all its cyclic shifts. In combinatorics this is called a necklace. The number of necklaces of length n over three symbols grows roughly as 3ⁿ/n. At depth 19 there are 61,171,659 of them.

Two layers. The net has two layers. The address layer is frozen topology: the closed loops that keep coming back. The tension layer is where anything happens: exchange, decay, change. Earlier work (“The Net Has No Hidden Clock”, September 2026) exhaustively checked what the address layer can do. On a ring of five positions, every local, conserving, reversible rule was enumerated — all 1,594,323 sign-symmetric rules and all 144 conserving flux-form rules. Only the identity and the shift survive. In the address layer there is only passage. Change lives in the tension layer.

4. The identity: the axiom is the theorem

Put the two descriptions side by side. Three statements follow almost at once.

4.1 The axiom is the theorem. The net’s axiom says everything is a closed loop. ‘t Hooft’s theorem says every finite, deterministic, reversible system is a set of closed cycles. He reaches this as a consequence of determinism and finiteness. The net takes it as its ground. They state the same fact about the world.

4.2 An address is a cogwheel. Take one address: a necklace of trits on a ring of length n. One step of the loop moves the word one position along the ring. After a number of steps equal to the necklace’s period d, the word returns to its start. That is exactly a cogwheel with d teeth. Every address of the net is one cogwheel. The catalogue of addresses at depth n is the list of cogwheels at depth n.

4.3 The address dynamics is the cogwheel step. The exhaustive check found only one non-trivial local dynamics on a single ring: the shift. The shift is the cogwheel step. The address layer is therefore not merely compatible with ‘t Hooft’s automaton. It is the pure cogwheel automaton, with nothing added.

4.4 What this gives for free. ‘t Hooft’s machinery applies directly. Each address of period d carries an evenly spaced spectrum, 2πk/d. Energy is the number of full turns per tick: a frequency. The constant h enters only when frequencies are converted into joules. Quantum ladders, Zeeman levels and oscillators are what cogwheels look like in quantum notation. In the net, they are what loops look like.

One further detail matters. In his treatment of fermions, ‘t Hooft writes evolution operators whose entries are −1, 0 and +1. These are balanced-ternary values. The trit is already present in his mathematics.

5. A worked example: depth three

Abstract identities convince less than a small case worked out in full. Take words of three trits. There are 3³ = 27 of them. Group them into rotation classes:

Class (representative)	Period (teeth)	Trit sum	Mirror partner
0 0 0	1	0	itself
+ + +	1	3	− − −
− − −	1	−3	+ + +
0 0 +	3	1	0 0 −
0 0 −	3	−1	0 0 +
+ + 0	3	2	− − 0
− − 0	3	−2	+ + 0
+ + −	3	1	− − +
− − +	3	−1	+ + −
0 + −	3	0	0 − +
0 − +	3	0	0 + −

There are eleven classes: three cogwheels with one tooth and eight with three teeth. Check: 3 × 1 + 8 × 3 = 27. The formula agrees: g(3) = (1/3)(3³ + 2·3) = 11.

Now read off the spectrum. A one-tooth cogwheel has a single level, E = 0. A three-tooth cogwheel has three levels: 0, 2π/3 and 4π/3. Across all 27 states:

Level (per tick)	Number of states
0	11
2π/3	8
4π/3	8

That is the complete “quantum” description of the net at depth three: a finite collection of loops, each turning on its own ring, each with its own small ladder. Note that the spacing 2π/3 is exactly the phase step of one ternary position: one third of a full turn.

Three features of the table matter for what follows.

• Mirror pairs. Negating every trit maps each class to a class with the opposite trit sum. Negation does not interfere with rotation. A class and its mirror therefore carry identical spectra. Mirroring changes the sign. It never changes the ladder.
• Self-mirrored neutrality. The class 0 0 0 is its own mirror. The classes 0 + − and 0 − + are each other’s mirror and each other’s reversal. Mirror and reversal together map each onto itself.
• Neutral words. Three classes carry trit sum 0. They are loops that exist but carry no net sign.

6. The experiment, read as cogwheels

With the identity in place, the Northwestern experiment can be read line by line.

The cyclotron ladder is a cogwheel. The positron’s orbit is a closed loop. Its quantum number counts closures. The evenly spaced ladder ħω_c(n + ½) is precisely the spectrum ‘t Hooft derives for the continuum limit of a cogwheel, including the half-step offset. The experiment shows, rung by rung, what a single cogwheel looks like when one tooth is excited at a time.

The spin is a second cogwheel. Spin and orbit are two loops turning at nearly the same rate. Their frequency ratio is g/2, about 1.00116. The anomaly frequency is the beat between two cogwheels whose tooth counts differ slightly. What precision physics measures, to twelve or thirteen digits, is the ratio of two loops. That ratio is a pure number. It needs no constant of nature to be stated.

The positron is a mirror loop. To keep ω_c positive, the authors orient their axis opposite to the magnetic field. The positron turns the other way round. In the net, antimatter is the mirrored address: every trit negated. Negation commutes with the cogwheel step, so a mirrored address carries the same ladder. The net therefore gives electron and positron exactly the same spectrum, and so exactly the same magnetic moment. The comparison the Northwestern group is preparing tests precisely this.

Loading through a zero. The positron enters the trap bound to an electron. The pair is neutral: +1 and −1 make 0. The trap’s field does not act on it. Only inside does the pair split, and only then does charge appear. This is the net’s neutral word made visible in the laboratory. A loop with sum zero passes where a signed loop would be stopped.

The cavity is context. An excited cyclotron state decays by radiating into the electromagnetic field. In the language of loops, decay is an exchange between the positron’s cogwheel and cogwheels of the surrounding field. That exchange belongs to the tension layer. The cavity removes the field’s partner loops near 156 GHz, so the exchange is suppressed: 83 milliseconds becomes four seconds. The decay rate is not a property of the positron alone. It is a property of the positron together with its surroundings. In the net this is no surprise. Context belongs to the path.

The spin is frozen. The spin barely exchanges with anything. Its calculated lifetime is about two years; the lowest spin-down state does not decay at all. The spin behaves like the address layer: frozen, passing, unchanged. The orbit behaves like the tension layer: it exchanges, rises and falls. The two-layer structure of the net is visible in a single particle.

Reading through another loop. The cyclotron and spin states are read through the shift they cause in a third loop, the slow axial oscillation. No instrument stands outside. One loop is read through the count of another loop coupled to it. This is what the net means by saying the observer is part of the path. It is also what ‘t Hooft means when he refuses to place the experimenter outside the deterministic system.

7. Where the net goes further

The identity is close, but not complete. In two respects the net needs less than ‘t Hooft’s automaton.

No universal clock. ‘t Hooft’s cogwheels all turn on one common tick, which he places near the Planck time. The net has no such clock. Time in the net is the counting of turns of the loop itself. Each address returns on its own period. ‘t Hooft’s own mathematics points this way: each cogwheel’s energy can be shifted by its own arbitrary constant with no physical effect, because which cogwheel a state belongs to never changes. Each cogwheel is, in that sense, its own clock. The net takes this seriously and drops the common tick.

No prior lattice. ‘t Hooft places his automaton on a lattice given in advance: a grid of sites in space. The net has no grid. The net is the path the movement traces. Addresses are not positions on a pre-existing structure; they are the structure. ‘t Hooft must eventually explain why his lattice has the geometry it has. The net does not carry that burden.

Superposition. For ‘t Hooft, a superposition is a template, not a state of the world. The net arrives at the same place from its own start. Outside the path there is nothing. An address is therefore either traced or not traced: 1 or 0. A weighted mixture of addresses is a statement about what we know, not a configuration of the net. Probabilities measure ignorance of which loop is being traced.

8. What both still have to write

There is one place where neither ‘t Hooft nor the net is finished, and it is the same place.

Free cogwheels are simple. The world, however, is full of exchange: emission, absorption, scattering, chemistry, life. In ‘t Hooft’s account, this is the problem of building a cellular automaton with local interactions whose quantum Hamiltonian is both local and bounded from below. He states openly that he has not found it for realistic particles.

In the net, this is the tension layer. The address layer allows only passage. All change must therefore come from the rule by which loops exchange trits while the total balance stays conserved. The exhaustive result adds its own constraint: this rule cannot live inside a single ring with nearest-neighbour steps. It must act between rings, or reach further along them.

The cavity in the positron experiment shows what such a rule must do. It must let the cyclotron loop give up a tooth to the field. It must make that exchange depend on which partner loops are present. And it must leave the spin loop almost untouched. Writing that coupling rule explicitly, in balanced ternary, between rings, is the next piece of work. It would complete the automaton exactly where ‘t Hooft’s remains open.

9. Conclusion

A single positron on a ladder looks like the purest quantum object there is. Read with ‘t Hooft, it is a cogwheel: a closed cycle of states that quantum notation renders as evenly spaced levels. Read with the Vacuum.Net theory, it is a loop: a thing that exists because it keeps coming back. These are not two readings. They are one.

The net’s single axiom is ‘t Hooft’s structural theorem. The net’s addresses are his cogwheels. The net’s address dynamics is his cogwheel step, and nothing else. The positron experiment — its loading, its cavity, its frozen spin, its indirect reading — shows the two layers of the net working in one particle. The coming comparison of electron and positron tests a symmetry the net carries in its arithmetic: negating the trits changes the sign, never the ladder.

Quantum mechanics, on this view, is what loops look like when you calculate with them. The loops themselves are what there is.

---

Annotated references

The experiment

1. T. G. Myers, L. Soucy, B. A. D. Sukra, B. Sinha, G. Gabrielse, “One-Positron Quantum Cyclotron,” Phys. Rev. Lett. 137, 173001 (2026). doi:10.1103/gbkc-6nlj. Open access.
   The source of every experimental detail in Sections 1 and 6: the 5.6 T field, 300 mK temperature, sodium-22 source, positronium loading through a 640 µm hole, cavity-inhibited decay (83 ms to ~4 s), the ~2.1-year spin lifetime, and quantum non-demolition reading via the axial frequency. This is the primary paper on which the entire essay rests.
2. “A Quantum Cyclotron for Antimatter,” Physics 19, 132 (2026).
   The APS viewpoint accompanying the paper. A short, non-specialist account of the experiment’s significance. Useful for readers who want an accessible entry point before tackling the PRL paper itself.
3. X. Fan, T. G. Myers, B. A. D. Sukra, G. Gabrielse, “Measurement of the Electron Magnetic Moment,” Phys. Rev. Lett. 130, 071801 (2023).
   The electron counterpart of the positron measurement, at about one part in 10¹³. The positron result will be compared against this value to test matter–antimatter symmetry in the lepton sector.
4. S. Peil, G. Gabrielse, “Observing the Quantum Limit of an Electron Cyclotron,” Phys. Rev. Lett. 83, 1287 (1999).
   The first one-electron quantum cyclotron. This paper established the experimental template that the positron version follows.
5. G. Gabrielse, H. Dehmelt, “Observation of Inhibited Spontaneous Emission,” Phys. Rev. Lett. 55, 67 (1985).
   The original demonstration that a trap cavity suppresses cyclotron decay. Provides the physical basis for the “cavity as context” reading developed in Section 6.
6. L. S. Brown, G. Gabrielse, “Geonium Theory: Physics of a Single Electron or Ion in a Penning Trap,” Rev. Mod. Phys. 58, 233 (1986).
   The standard reference for the energy levels, frequencies and magnetic-bottle coupling used throughout the essay. The technical backbone of single-particle trap physics.

The cellular automaton

7. G. ‘t Hooft, The Cellular Automaton Interpretation of Quantum Mechanics, Fundamental Theories of Physics 185 (Springer, 2016); arXiv:1405.1548. Open access.
   The fullest statement of ‘t Hooft’s position. Chapter 2 gives the cogwheel model, its evenly spaced spectrum, and the theorem that every finite deterministic reversible system is a set of cogwheels. Chapter 13 gives the continuum limit and the harmonic oscillator. Chapter 15 uses evolution operators with entries −1, 0, +1 — balanced-ternary values. Chapters 3–5 treat templates, the Born rule and superdeterminism. Chapters 8–9 state the open problems with interactions.
8. A. Einstein, B. Podolsky, N. Rosen, “Can Quantum-Mechanical Description of Physical Reality Be Considered Complete?” Phys. Rev. 47, 777 (1935).
   The classic statement that quantum theory does not describe what is. ‘t Hooft explicitly places his work in this line of thought.
9. J. S. Bell, “On the Einstein Podolsky Rosen Paradox,” Physics 1, 195 (1964).
   The theorem ‘t Hooft answers with superdeterminism, and which the net answers by placing the observer on the path. Essential context for the debate over determinism and quantum foundations.
10. E. Fredkin, “Digital Mechanics,” Physica D 45, 254 (1990).
    An early programme for discrete, deterministic physics. Acknowledged by ‘t Hooft as a direct influence on the cellular automaton interpretation.

The net

11. J. Konstapel, “The Vacuum.Net Theory: Foundational Paper” (Leiden, 2026), constable.blog.
    Sets out the axioms of the net, the role of closure, the double cover and the quaternionic ground. The single axiom — everything is a closed line — originates here.
12. J. Konstapel, “Everything Has Its Number in the Maze” (Leiden, 2026), constable.blog.
    The balanced-ternary ground of the net and the address theorem. Explains why three values, and why no minus sign is needed.
13. J. Konstapel, “The Net Has No Hidden Clock” (Leiden, September 2026), constable.blog.
    The exhaustive check that only the identity and the shift survive as local, conserving, reversible rules on the ring. The basis of Sections 3 and 4.3 of this essay.
14. J. Konstapel, “The Current State of the Vacuum.Net Theory” (Leiden, 3 September 2026), constable.blog.
    The two-layer law — address layer and tension layer — used throughout Sections 3, 6 and 8.
15. P. Rowlands, Zero to Infinity: The Foundations of Physics (World Scientific, 2007).
    The nilpotent rewriting of the Dirac equation and the principle of zero totality. Behind the reading of the neutral positronium pair as +1 and −1 making 0.
16. T. Hattori, S. Takesue, “Additive Conserved Quantities in Discrete-Time Lattice Dynamical Systems,” Physica D 49, 295 (1991).
    The flux-form characterisation of conserving rules, used in the exhaustive check of Section 3.
17. D. Das et al., “Algorithms for Ternary Number System” (2012).
    On counting ternary words up to rotation. The source of the necklace correction to the address count used in Section 5.

Jump to the article here. ( DOI: 10.13140/RG.2.2.18966.92483 ).

J.Konstapel,Leidem,25-10-2026.

In May 2026, Roma Continua won A Vision for Rome, the international competition of the Fondazione Roma REgeneration. The team, led by IT’S and OMA with OKRA as landscape architect,

Introduction

Rome is usually told as a story of rise, fall and rebirth. This essay takes a different cut. It follows the document The Path of Rome (J. Konstapel, Leiden, 2026) and reads the whole history of the city as one rung on a scale of threes. Layer 56, in the ladder of the Vacuum.Net theory. On that rung, Rome moves in three great steps of roughly 840 years each. A rise from 753 BCE to about 150 CE. A fall from 150 to about 1000. A long standstill until 1870. A fourth step began in 1870 and is still running.

1. The Ground: Volcano, Tuff and River

Rome’s story begins deep in the Pleistocene. The city stands on the ash of two volcanoes, in the valley of one river.

The Colli Albani, twenty to thirty kilometres south-east of the centre, have been active for about 600,000 years. Their great explosive phase left a caldera of roughly 10 by 12 kilometres. The Monti Sabatini to the north-west last erupted about 70,000 years ago. Beneath the city itself lie volcanic units: the Palatino tuff (about 533,000 years old) and the Tufo Lionato of the Capitoline (about 355,000–366,000 years old). Roman concrete owes its durability to ash from the Alban Hills (Jackson et al. 2014). The standard work on this ground is Heiken, Funiciello and De Rita, The Seven Hills of Rome (2005).

The volcano is dormant, not extinct. The last eruption of the Alban Hills was about 36,000 years ago, at the Albano maar. Marra et al. (2016) compute a recurrence of roughly 31,000 years and measure uplift of 2–3 mm per year above magma a few kilometres deep. INGV sees no eruption in the short or medium term. At Cava dei Selci, an average of 12 tonnes of CO₂ escapes from the soil every day. Homes were evacuated there. The debate over Holocene activity is unresolved; Funiciello et al. (2003) date a lake overflow to around 398 BCE, after which the Romans dug a drainage tunnel. Marra and Karner question those dates.

The river is the second actor. The Tiber is 406 km long and carries about 239 m³ per second through the city. About 2,600 years ago — strikingly close to the traditional founding — 4 to 6 metres of silt were laid down in a single century. By the late third century BCE a stable plain lay 6–8 m above sea level (Marra et al. 2018). The Forum basin was a marsh, raised with up to 20,000 m³ of fill in the regal period (Ammerman 1990).

Floods shaped the city’s memory. Aldrete (2007) collects 42 flood reports between 414 BCE and 398 CE. The record at the Ripetta gauge is 19.56 m, set at Christmas 1598. In December 1870, three months after the breach of Porta Pia, the water stood at 17.22 m. The muraglioni embankments, built from 1876 to 1926, ended the great floods. Since then the river has overtopped its banks only three times, last in 1937.

Two movements frame the site. The volcano pushes the land up. The delta at Fiumicino sinks by up to 15–19 mm per year (Delgado Blasco et al. 2019). The city lies on the hinge between them. That is a reading, but the measurements behind it are hard.

2. The Beginning: A Crossing, Not a Hill

Popular imagination puts Rome’s origin on a hill. The archaeology says otherwise. The oldest evidence of settlement sits at the river: a ford, a salt route, a market.

The deep prehistory is rich. At Castel di Guido, west of the city, hominins butchered straight-tusked elephants about 400,000 years ago and made 98 bone tools (Villa et al. 2021). La Polledrara preserves an elephant trapped in mud with more than 500 stone tools. The two Neanderthal skulls of Saccopastore have been redated to about 250,000 years (Marra et al. 2015). Casal de’ Pazzi, about 200,000 years old, yielded a skull fragment beside elephant, aurochs and hippopotamus.

Closer to the founding, cores at Sant’Omobono — at the foot of the Capitoline, near the Tiber Island — hit occupation layers from c. 1200–1000 BCE, alternating with flood deposits (Brock & Terrenato 2016). The Capitoline carries a terraced Late Bronze Age settlement with 9th-century metalworking. From the salt pans at the river mouth the Via Salaria ran inland. Herders crossed at the Forum Boarium, where the water was shallow and slow.

Was there a founding act at all? Varro reckoned back to 21 April 753 BCE. Carandini defends a real act of founding, a wall on the Palatine. Fulminante (2014) sees a gradual coalescence, with a step in the 10th–9th century and a larger one in the mid-8th. The Lapis Niger (c. 570–550 BCE) carries the oldest Latin, including the word for king. The debate is open. What is not in doubt is the pattern: people gathered at the crossing.

The Vacuum.Net reading makes this literal. The theory’s first rule of arising says: a movement leaves a place through a crossing. Rome’s oldest sign is the ford. Its oldest god, Janus, is the god of the passage. The fit is elegant. It is an interpretation, and the document says so.

3. The Rise: 753 BCE to c. 150 CE

From village at the ford to the first city of a million people took just over 900 years. The engine was a closed loop: grain and water in, waste out.

How large the city became is debated. Morley (1996) and most scholars hold 850,000 to one million under Augustus. Storey (1997), working from the housing densities of Pompeii and Ostia, arrives at about 450,000. D’Eramo names a peak of 1.5 million in the second century. The spread is wide, and the document records it honestly.

The grain dole under Augustus reached about 200,000 recipients. Imports ran at an estimated 150,000 to 272,000 tonnes per year. Josephus reports grain from Africa for eight months and from Egypt for four. The harbour moved to Portus: Claudius began in 42 CE, Trajan added a hexagonal basin of 715 m between 100 and 112.

Water was the second half of the loop. Eleven aqueducts, from the Aqua Appia (312 BCE) to the Alexandrina (226 CE), delivered a nominal 1.1 million m³ per day; hydraulic studies put the real flow at about two-thirds of that. The Cloaca Maxima carried it away. The Aurelian Wall (271–275) enclosed 13.7 km².

Two independent cycle models describe this era. Turchin and Nefedov (2009) distinguish a Republican cycle (350–30 BCE) and a cycle of the Principate (30 BCE–285 CE), each with population growth, elite overproduction and fiscal crisis. The Vacuum.Net reading notes that both cycles — about 315–320 years — land on the same ladder layer, layer 54, as a phase of the Bakhtin cycle. Three layer-54 steps make one layer-55 step. That equals the rise. Computed, and neat. Whether it is more than arithmetic is the open question at the heart of the theory.

4. The Fall: c. 150 to c. 1000

In about 850 years Rome shrank from a (debated) million to 20,000–30,000 inhabitants. The fall came in waves.

Plagues struck first. The Antonine Plague reached Rome in 166; Cassius Dio reports up to 2,000 deaths a day in 189. The Plague of Cyprian arrived in 251. The Justinianic Plague from 541 is sharply contested: older estimates run from a third to half of the population dead; Mordechai et al. (2019) call it “an inconsequential pandemic.”

Climate compounded the shocks. Kyle Harper (The Fate of Rome, 2017) explains the fall through the end of the Roman Climate Optimum, the pandemics, and the Late Antique Little Ice Age (536–660). Newfield and others call this too deterministic. Büntgen et al. (2016) document the cold after the eruptions of 536, 540 and 547; McCormick et al. (2012) document the earlier optimum with honest caveats.

Then came the sacks. Alaric in August 410. The Vandals in June 455. The decisive blow, in the reading of this document, was the siege of Vitiges (537–538): the aqueducts were cut, the mills on the Janiculum stopped, and Belisarius built floating mills on the Tiber. According to Bury, the Romans then drew water from the river and from wells for a thousand years. When Totila took the city in December 546, Procopius counted about 500 citizens left.

The population curve is steep. About 800,000 under Constantine. Perhaps 60,000 in the early 6th century. 20,000–30,000 around 1000 (Wickham). Ward-Perkins (2005) reads a real material collapse: roof tiles and coins disappear, even cattle shrink. Wickham (2005) places the break at the end of the Roman tax state. Scheidel (2017) calls it a case of state failure as a “great leveler.”

The Vacuum.Net reading adds a structural note: rise (903 years) and fall (850 years) are two nearly equal steps on the same ladder rung, with opposite signs. And the fall of 537 is read as the loss of a closure: the water loop was cut, and the city fell back from knot to loose windings. It is a compelling image. It is not a measurement.

5. The Standing City: c. 1000 to 1870

After the fall Rome became neither imperial city nor ruin. It became a third thing: the city that stood.

Around 1000, 20,000 to 30,000 people lived in the abitato in the Tiber bend. Around it, to the Aurelian Wall, lay the half-rural disabitato. The Saracen raid of 846 led to the Leonine Wall. The Normans sacked the city in 1084, though Hamilton (2003) argues the destruction was mostly later memory. In 1143 a Senate rose again on the Capitoline. In 1347 Cola di Rienzo seized power by appealing to ancient law.

The centre left and returned. The pope resided in Avignon from 1309 to 1377. The Western Schism lasted until 1417. In 1300 Boniface VIII proclaimed the first Jubilee, and the pilgrim became the city’s new circulation.

Recovery tracked water. Nicholas V restored the Acqua Vergine in 1453. Sixtus V built the Acqua Felice (1585–87), Paul V the Acqua Paola (1612). With them the population rose from 55,035 (1526–27) to about 110,000 around 1600. The Sack of 1527 killed 6,000 to 12,000; recovery to the earlier level took about 33 years — one layer-52 step in the reading, the same layer as the Kondratiev wave.

Sixtus V then redrew the city in five years. With Domenico Fontana he drew straight streets between the basilicas and re-erected the ancient obelisks at the nodes. A priest returning in 1595 said he could no longer find Rome in Rome. In the reading of the theory, this is the first rule of arising applied deliberately: lines between nodes, crossings marked. Fascism later drew lines through the nodes and displaced the residents. The difference matters, and the document draws it well.

After 1600 Rome hovered around 150,000–185,000. Under French occupation it fell to 135,000 by 1814. In 1870 it was only the fourth city of Italy.

6. The Third Rome: 1870 to 1981

Then came the fastest growth in the city’s history. On 20 September 1870 the bersaglieri broke through at Porta Pia. In 1871 Rome became the capital of Italy, with 213,633 residents by census. The censuses then climb relentlessly: 386,626 (1891), 663,848 (1921), 1,403,307 (1941), 2,155,093 (1961), and the peak of 2,797,337 in 1981.

The plans struggled to keep up. The Viviani plan (1873) never became law. The plan of 1883 held for 25 years. The Sanjust plan (1909) first allowed growth beyond the Aurelian Wall. The plan of 1931 followed.

Fascism cut through the fabric. The Via dell’Impero (1932) removed the Alessandrino quarter. The Via della Conciliazione displaced 1,236 families — 4,992 people — by a municipal count of 1937. Twelve borgate were built in the Agro Romano between 1930 and 1937. The E42 exhibition requisitioned 400 hectares in 1938 and never took place.

After the war, building outran the state. Public bodies own 2.7 per cent of Italy’s housing stock. Much of Rome’s growth ran through abusivismo, from the enclosed balcony to districts of ten-storey blocks (D’Eramo 2017). Pasolini and Fellini found the living city in exactly those borgate and at the edge — not in the centre.

Again the water came first. The new growth began with the muraglioni closing the river: embankments instead of aqueducts, but the same lever. The price, in the document’s reading, was the loss of the river as a living edge.

7. The Present: 1981 to 2026

Since 1981 Rome has not grown. The municipality counts 2,747,395 inhabitants (1 January 2025); the metropolitan city 4.22 million. The share aged 65 and over rose from 19.0 per cent (2002) to 24.0 per cent (2025). The labour-market area around Rome, meanwhile, grew from 3.48 million (2001) to 4.08 million (2023). The city ages and the edge grows.

Visitors replace growth. In 2025 Rome recorded 22.9 million arrivals and 52.9 million overnight stays, a record. The Jubilee drew 33.4 million pilgrims from 185 countries. There are 47,090 Airbnb listings, 48 per cent of them in the historic centre.

Housing is under strain. Between 2011 and 2022 there were 69,000 eviction orders for arrears. About 162,000 of 1.44 million dwellings stand empty. Social housing is 5.7 per cent of the stock. Some 114,000 households are at high risk (Sapienza/MEMOTEF 2025).

The big works are finally moving. Metro C opened the Colosseo and Porta Metronia stations in December 2025; the full 29 km line follows around 2035. The Santa Palomba waste-to-energy plant has been under construction since May 2026. A second Peschiera aqueduct line starts in 2027; network water losses fell from 43 to 27 per cent.

Rome lost Expo 2030 to Riyadh in November 2023, and regrouped. Stefano Boeri’s Laboratorio Roma050 (June 2025) proposes the Tiber as ecological axis and more than 250 “micro-cities.” And Roma Continua won its competition in May 2026: five green corridors along the Tiber and its tributaries, five mobility hubs, five knowledge clusters, reuse of vacant heritage, a horizon of 25 years. The core principle is “recalibration instead of expansion.”

8. The Instrument and Its Limits

What should a business-minded reader make of the Vacuum.Net apparatus — the ladder of three, the layers, the closure rule? Three observations.

The arithmetic is real, the reading is chosen. The document computes every duration on a logarithmic scale of threes, anchored to a natural unit of length from hydrogen. The positions are correct arithmetic. But the turning points — the peak around 150, the low around 1000 — are chosen by the author, not produced by a rule. The document admits this in a candid sensitivity table: with a low around 600 or 1300, the equality of the three steps dissolves. The mean step (874 years) is fixed only by the start (753 BCE) and end (1870), not by the middle.

The family patterns may be selection. About 35 quantities were placed on the ladder and the hits picked out. With three families, roughly a third falls into each by chance. The document flags this itself.

A real test is specified. Fix a rule in advance for picking turning points from a smoothed population curve. Then compare with the 840.6-year step. Then apply the same rule to Athens, Constantinople and Alexandria. If layer-55 steps appear there too, the reading becomes a hypothesis with evidence. Until then it is a lens, not a law.

That honesty is the document’s strength. Every statement carries a status: source, computed, reading or open. A reader can disagree with the lens while trusting the facts.

9. What the Reading Gets Right

Even as a lens, the Vacuum.Net reading produces insights with practical value.

Water is the city’s strand. This is the strongest historical claim in the document, and it survives without the ladder. Rome grew when eleven aqueducts closed the water loop. It collapsed when the loop was cut in 537. It recovered with the Vergine, Felice and Paola. Its modern growth began with the muraglioni. Today 70 per cent of the water comes from Peschiera and 27 per cent leaks away — and the city is again investing in the loop, in supply and in outflow. Any 25-year plan that does not name water as its first principle is missing the one variable that has decided Rome’s fate for 2,500 years.

The pair and the crossing. The mythic reading is that Rome’s pairs either hold or break. Romulus kills Remus at the boundary; Leo the Great replaces the fratricidal pair with Peter and Paul, the pair that holds. The city prospers when its pairs support each other — city and province, pope and pilgrim, centre and periphery — and fails when they break. Applied to Roma Continua: twin each peripheral hub with a place in the centre, so support runs both ways and does not break at the GRA ring road.

The pilgrim as circulation. The standing city lived from those passing through. Today’s tourist is the same stream — 52.9 million overnight stays — but it runs through the centre and starves the districts. The 2050 Jubilee falls inside the plan’s horizon. It is the natural test case for spreading the stream.

Count the scales. The document’s scale argument is concrete. The ancient city (13.9 km²) and the 5-kilometre cycling reach of a forum share the same size class. The municipality could hold about 93 ancient-city units; the metropolis about 386. That points to a hundred forums rather than five, in the order of Boeri’s 250 micro-cities — a fine mesh rather than a ring. Whether one accepts the ladder or not, the arithmetic of areas is plain.

10. Assessment of Roma Continua

Roma Continua chooses the right form for the moment. A city that stopped growing in 1981, that is ageing, and that loses its growth to the edge should densify its closures rather than move its boundary. Recalibration instead of expansion is not a slogan; it is the only strategy that matches the demographic facts.

The palimpsest principle — writing on what exists, reusing vacant heritage — matches both the city’s own myth of renovatio and the pragmatic need. Freud described Rome in 1930 as a city in which no layer is ever lost. The plan embodies that image ninety years later.

Three gaps stand out.

1. Water is not the first principle. Care, beauty, knowledge, movement and reuse: water hides inside the green corridors. The city’s whole history argues it should be first. Each hub should close its own loop of rainwater, storage, shade and outflow, and the Tiber should become an edge again rather than a gutter behind the muraglioni.
2. Fives where the structure argues for threes — and for many more than five. Five corridors, five hubs, five clusters. The scale arithmetic of the metropolis points to a hundred forums, not five. Boeri’s finer grain is the closer fit.
3. The prediction the plan must face. In the document’s reading, the current sub-step closes around 2057 with a contraction, not new growth. ISTAT expects Italy to fall to about 54.7 million by 2050, with 34.5 per cent aged 65 and over. A 25-year plan calibrated for contraction is resilient. A plan that quietly assumes return to growth is not. Roma Continua, to its credit, does not assume growth.

Conclusion

The facts of Rome’s path are well established. A crossing at a ford, on volcanic ground, at the end of a young plain. A rise of 900 years on grain and water. A fall in which the cutting of the aqueducts mattered as much as any plague. A thousand years of standing still, lived on pilgrims and returned water. A century of explosive, planless growth. And now four decades of standstill, ageing and visitors, with the water loop again under investment.

The Vacuum.Net reading places all of this on one rung of a ladder of threes, in steps of about 840 years. As measurement, it does not yet hold; the document says so itself. As a reading, it has three durable yields. It puts water where it belongs: as the first principle of any Roman future. It makes the pair — centre and periphery, hub and counterpart — the unit that must not break. And it reframes Roma Continua’s greatest virtue: a city that refounds itself from what is already there, again and again.

That is the genius of the city, against the spirit of the age. The spirit of the age asks Rome to grow, compete and win expos. The genius of the city has never been growth. It is the passage, the pair, the fire kept and the water fetched. The next 25 years will be judged by how well the plan keeps the thresholds open and the pairs intact.

---

Annotated Reference List

Theory (constable.blog, J. Konstapel)

• Konstapel, J. (2026). Vacuum.Net Theory. The framework essay: mass, energy and space as one net; the measuring rod of 45.6 nm; the two ways energy is released. Annotation: the foundation on which the ladder rests; read first.
• Konstapel, J. (2026). Geoneutrinos and the Ladder of Three. The rule for positions and layers, the three families (A, B, C), and the assignment of Italy the state +1. Annotation: supplies the method used for every layer computation in the source document; the family division drives the “volcano and city breathe together” claim.
• Konstapel, J. (2026). The Way Before the Net. The three rules of arising: crossing (R1), return gives birth (R2), mutual support (R3). Annotation: these three rules are applied throughout the essay — to the ford, to renovatio, to the Romulus/Remus pair.
• Konstapel, J. (2007). The Worldwide Carnival. The Bakhtin cycle of Western culture in phases of 250 years. Annotation: provides the second, independent cycle model that coincides with Turchin’s Roman cycles on layer 54.
• Konstapel, J. (2024). History and Future are a Fractal Process. Kondratiev, precession and the five phases. Annotation: the time side of the argument; the Kondratiev wave and the 25-year plan term share layer 52.
• Konstapel, J. (2025). De Resonante Stad. The city as rhythm, fine mesh and kami. Annotation: the basis for the advice on the number of forums and for the treatment of genius loci.

Geology, Climate and Ground

• Heiken, G., Funiciello, R. & De Rita, D. (2005). The Seven Hills of Rome: A Geological Tour of the Eternal City. Princeton University Press. Annotation: the standard work on the volcanic ground beneath every hill and monument; the best entry point for the geology.
• Marra, F. et al. (2018). “Rome in its setting.” PLoS ONE 13(3). Annotation: documents the rapid raising of the Tiber plain shortly before the founding — the fact that ties the founding to the river’s return of silt.
• Marra, F. et al. (2016). “Assessing the volcanic hazard for Rome.” Geophysical Research Letters 43. Annotation: computes the ~31,000-year recurrence and measures 2–3 mm/yr uplift of the Alban Hills; read with Giordano’s dissenting comments in Scientific American.
• Marra, F. & Karner, D. B. (2005). “The Albano maar: active or dormant volcano?” Il Quaternario 18(2). Annotation: the alternative ~45,000-year repose estimate; the disagreement itself feeds the layer-58/59 family reading.
• Aldrete, G. S. (2007). Floods of the Tiber in Ancient Rome. Johns Hopkins University Press. Annotation: collects all 42 ancient flood reports, 414 BCE–398 CE; the empirical base for every flood statement.
• McCormick, M. et al. (2012). “Climate Change during and after the Roman Empire.” Journal of Interdisciplinary History 43(2). Annotation: the Roman Climate Optimum and the instability from 150–200 CE, stated with explicit caveats.
• Büntgen, U. et al. (2016). “Cooling and societal change during the Late Antique Little Ice Age.” Nature Geoscience. Annotation: the cold decades after the eruptions of 536, 540 and 547; the physical core of the “fall” climate argument.
• Harper, K. (2017). The Fate of Rome: Climate, Disease, and the End of an Empire. Princeton University Press. Annotation: the leading climate-and-disease account of the fall; read with Newfield’s critique (BMCR 2019.10.21), which calls it too deterministic.
• Delgado Blasco, J. M. et al. (2019). “Measuring urban subsidence in the Rome metropolitan area.” Remote Sensing 11(2). Annotation: satellite measurement of the sinking Fiumicino delta, up to 15–19 mm/yr.
• CMCC (2025). Rome’s warming climate in numbers. Annotation: measured and projected heat to 2050: +1.5 to +1.9 °C and 22–28 more tropical nights.

Origins and Antiquity

• Brock, A. L. & Terrenato, N. (2016). “Rome in the Bronze Age.” Antiquity 90. Annotation: the Sant’Omobono cores — settlement at the river from c. 1200–1000 BCE; the key evidence against the hill-first story.
• Fulminante, F. (2014). The Urbanisation of Rome and Latium Vetus. Cambridge University Press. Annotation: gradual coalescence rather than a founding act; the main scholarly counterweight to Carandini.
• Marra, F. et al. (2015). Saccopastore redating. Quaternary Science Reviews 129. Annotation: the Neanderthal skulls of Rome redated to ~250,000 years.
• Storey, G. R. (1997). “The population of ancient Rome.” Antiquity 71. Annotation: the low estimate (~450,000) from Pompeii/Ostia housing densities; the essential counterpoint to Morley (1996), Metropolis and Hinterland.
• Turchin, P. & Nefedov, S. (2009). Secular Cycles. Princeton University Press. Annotation: the two Roman cycles (Republic, Principate); chapters 6–7 supply the independent cycle model that the ladder reading places on layer 54.
• Ward-Perkins, B. (2005). The Fall of Rome and the End of Civilization. Oxford University Press. Annotation: the fall as material collapse — tiles, coins, cattle; the strongest “real decline” position.
• Wickham, C. (2005). Framing the Early Middle Ages. Oxford University Press. Annotation: the fall as the end of the Roman tax state; also the source for the 20,000–30,000 population around 1000.
• Mordechai, L. et al. (2019). “The Justinianic Plague: an inconsequential pandemic?” PNAS. Annotation: the minimalist reading of the plague of 541; the debate it triggered is exactly why the document marks the plague figures “source, with debate.”

Middle Ages to 1945

• Krautheimer, R. (1980). Rome: Profile of a City, 312–1308. Princeton University Press. Annotation: the classic picture of abitato and disabitato — the inhabited core within the half-rural shell.
• Wickham, C. (2015). Medieval Rome: Stability and Crisis of a City, 900–1150. Oxford University Press. Annotation: the population plateau and the commune of 1143.
• Karmon, D. (2005). “Restoring the Ancient Water Supply System in Renaissance Rome.” The Waters of Rome 3. Annotation: water as papal politics; the empirical spine of the reading “water is the strand.”
• Insolera, I. (1962; 2011 with P. Berdini). Roma moderna. Einaudi. Annotation: the standard work on planning after 1870; indispensable for the plans-and-failures narrative.
• Kallis, A. (2014). The Third Rome, 1922–43. Palgrave. Annotation: Fascist Rome as model city and renovatio; the key source for the demolitions as symbolic landscape.
• Baxa, P. (2010). Roads and Ruins. University of Toronto Press. Annotation: the fascist road-building programme as the destruction of living quarters.
• Moure Cecchini, L. The Via della Conciliazione. Annotation: the displacement of 4,992 residents, documented from the 1937 municipal count.

Myth and the Genius of the Place

• Leo the Great. Sermon 82. Annotation: Peter and Paul as the true founders who replace the fratricidal first pair — the mythic pivot of the “pair that holds.”
• Pliny the Elder. Natural History 3.65. Annotation: the secret name of Rome and the execution of Valerius Soranus for revealing it; the Pisa article reconsiders all ancient sources.
• Warde Fowler, W. (1912). “Mundus patet.” Journal of Roman Studies 2. Annotation: the pit to the underworld opened three days a year; the “three” that the net reading leans on.
• Servius, on Aeneid 2.351. Annotation: the Capitoline shield to the genius of Rome, sive mas sive femina — the spirit honoured without being named.
• Bakhtin, M. (1965). Rabelais and His World. Annotation: carnival as the zero of order; the theory behind the Saturnalia reading and the Worldwide Carnival cycle.
• Freud, S. (1930). Civilization and Its Discontents, chapter 1. Annotation: Rome as the image of the mind in which no layer is ever lost — the palimpsest, ninety years before Roma Continua.
• Jung, C. G. (1962). Memories, Dreams, Reflections. Annotation: the man who fainted before he could visit Rome — the force of the place as a biographical fact.

After 1945 and the Future

• D’Eramo, M. (2017). “The Not So Eternal City.” New Left Review 106. Annotation: abusivismo, standstill and the city without a state; the sharpest short analysis of modern Rome, and the source for the 1870 fourth-city ranking.
• Clough Marinaro, I. & Thomassen, B. (eds.) (2014). Global Rome. Indiana University Press. Annotation: Rome as a world city after 2000; the academic frame for the present phase.
• Cellamare, C. (ed.) (2016). Fuori raccordo. Donzelli. Annotation: self-organised districts beyond the GRA as a force of their own; the alternative to top-down corridors.
• Sapienza/MEMOTEF (2025). Disagio abitativo a Roma. Annotation: the housing emergency in numbers — 69,000 eviction orders, 162,000 empty dwellings, 5.7 per cent social housing.
• Politecnico di Torino (2026). Airmap Roma. Annotation: short-term rentals and the growth of the edge; 47,090 Airbnb listings, 48 per cent in the historic centre.
• IT’S, OMA, OKRA et al. (2026). Roma Continua. Annotation: the winning competition entry and the occasion of this essay; five corridors, five hubs, 25 years, recalibration instead of expansion.
• Boeri, S. et al. (2025). Laboratorio Roma050. Annotation: the municipality’s own vision — the Tiber, archaeology and 250+ micro-cities; the finer-grained alternative that the scale arithmetic favours.

Article

Dit is een toepassing van de Maze.

J.Konstapel,Leiden,25-10-2026.

Aanleiding: Leiden Bio Science Park maakt investeringsagenda van 2,5 miljard euro; zonder verandering dreigt stilstand

3.3.1 Stand van zaken bij Voortgangsrapportage

Van regionale Human Capital-aanpak naar nationale biotechstrategie

Ik kreeg het niet voor elkaar om de investeringagenda te pakken te krijgen vandaar dat ik zelf maar een heb gemaakt.

Het Leiden Bio Science Park vraagt 2,5 miljard euro voor de komende tien jaar. Daarvan gaat 1,754 miljard naar bereikbaarheid, gebouwen en het aantrekken van mensen, en 633 miljoen naar onderzoek. De agenda zelf is niet openbaar.

Daarom heb ik de voorafgaande vraag onderzocht. Welke innovaties komen er in de medische (rode) en de agrarische (groene) biotech? Waar lopen ze vast? En wat verandert de vacuum.net-theorie daaraan?

Wat er aankomt

In totaal heb ik 38 innovaties in kaart gebracht. Een aantal gaat in 2026–2027 over een drempel:

• Gentherapie in het lichaam zelf. De FDA beslist in maart 2027.
• CAR-T bij auto-immuunziekten. Patiënten zijn na een jaar medicijnvrij.
• Een kankervaccin per patiënt. Het werkt in een fase 3-studie.
• Medicijnen door de bloed-hersenbarrière. Het eerste is goedgekeurd.
• Een gen-bewerkt varken. Goedgekeurd in vijf landen.
• Gen-bewerkte gewassen in Europa. Toegestaan vanaf juli 2028.

Kweekvlees en verticale landbouw krimpen: Meatable in Leiden sloot in december 2025.

De innovatie komt niet uit de grote bedrijven. Twee derde van de nieuwe Amerikaanse medicijnen in 2024 kwam van kleine biotechbedrijven. Grote bedrijven kopen ongeveer de helft van hun pijplijn in.

De methode: eerst zien waar het vastloopt

Ik heb de theorie toegepast zoals in mijn stuk Read Before You Remove. Dat stuk neemt het kankervaccin per patiënt en ziet waar het vastloopt: een fabriek per patiënt, negen weken wachten, en 17% van de patiënten valt af. Dan volgt een volgorderegel uit de theorie: eerst behandelen, dan opereren. Zo wordt de tumor zijn eigen vaccin, zonder fabriek.

Hetzelfde heb ik bij alle 38 innovaties gedaan.

Wat eruit kwam

• Bij 30 van de 38 levert de theorie een concrete zet op. Het is altijd een volgorde, een structuur of een ontwerpregel, nooit een dosis.
• In 16 gevallen doet de beste praktijk dat al. Voorbeelden: resistentiegenen tegelijk inbouwen in plaats van na elkaar, een onverenigbare ent overbruggen met een derde plant, eerst in beeld brengen en dan bestralen, eerst behandelen en dan opereren. De theorie komt dus uit op wat werkt. Dat is geen ontdekking, maar het laat zien dat de regels niet willekeurig zijn.
• 12 zijn deels bekend, met iets nieuws erbij.
• Drie zetten vond ik nergens in de literatuur:
  1. Kankervaccins die mikken op wat alle tumoren gemeen hebben. Dan is er geen fabriek per patiënt meer nodig.
  2. Bacteriemengsels als medicijn, samengesteld op wat de bacteriën doen in plaats van welke soort ze zijn.
  3. Startculturen voor fermentatie die in de natuurlijke volgorde worden toegevoegd: eerst gist, dan melkzuurbacteriën, dan azijnzuurbacteriën. Nu gaat alles tegelijk.
• Eén zet bleek fout. Tegen insecten die resistent worden tegen RNA-sprays helpt het niet om meer RNA-doelwitten te stapelen. Ze blokkeren de opname zelf, dus één mutatie zet alle doelwitten tegelijk open. De sluiting moet bij de opname zitten.
• Bij 8 innovaties levert de theorie niets op. Daar gaat het om dosis, opbrengst, energie, kosten of goedkeuring: grootheden waarover de theorie niets beweert.

Eén getal verdient aandacht. Stikstofbacteriën als los product gaven in onafhankelijke proeven in 2 van de 61 gevallen meer opbrengst. Door toeval alleen verwacht je er al 3. Het product presteert dus niet boven toeval.

Wat nog getoetst moet worden

Vier voorspellingen zeggen iets wat de gangbare biologie niet zegt:

1. Celhoekjes. Waar drie cellen samenkomen, sluiten drie gelijke en drie verschillende cellen. Twee gelijke plus één andere lekken.
2. Kristalvlakken. Kubusvlakken lossen per oppervlak sneller op.
3. Stabiliteit. Die is te meten als sluiting boven toeval.
4. Kringen. Drie bacteriën in een kring houden stand, twee niet.

Alle vier zijn te toetsen met middelen die er al zijn, een deel in Leiden.

Volledig essay, met alle 38 innovaties, de knelpunten, de zetten en de bronnen: The Future of Biotech, versie 3.0 [PDF].

Korte Samenvatting

Aanleiding: Een fase 3-studie (INTerpath-001) toont aan dat een gepersonaliseerd mRNA-vaccin (intismeran autogene) bij hoog-risico melanoom het risico op terugkeer of overlijden met 42% verlaagt, een echte doorbraak.

Het structurele probleem: Dit vaccin is echter patiëntspecifiek, de productie duurt ongeveer negen weken en 17,3% van de gescreende patiënten kon geen vaccin krijgen omdat hun tumormonster ongeschikt was.

De onderzoeksvraag: Het essay onderzoekt of de ‘vacuum.net-theorie’ een bewezen alternatief kan bieden voor hetzelfde klinische effect, zonder dat er voor elke patiënt een fabriek nodig is.

De theorie levert geen nieuw medicijn, maar een ordeningsregel met drie stappen: (1) geef immunotherapie vóór de operatie zodat de tumor als eigen vaccin dient, (2) richt je op gedeelde kenmerken van alle melanomen in plaats van patiëntspecifieke mutaties, en (3) verhoog de immuuntolerantie alleen lokaal rond de tumor.

J.Konstapel,Leiden,25,10-2062

Jump to the Article here

Cause: New mRNA Vaccine Reduces Risk of Melanoma Recurrence by Over 40 Percent

On 24 October 2026, WIRED reported a phase 3 trial called INTerpath-001. Merck and Moderna had tested a personalised mRNA vaccine in patients with high-risk melanoma. The vaccine is named intismeran autogene. It reduced the risk of recurrence or death by 42 percent. This is a real advance. It is also an advance with a structural problem.

Each dose is made for one patient, from one tumour. Manufacturing takes about nine weeks. In the trial, 17.3 percent of screened patients could not receive a vaccine at all. Their tumour sample was not good enough to build one from.

This essay asks a focused question. Does the vacuum.net theory offer an alternative route to the same clinical effect? One that does not need a factory for every patient?

The answer is yes, with a precise limit. The theory does not supply a new molecule. It supplies an ordering rule. That rule points to three moves.

1. Read before you remove. Give immunotherapy while the tumour is still in the body. Operate afterwards. The tumour then serves as its own vaccine. It is made in place, for every patient whose tumour is still present. No sequencing. No factory. This move is already tested. In cross-trial comparison, its effect matches the vaccine’s. With two checkpoint drugs, it is larger. In the Netherlands, it has been the preferred treatment for resectable stage III melanoma since October 2025.
2. Read what is shared, not what changes. Target features that every melanoma carries. Do not target the mutations that differ between patients. One product could then serve everyone. The evidence here is mixed. One clear success. One modest result. One miss.
3. Set the zero locally. Lift the body’s immune tolerance only where the tumour sits. Do not lift it everywhere. The evidence is partial and the question remains open.

The theory also produced five predictions about the vaccine trial before its detailed data were read. Two came true. One partly. One could not be scored. One was not blind. None was a sharp test. That result is reported plainly in Section 6.

The strongest finding is the first move. A theory built far from biology arrives, by its own rule, at the treatment order that the best recent trials have now established. That is a correspondence, not a proof. It is still worth having.

---

2. The Occasion

The starting event deserves a precise description.

On 24 October 2026, WIRED published a report on INTerpath-001. Merck and Moderna tested their vaccine in 1,137 people. All had high-risk melanoma that had been completely removed by surgery. Two thirds received the vaccine plus pembrolizumab, a checkpoint inhibitor. One third received pembrolizumab alone.

The detailed results were presented at the ESMO congress in Madrid as late-breaking abstract LBA1. The numbers are strong.

• Median follow-up: 24.6 months.
• Risk of recurrence or death: down 42 percent (hazard ratio 0.58).
• Recurrence-free at two years: 82.2 percent with the vaccine, 70.0 percent without it.
• Distant spread: down 38 percent (hazard ratio 0.62).
• Overall survival: no significant difference yet (hazard ratio 0.77, p = 0.33).

These are real gains. They are reported as the first positive phase 3 result for an mRNA cancer therapy.

The WIRED article also contained errors. It stated that scientists chose “up to three” molecular targets per patient. In fact the vaccine encodes up to 34, and 91 percent of patients received all 34. It stated that the vaccine was awaiting FDA approval. No filing had been confirmed by 25 October. It stated that a pancreatic cancer vaccine had greatly improved survival. That was a phase 1 study of 16 patients, without a control group.

The article closed on the real problem. The melanoma researcher Marisol Soengas put it simply: the vaccine is literally bespoke. Each dose is made for one person from one tumour. The trial adds detail to that problem. The median time to the first vaccine dose was 64 days. And 302 of 1,749 screened patients were excluded because their tumour sample was not adequate for manufacture.

That is the question this essay puts to the theory. Is there a road to the same effect that does not require a factory per patient?

---

3. The Theory, in Full

The essay must stand on its own. The theory therefore comes first. It fits in a few paragraphs.

One strand. The vacuum.net theory describes everything that lasts as one strand that crosses itself. Picture a fishing net. The net is a single cord. Where the cord crosses itself, there is a crossing. Between crossings lie the open meshes. A stretch of cord that returns to where it started, in step, is a winding.

One register. At a crossing, the cord passes over, passes under, or does not cross. The theory writes these three states as +1, −1 and 0. Such a three-valued digit is a trit. Counting with trits is called balanced ternary.

One rule. Three crossings in a row form a window. A window is closed when its three trits add up to a multiple of three. The theory’s single selection rule, its axiom N2, states: what closes stays; what does not close passes its tension on to its neighbours.

Two kinds of closure. Of the 27 possible windows, exactly 9 close. This is a proven theorem, not a choice. A window closes if and only if its three trits are all equal or all different. The three all-equal windows are the form of a crystal. They hold and do not turn. The six all-different windows are the form of living matter. Their content keeps changing while their closure holds. A seam between the two kinds is always open, and carries tension.

Four further axioms.

• N1: there is one strand, and every distinction is a place where it touches itself.
• N3, memory: at every crossing, the first passage fixes the orientation. Later passages are read relative to it. Order matters.
• N4, context: what looks like a constant is a function of the local state.
• N5, scale: the same rules hold at every depth. What does not close at one scale becomes material for the next.

The test in a body. Now apply the rule to a living body. The body carries a reading layer: the immune system. It tests every form it meets. There are three outcomes. +1: the form fits, and stays. −1: it does not fit, and is taken apart. 0: it is not read, and simply remains.

The zero outcome is essential. The body must not attack the gut wall at every meal. It must not attack the eye. It must not attack a child in the womb, who is half foreign. So the body carries official zero-tokens that switch local reading off. Physicians call them checkpoints. PD-1 is one. Without them, the body attacks itself. With them, it has a blind spot.

Cancer in one sentence. Read this way, a tumour is the remainder of a selection. For years, every abnormal cell that did not carry a zero-token was read and removed. Every cell that did carry one remained. What a surgeon cuts out is that remainder.

What the two drugs do. Pembrolizumab removes the zero-token. It does so everywhere. That is why its side effects resemble autoimmunity. The mRNA vaccine adds words for the test to read: up to 34 fragments of the patient’s own mutated proteins.

---

4. What the Rule Says About the Vaccine

Now apply the theorem on the two kinds of closure.

A melanoma is living matter. Its windows are of the mixed kind. Their content changes while their closure holds. The mutations a vaccine is built from are content. They differ between patients. They drift within one patient over time.

So a vaccine built from mutations reads the part of the tumour that changes. That is why it must be made per person. That is why it takes nine weeks. That is why one patient in six could not get one. The cost follows from what is read, not from the mRNA technology.

The rule suggests three other things to read.

The remainder itself, in place (axiom N3). Order matters. If the tumour is removed first and the immune system is trained afterwards, the trainer needs a copy of the tumour: a sample, a sequence, a factory. If the zero is lifted first, while the tumour is still there, the immune system reads the original. The tumour then is its own vaccine. It is complete and current. It needs no manufacture.

What every melanoma shares. A melanoma descends from a pigment cell. It keeps features of that lineage. It often carries proteins that many melanomas share. Reading those is reading the closure, not the content. One product could serve many patients.

The local zero (axiom N4). A zero-token is a function of the local state. It is not a fixed property of the body. That suggests lifting it only at the tumour, not everywhere.

---

5. The Experiment

The work followed the Respondent-Judge Protocol. A respondent fixed predictions in writing before the detailed trial data were looked up. The file was locked with a cryptographic hash at 11:16 UTC on 25 October 2026. The SHA-256 hash begins with e7d5f06b. A separate judge then verified the hash, found the data independently, and scored each prediction.

The predictions, with their failure criteria:

• P1. The vaccine’s relative benefit is about the same at every stage of disease.
• P2. The benefit does not rise steadily with the number of mutations in the tumour.
• P3. The survival curves do not separate in the first three months, and separate after that.
• P4. Overall survival is not yet significant at this interim analysis.
• P5. Reading the tumour in place gives a relative benefit at least as large as adding the bespoke vaccine after removal. This prediction was known to be not blind.

---

6. Evaluation and Adjustment

Prediction	Verdict	Evidence
P1	Hit, coarse	Hazard ratio 0.60 in the lower-risk group, 0.57 in the higher-risk group. Only two pooled groups are public.
P2	Not decidable	No phase 3 data on mutation load are public. The phase 2b trial fits the prediction (0.65 against 0.59) but cannot score it.
P3	Partial	No early separation: hazard ratio 0.99 over months 0–6. But the benefit appears only after month 6 (0.42), not after month 3.
P4	Hit	Overall survival hazard ratio 0.77, p = 0.33.
P5	Hit, not blind	Neoadjuvant pembrolizumab (SWOG S1801): 0.58. Neoadjuvant ipilimumab plus nivolumab (NADINA): 0.32, later 0.40. Vaccine: 0.58.

The verdict must be honest. P1 and P4 were hits. But P4 was what any oncologist would have expected. P3 failed on its timing. Its first half held for a reason the theory did not need: the vaccine simply starts nine weeks late. None of the blind predictions discriminated sharply between this theory and ordinary immunology.

One adjustment follows. The delay in P3 is larger than the manufacturing delay alone. Months 0 to 6 show no measurable effect, though most patients had their first dose by month 2 to 3. Reading takes time after the words arrive. Future predictions from this theory should allow a reading time of about three months after the first dose, not zero.

P5 carries the weight of this essay. It must be read with care. The trials differ in endpoint, in comparator and in patients. S1801 and NADINA enrolled people whose tumour could still be seen and felt. S1801 also included resectable stage IV. INTerpath-001 also enrolled stage IIB and IIC, where only the primary skin tumour had been removed. Cross-trial comparison is suggestive, never decisive.

With that caution, the picture is still clear. Reading the tumour in place matches the vaccine’s effect with pembrolizumab alone. With two checkpoint drugs, it exceeds it, though part of that gain comes from the second drug. The Dutch melanoma guideline drew the practical conclusion on 9 October 2025: for resectable stage III melanoma, treatment before surgery is preferred.

---

7. The Alternative

The answer to the research question can now be stated.

Move 1 — Read before you remove

For melanoma that can still be found in the lymph nodes, give the checkpoint drug before the operation. The tumour does the work of the vaccine. No sample needs to qualify for manufacture. No sequence needs to be read. No dose needs to be made. Status: corresponded, and guideline-preferred in the Netherlands for resectable stage III. The theory did not discover it. It arrives at it by its own rule.

Move 2 — Read what is shared

For patients who have no tumour left to read in place, the bespoke vaccine remains one route. The other is a product aimed at what melanomas share. The evidence is uneven.

• Tebentafusp, aimed at the pigment-cell protein gp100, reduced the risk of death by 49 percent in uveal melanoma of the eye.
• BNT111, built on four shared melanoma proteins, gave a response rate of 18.1 percent with cemiplimab, against 13.6 percent for cemiplimab alone.
• IO102-IO103, aimed at two shared immune-escape proteins, narrowly missed its phase 3 goal (hazard ratio 0.77, p = 0.056).

Status: open.

Move 3 — Set the zero locally

Treat the tumour where it lies. Open it to reading without lifting the body’s tolerance everywhere. The first approved example is T-VEC, a modified virus injected into the tumour. In its main trial, durable responses rose from 2.1 to 16.3 percent. Before surgery, in a phase 2 trial, it lowered the risk of death with a hazard ratio of 0.54. That figure was measured with a wide 80 percent confidence interval, and survival was not the trial’s primary endpoint. Status: open.

Upstream of all three

Fewer remainders mean fewer recurrences. In the Nambour trial in Australia, daily sunscreen halved new melanomas: 11 against 22. Among Australians aged 15 to 24, melanoma incidence now falls by about 5 to 7 percent a year. Those studies do not prove the cause. Status: corresponded.

---

8. One Prediction for the Next Trial

A theory must say something before the data arrive. Here is one, fixed with this essay.

The rule says that once the tumour has been read in place, the reading has happened. A bespoke vaccine added afterwards supplies words the immune system already has.

Hypothesis H-1. In patients who receive neoadjuvant immunotherapy and have no viable tumour left at surgery, adding a bespoke neoantigen vaccine afterwards will give a hazard ratio for recurrence of 0.80 or higher. In patients with substantial viable tumour left at surgery, it will give 0.65 or lower.

Failure. The prediction fails if the vaccine’s benefit is equal in both groups, or larger in the first.

This is checkable. It costs nothing to state. And it turns the theory’s reading into a claim that a trial can refute.

---

9. What the Theory Does Not Do

The theory supplied no dose, no molecule and no number. Every figure in this essay comes from the trials. The theory’s own ledger says the same. Where it counts closures, it holds. Where it must fix a size, it has not yet done better than chance.

Its contribution is narrower and real. It asks of every treatment one question: does the chain close at the point of need? For a bespoke vaccine, the chain runs from tumour to sequencer to factory and back, in nine weeks. For treatment before surgery, the chain closes inside the patient.

This essay is a reading of published trials through a theory. It is not treatment advice. A patient with melanoma decides with an oncologist.

---

Annotated References

1. Long, G. V. et al. (2026). INTerpath-001: intismeran autogene plus pembrolizumab versus pembrolizumab in resected high-risk melanoma. ESMO Congress 2026, LBA1. Reported by OncLive and CancerNetwork, 24–25 October 2026. OncLive report. Why read? The trial that prompted this essay, with the piecewise hazard ratios used in Section 6. Reading advice: the full paper is due in the New England Journal of Medicine on 11 November 2026; read its subgroup tables first.
2. González, F. (2026). “New mRNA Vaccine Reduces Risk of Melanoma Recurrence by Over 40 Percent.” WIRED, 24 October 2026. Why read? The occasion. Reading advice: read it for the closing remark of Soengas on the bespoke vaccine; check its number of targets against the trial.
3. Weber, J. S. et al. (2024). Individualised neoantigen therapy mRNA-4157 (V940) plus pembrolizumab versus pembrolizumab monotherapy in resected melanoma (KEYNOTE-942). The Lancet 403:632–644. Why read? The phase 2b predecessor, with the mutation-load subgroup and the delayed separation of the curves.
4. Patel, S. P. et al. (2023). Neoadjuvant-adjuvant or adjuvant-only pembrolizumab in advanced melanoma (SWOG S1801). New England Journal of Medicine 388:813–823. Why read? The same drug, given before and after surgery rather than after only: two-year event-free survival 72 against 49 percent. The cleanest test of Move 1.
5. Blank, C. U. et al. (2024). Neoadjuvant nivolumab and ipilimumab in resectable stage III melanoma (NADINA). New England Journal of Medicine. DOI 10.1056/NEJMoa2402604. Why read? A phase 3 trial sponsored by the Netherlands Cancer Institute and co-sponsored by Melanoma Institute Australia. Twelve-month event-free survival 83.7 against 57.2 percent. Reading advice: read with the ESMO 2025 update (two-year 77.3 against 55.7 percent).
6. Richtlijnendatabase (2025). Richtlijn Melanoom, module “Neoadjuvante behandeling stadium III”, 9 October 2025. Guideline page. Why read? The Dutch guideline that made treatment before surgery the preferred route, with referral to one of 14 melanoma centres.
7. Nathan, P. et al. (2021). Overall survival benefit with tebentafusp in metastatic uveal melanoma. New England Journal of Medicine 385:1196–1206. Why read? The clearest success of a shared, lineage target, reached with a T-cell receptor bispecific rather than a vaccine: one product for every eligible patient. Reading advice: note the restriction to one tissue type (HLA-A*02:01).
8. CancerNetwork (2025–2026). BNT111 phase 2 and IO102-IO103 phase 3 results. Why read? The modest and the missed results for shared-antigen vaccines. Together they keep Move 2 open.
9. Andtbacka, R. H. I. et al. (2015). Talimogene laherparepvec improves durable response rate in patients with advanced melanoma (OPTIM). Journal of Clinical Oncology 33:2780–2788. Why read? Treatment of the tumour where it lies. Reading advice: read with the five-year neoadjuvant T-VEC phase 2 data (ASCO Post, August 2023), and note the 80 percent confidence intervals.
10. Rohaan, M. W. et al. (2022). Tumor-infiltrating lymphocyte therapy or ipilimumab in advanced melanoma. New England Journal of Medicine. DOI 10.1056/NEJMoa2210233. Why read? The patient’s own reading cells, grown and returned: progression-free survival hazard ratio 0.50. Personalised too, but without a mutation sequence. A useful contrast.
11. Green, A. C. et al. (2011). Reduced melanoma after regular sunscreen use (Nambour trial). Journal of Clinical Oncology 29:257–263; online 6 December 2010. Why read? The upstream move: 11 against 22 new melanomas in ten years. Reading advice: the 73 percent drop in invasive melanoma rests on 14 cases.
12. Konstapel, J. (2026). The Vacuum.Net Theory: Definition, Theorems and Tests. Version 4, 17 October 2026. PDF. Why read? The axioms N1–N5 and the theorem on the two kinds of closure used in Section 3, with the theory’s own ledger of tests passed and failed.
13. Konstapel, J. (2026). Where Illness Comes From, Completed. October 2026. Essay. Why read? The reading of illness as a test with three outcomes, and of a tumour as the remainder of a selection. Reading advice: start with Part 0.
14. Konstapel, J. (2026). The Respondent-Judge Protocol. Version 1.0, 17 October 2026. PDF. Why read? The method of Sections 5 and 6: predictions fixed and hashed before the data, scored by a separate judge.

Article

het universum kan worden voorgesteld als een bijna oneindig lange draad die een bijna oneindig aantal lussen maakt;

Door deze draad beweegt het licht, of het licht is de draad of het is alleen beweging.

Jump to the Scientific Article ( DOI: 10.13140/RG.2.2.14979.87849 ) here

J. Konstapel,24 October 2026

© 2026 J. Konstapel, Constable Research, Leiden. All rights reserved. Quotation with
attribution is permitted; reproduction of the whole requires written permission.

Most theories begin with many ideas. The Vacuum.Net begins with one: the loop. A single strand runs through a set of places. Whenever the strand returns to a place it has already visited, a loop occurs. That return is the only event in the theory. Every other concept — path, knot, number, tension, scale — is derived from it.

The strand is not in space. The strand is the universe. There is nothing outside it. This is the first axiom. Every distinction we can make is itself a loop: to distinguish something is to come back to it a second time.

Five axioms fix the structure:

1. Strand. There is exactly one strand. Every distinction is a loop.
2. Closure. A loop persists only if it lies in a closed window. What does not close is passed on as tension.
3. Memory. At every knot the passages are ordered. The first passage fixes the direction.
4. Context. Every constant is a function of the strand.
5. Scale. Merged into blocks, the coarse strand obeys the axioms again. What does not close at one scale is the material of the next.

This essay follows the consequences. What is a path? Where do paths connect? How do we compute a connection point? And why does the whole net never close?

---

2. Direction Needs Three Values

Direction is the first consequence. A loop has a direction. A direction must differ from its own reverse. This sounds trivial. It is not.

Mathematics offers a natural test. Which is the smallest cyclic group in which some element differs from its reverse? In the group with one element, every element is its own reverse. In the group with two elements, the nonzero element equals its reverse. Only the group with three elements — the integers modulo 3 — passes the test. There, +1 and −1 are distinct.

So direction has exactly three values. The theory calls them trits: +1, 0 and −1. A passage through a knot is read relative to the first passage through that knot. It either leaves the knot the same way (+1), leaves the way the first passage entered (−1), or leaves some other way (0). The rule uses only the strand itself. It is mechanical. Nothing is interpreted; everything is computed.

---

3. Closure Is a Loop of Trits

Trits behave like steps on the three-point cycle. Walk a sequence of trits and count your position modulo 3. The walk returns to its start exactly when the sum of the trits is 0 modulo 3. A closed run of trits is therefore itself a loop — one scale up. The scale axiom at work.

The shortest nontrivial return takes three steps: +1, +1, +1. This fixes the window at three trits. A window closes when its three trits sum to 0 modulo 3.

There are 27 possible windows. Only 9 close:

• the three constant windows (000, +1+1+1, −1−1−1);
• the six orderings of (−1, 0, +1).

The other 18 windows all have two equal trits and one different trit. None of them closes. Open windows are not harmless. By the second axiom, whatever does not close is passed on as tension. Tension is the engine of the theory. It guarantees that something always happens next.

---

4. Three Elementary Closures

The first theorem is the grammar of the theory. Every closed trit word reduces completely — to nothing — by repeatedly removing one of three pieces:

• a standing: a single 0 (the path pauses);
• a meeting: x followed by −x (the path goes and comes back);
• a ring: x, x, x (the path goes round once).

No other word reduces. This was proven, and then computed for all 265,720 closed words up to length 12. Proof and enumeration agree.

Lift the steps to the integers and the picture sharpens. A closed walk with nonzero steps moves a multiple of three places. That multiple is its winding number. The meeting is the shortest closure with winding number 0. The ring is the shortest with winding number ±1. So apart from standing still, there are exactly two ways to close: to meet, in two steps, or to go round, in three steps. Everything else is a combination of these two moves.

---

5. Every Number Is a Path

The theory counts loops, so it must count. It counts in balanced ternary. Every integer has exactly one representation as a sum of digits −1, 0, +1 times powers of three. The word “balanced” is literal: the digits balance around zero.

• Words of k digits reach exactly the integers from −(3^k − 1)/2 to +(3^k − 1)/2, with no gaps and no doubles.
• Every number is a trit word; every trit word is a number.

Digit k stands at scale 3^k — the scale axiom again, now applied to arithmetic. The number of digits of an integer n is ⌊log₃(2n − 1)⌋ + 1, computable in whole numbers without any rounding. Depth in scales is therefore a logarithmic position: p = log₃ q. This position will carry the entire bridge to measurement in Section 8.

A few examples, writing T for −1: 2 = 1T, 3 = 10, 8 = 10T, −8 = T01.

---

6. The Ladder, and Why the Net Never Closes

Now merge the strand. At each scale, collect the elementary closures into blocks of equal size: a meeting merges two finest passages, a ring merges three. After a merges of meetings and b merges of rings, one coarse passage contains exactly 2^a · 3^b finest passages. This is the ladder theorem. The condition of equal block size is essential; drop it and the count fails (a ring joined to a single passage of a meeting gives 4, not 6).

The next theorem is the heart of the theory. Can the meetings and the rings ever return together? That would require 2^a = 3^b with a and b positive. This has no solution. A power of two is even; a power of three is odd. They meet only at 1, where a = b = 0.

The net never closes completely. There is always a remainder, and by the closure axiom the remainder passes on as tension. There is always a next passage. The strand has no last passage. The net is infinite — not by assumption, but by arithmetic.

Yet the net comes close to closing. It comes closest where b·log₂3 is nearest a whole number. These best approximations are the convergents of the continued fraction of log₂3. The sequence of near-closures runs at depths:

Depth b	a	Remainder
1	2	498.045 cents
2	3	203.910 cents
5	8	90.225 cents
12	19	23.460 cents
41	65	19.845 cents
53	84	3.615 cents
306	485	1.770 cents
665	1054	0.0756 cents
15601	24727	0.0315 cents

The remainder at depth 12 is the Pythagorean comma — the small excess of twelve musical fifths over seven octaves, known to music theory for two millennia. Each near-closure is finite and enormous. A cycle at depth 665 holds about 1.9 × 10³¹⁷ finest passages.

---

7. What a Path Is

With the machinery in place, the definition is simple. A path is the strand on an interval: a contiguous piece of the one strand. Its trit word is the sequence of its trits, computed by the rule of Section 2.

A path is closed in a window when that window’s trits sum to 0. It is closed everywhere when every window closes. Its open windows are its tension. Two paths can meet only where the strand meets itself — at a knot.

The counts are exact:

• A strand of N passages contains N(N + 1)/2 paths.
• Of the 3^n possible trit words of length n ≥ 3, exactly 3^(n−1) close their first window.
• Exactly nine words close everywhere. A word closes in every window if and only if its first window closes and it has period 3. There are nine closed windows, hence nine such words for every length: the constant runs and the repetitions of (−1, 0, +1) in its six orders.
• A knot visited m times holds m(m − 1)/2 loops.

The nine everywhere-closed words carry a quiet message. A path without tension is possible — but only in nine rigid forms. Every other path carries tension. Tension is not the exception. It is the rule.

---

8. The Same Path in Position Space

The theory also lives in measurement. Take a phenomenon described by n magnitudes. Each magnitude is a ratio to a reference of its own kind, hence a pure number q. Its position is p = log₃ q — the depth of Section 5, now applied to data. The phenomenon is a point in an n-dimensional position space. When it changes, the point traces a path.

Between two magnitudes qᵢ and qⱼ along such a path, the local exponent is the ratio of their position changes. The slope trit is the direction of pⱼ relative to pᵢ: rising +1, flat 0, falling −1.

A point alone is a coordinate. A point carries no content. The path carries direction and shape. The point disappears; the path remains. This is the theory’s bridge from arithmetic to physics.

---

9. Connection Points on One Path

Where does a path connect with itself? Four answers, in the net and in position space.

In the net:

• A knot — a place visited at least twice — holds m(m − 1)/2 loops.
• A closed window — three trits summing to 0 — is one of the nine forms of Section 3.
• An open window — any other three trits — carries tension.

A propagation theorem links neighbouring windows: the window sum shifts by the entering trit minus the leaving trit, sₖ₊₁ ≡ sₖ + tₖ₊₃ − tₖ (mod 3). Tension therefore does not stay put. It moves along the word.

In position space:

• A regular point: the slope keeps its direction. Nothing closes.
• A turning point: the slope changes direction. The trits read +1, 0, −1 (or reversed) — a meeting. The path goes and comes back.
• A zero: the magnitude changes sign. The position p = log₃|q| runs to −∞. Again a meeting.
• A return: the path passes a point it passed before — a knot at this level.

One theorem unifies the closed cases. On a closed path of a single magnitude, rises and falls alternate and are equal in number. The reduced slope word breaks down into meetings only, and the winding number is 0. Proven, and computed on 2,000 random periodic paths, then rechecked on 3,000. A closed path in one dimension is a chain of meetings. It cannot go round; there is nothing to go round in.

---

10. Connection Points Between Paths

Two paths are two pieces of the one strand. They meet where the strand meets itself: at a shared knot, where Section 9 applies unchanged.

A subtler connection is the seam. Join the end of one closed window A to the start of another closed window B. Two new windows straddle the joint. The seam rule counts the open ones: the number of open windows at the seam is [A₁ ≠ B₁] + [A₃ ≠ B₃]. The seam is closed if and only if B = A. A window continued in its own order produces no tension. Two different closed windows always produce tension at their seam.

Computed over all 81 ordered pairs: 9 closed seams, 36 with one open window, 36 with two. The pattern is exact and complete.

In position space, the between-path connections become laws. A power law qⱼ = C · qᵢ^e is a straight line in the plane of the positions, with slope e. Two such laws between the same magnitudes are identical, parallel, or meet in exactly one point:

Relation	Exponents	Constants	Connection
identical	equal	equal	the same law
parallel	equal	different	same exponent, different constant
intersecting	different	—	one point where both hold

And the no-closure theorem returns between two paths. Two periodic paths with frequencies f₁ and f₂ return together if and only if f₁/f₂ is rational. Twelve musical fifths never land on a whole number of octaves: 3¹² = 531,441 against 2¹⁹ = 524,288. Their ratio is the Pythagorean comma, 23.460 cents. The arithmetic of Section 6 is not an abstraction. It is audible in every piano tuned in fifths and played in octaves.

---

11. Computing a Connection Point

The method has four fixed steps. No step involves a choice.

1. Strand. Take the input: a word sequence, a measured series, or a set of magnitudes.
2. Trits. For a strand, apply the rule of Section 2. For magnitudes, use the slope trit. For an exact number, use its balanced-ternary word.
3. Windows. Mark every window closed or open.
4. Connection points. Read knots, seams, turning points, zeros and returns. Break every closure into standings, meetings and rings.

Five worked cases show the range:

Case	Input	Result
A word sequence	an 11-passage sentence	4 knots, 6 loops, one window closed, one open
Light	wavelength against frequency	slope −1; λ·f constant; a straight line
Kepler	eight planets, NASA data	measured slope 1.4978 ± 0.0007 against the law’s 3/2
π junction	a superconducting junction on cooling	a zero: a meeting
Little–Parks	resistance of a ring against flux	a closed path: meetings only

The Kepler case deserves emphasis. The exponent 3/2 of the third law is exactly the ratio of a ring (length 3) to a meeting (length 2). The law of planetary motion reads as a balance of the two elementary closures. Whether this is coincidence or consequence is the kind of question the theory is built to ask.

---

12. Verification

Every theorem in the source has a proof. Every computed result was computed twice, by two independent programs: a respondent script, and a separate judge that wrote its own code from scratch without reading the first. The judge verified, among others:

• the uniqueness of balanced ternary, by round trip on ±100,000;
• the reduction of all 265,720 closed words up to length 12;
• the seam rule over all 81 ordered pairs;
• the nine everywhere-closed words for lengths 3 to 9;
• the Kepler slope on NASA data, to five decimals;
• the continued fraction of log₂3 and the depth table.

All 72 references were checked one by one against Crossref, publisher records or the source itself.

This discipline matters. A theory that claims the universe is a loop invites suspicion. Its response is to make every claim fixed, mechanical and checkable. Claims are fixed before the data are read; every number has two independent origins.

---

13. Conclusion: The Open Net

The Vacuum.Net yields a compact picture. Direction needs three values. Closure is arithmetic mod 3, and it comes in exactly three elementary forms: the standing, the meeting and the ring. Numbers are paths in balanced ternary, and their depth is logarithmic. Nine paths close everywhere; every seam between different closures carries tension.

And the whole never closes. Two to the a can never equal three to the b. The remainder is passed on as tension, forever. The net is infinite by a parity argument — the same argument that puts 23.46 cents between twelve fifths and seven octaves.

This is the theory’s most consequential claim. The universe is not a closed structure that happens to contain process. It is a process that keeps failing, by exactly a comma, to close — and each near-closure leaves a residue that becomes the material of the next scale. The curve in the Vacuum.Net does not end. It returns, misses by a remainder, and passes the difference on.

---

Annotated References

Grouped as in the source. Each entry carries a note on why it is cited.

The theory

1. Konstapel, J. (2026). The Vacuum.Net Theory: Definition, Theorems and Tests, version 4. Constable Research, Leiden. — The founding document: the definition of the strand and the five axioms on which Sections 1 and 2 rest.
2. Konstapel, J. (2026). “The Way Before the Net.” constable.blog, 20 October 2026. — Zero as the meeting of an act with its mirror; the two roads to closure (meeting and going round) behind Section 4.
3. Konstapel, J. (2026). “The Mathematics of Closure.” constable.blog, 15 October 2026. — The convergents of log₂3 that generate the depth table of Section 6.
4. Konstapel, J. (2026). “The Possibilities and Impossibilities of the Net.” constable.blog, 19 October 2026. — What the net can do in its own quantities: read a place, hold distant structures in one state, select by address, keep a configuration in circulation.
5. Konstapel, J. (2026). The Respondent–Judge Protocol. Constable Research, Leiden. — The double-computation discipline verified in Section 12.

Numbers and the register

6. Hardy, G. H., and Wright, E. M. An Introduction to the Theory of Numbers, 6th ed. Oxford University Press, 2008. — Unique factorisation, irrationality and continued fractions; the number theory behind the no-closure theorem.
7. Knuth, D. E. The Art of Computer Programming, Vol. 2: Seminumerical Algorithms, 3rd ed. Addison-Wesley, 1998. — Balanced ternary as a practical register; Section 4.1.
8. Spencer-Brown, G. Laws of Form. Allen & Unwin, 1969. — A calculus built on one act of distinction; the philosophical forerunner of the first axiom.
9. Kauffman, L. H. (2005). “EigenForm.” Kybernetes 34(1/2), 129–150. — Forms that return on themselves: closure as a fixed point.
10. Varela, F. J. (1975). “A calculus for self-reference.” International Journal of General Systems 2, 5–24. — A third value beside true and false, close to the trit 0.

Paths, returns and knots

11. Pólya, G. (1921). “Über eine Aufgabe der Wahrscheinlichkeitsrechnung betreffend die Irrfahrt im Straßennetz.” Mathematische Annalen 84, 149–160. — A random path returns with certainty in one and two dimensions, not in three; the basis of the free-path model of the strand.
12. Watson, G. N. (1939). “Three triple integrals.” Quarterly Journal of Mathematics 10, 266–276. — The integral behind the 3D return probability 0.3405.
13. Montroll, E. W. (1956). “Random walks in multidimensional spaces.” Journal of the SIAM 4(4), 241–260. — Return probabilities in four and five dimensions.
14. Rolfsen, D. Knots and Links. Publish or Perish, 1976. — The standard theory of closed curves in space and their winding.
15. Adams, C. C. The Knot Book. W. H. Freeman, 1994. — A readable entry to knots, crossings and invariants.
16. Milnor, J. Topology from the Differentiable Viewpoint. University Press of Virginia, 1965. — Degree and winding number; background to the elementary-closures theorem.

Curves, cycles and special points

17. Strogatz, S. H. Nonlinear Dynamics and Chaos, 2nd ed. Westview Press, 2015. — Fixed points, limit cycles and quasi-periodic motion; background to the position-space sections.
18. Lissajous, J. A. (1857). “Mémoire sur l’étude optique des mouvements vibratoires.” Annales de chimie et de physique (3rd series) 51, 147–231. — Combined vibrations that close only at rational ratios; the visual form of the two-cycles theorem.
19. Thom, R. (1972). Stabilité structurelle et morphogenèse. Benjamin. — Folds and cusps: the stable kinds of sudden change in a curve.
20. Arnold, V. I. (1992). Catastrophe Theory, 3rd ed. Springer. — A short, exact account of the special points of smooth maps.
21. Feigenbaum, M. J. (1978). “Quantitative universality for a class of nonlinear transformations.” Journal of Statistical Physics 19, 25–52. — Period doubling: branching by factor two, one half of the ladder.
22. Mandelbrot, B. B. (1982). The Fractal Geometry of Nature. W. H. Freeman. — Curves alike at every scale; background to the scale axiom.
23. Verhulst, P.-F. (1838). “Notice sur la loi que la population suit dans son accroissement.” Correspondance mathématique et physique 10, 113–121. — The S-curve of bounded growth.

Scaling, laws and dimensions

24. Buckingham, E. (1914). “On physically similar systems.” Physical Review 4, 345–376. — The Π theorem: laws written in dimensionless invariants; the basis of the pure-number positions.
25. Bridgman, P. W. (1922). Dimensional Analysis. Yale University Press. — Why dimension is multiplicative.
26. Barenblatt, G. I. (1996). Scaling, Self-Similarity, and Intermediate Asymptotics. Cambridge University Press. — Power laws as straight lines in logarithmic space, and when exponents are exact.
27. Clauset, A., Shalizi, C. R., and Newman, M. E. J. (2009). “Power-law distributions in empirical data.” SIAM Review 51(4), 661–703. — How to fit an exponent honestly; required before any exponent is called exact.
28. Stanley, H. E. (1971). Introduction to Phase Transitions and Critical Phenomena. Oxford University Press. — Critical exponents as straight stretches next to a threshold.
29. Larson, D. B. (1959). The Structure of the Physical Universe. North Pacific Publishers. — Motion as the one component of the universe, with space and time as reciprocal aspects.
30. Hestenes, D. (1966). Space-Time Algebra. Gordon and Breach. — Rotation as two reflections; the rotor that returns only after two turns.
31. Grassmann, H. (1844). Die lineale Ausdehnungslehre. Wigand, Leipzig. — The extended form as the trace of a moving element: the strand as a map.

Music and the comma

32. Barbour, J. M. (1951). Tuning and Temperament: A Historical Survey. Michigan State College Press. — The Pythagorean comma and the historical cycles of 12, 41 and 53 fifths.
33. Benson, D. J. (2006). Music: A Mathematical Offering. Cambridge University Press. — The arithmetic of intervals, cents and temperaments.
34. Kepler, J. (1619). Harmonices Mundi. Linz. — The third law, published in a book on musical ratios in the heavens.

Measurements used

35. Williams, D. R. NASA, NSSDCA (updated 2025). Planetary Fact Sheet. — The distances and periods behind the Kepler case.
36. Ryazanov, V. V., et al. (2001). “Coupling of two superconductors through a ferromagnet: evidence for a π junction.” Physical Review Letters 86(11), 2427–2430. — A critical current that passes through zero and changes sign: a zero in the sense of the theory.
37. Little, W. A., and Parks, R. D. (1962). “Observation of quantum periodicity in the transition temperature of a superconducting cylinder.” Physical Review Letters 9, 9–12. — A closed curve in flux: the periodic worked case.
38. Hertlein, C., et al. (2008). “Direct measurement of critical Casimir forces.” Nature 451, 172–175. — Critical Casimir forces near a critical point, with the 3D Ising exponent taken as given.

Method

39. Tarski, A., Mostowski, A., and Robinson, R. M. (1953). Undecidable Theories. North-Holland. — What it means to translate one theory into another.
40. Lakatos, I. (1970). “Falsification and the Methodology of Scientific Research Programmes.” In Criticism and the Growth of Knowledge, Cambridge University Press, 91–196. — Why claims must be fixed before the data are read.

Topology of closed curves

41. Gauss, C. F. (1833). Note of 22 January 1833 on the linking of two closed curves. Werke, Vol. 5, p. 605. — The linking number: the first invariant of two curves.
42. Alexander, J. W. (1928). “Topological invariants of knots and links.” Transactions of the AMS 30(2), 275–306. — The first polynomial that distinguishes knots.
43. Jones, V. F. R. (1985). “A polynomial invariant for knots via von Neumann algebras.” Bulletin of the AMS 12(1), 103–111. — A sharper knot invariant, later linked to physics.
44. White, J. H. (1969). “Self-linking and the Gauss integral in higher dimensions.” American Journal of Mathematics 91(3), 693–728. — Linking = twist + writhe; the split of a closed ribbon’s winding, used for DNA.

Cycles, rotation and locking

45. Poincaré, H. (1885). “Sur les courbes définies par les équations différentielles (3e partie).” Journal de Mathématiques Pures et Appliquées (4th series) 1, 167–244. — The rotation number; a curve on a torus closes iff that number is rational. The exact form of the two-cycles theorem.
46. Kronecker, L. (1884). “Näherungsweise ganzzahlige Auflösung linearer Gleichungen.” Sitzungsberichte der Akademie der Wissenschaften zu Berlin, 1179–1193 and 1271–1299. — At an irrational ratio the curve never closes but comes arbitrarily close to every point.
47. Weyl, H. (1916). “Über die Gleichverteilung von Zahlen mod. Eins.” Mathematische Annalen 77(3), 313–352. — Multiples of an irrational spread evenly; background to the remainder table.
48. Arnold, V. I. (1961). “Small denominators. I.” Izvestiya Akad. Nauk SSSR 25(1), 21–86. — Arnold tongues: coupled cycles lock onto a rational ratio over a finite range; closure becomes stable, not exceptional.
49. Kuramoto, Y. (1975). “Self-entrainment of a population of coupled non-linear oscillators.” Lecture Notes in Physics 39, Springer, 420–422. — Many cycles fall into one common rhythm above a coupling threshold.
50. Bennett, M., Schatz, M. F., Rockwood, H., and Wiesenfeld, K. (2002). “Huygens’s clocks.” Proceedings of the Royal Society A 458, 563–579. — Huygens’ 1665 observation repeated: two clocks on one beam lock in anti-phase; a measured two-cycle with coupling.
51. Noether, E. (1918). “Invariante Variationsprobleme.” Nachrichten der Gesellschaft der Wissenschaften zu Göttingen, 235–257. — Every continuous symmetry gives a conserved quantity; the physical source of the constants in the two-laws theorem.

Physics: closed paths and whole numbers

52. Thomson, W. (Lord Kelvin) (1869). “On vortex atoms.” Proceedings of the Royal Society of Edinburgh 6, 94–105. — Atoms as knotted closed curves in a medium; the historical forerunner of the knot in the net.
53. Aharonov, Y., and Bohm, D. (1959). “Significance of electromagnetic potentials in the quantum theory.” Physical Review 115(3), 485–491. — A particle on a closed path picks up a phase from a field it never touches; only the closed path counts.
54. Deaver, B. S., and Fairbank, W. M. (1961). “Experimental evidence for quantized flux in superconducting cylinders.” Physical Review Letters 7(2), 43–46. — Flux in a superconducting ring comes in whole units of h/2e.
55. Doll, R., and Näbauer, M. (1961). “Experimental proof of magnetic flux quantization in a superconducting ring.” Physical Review Letters 7(2), 51–52. — The same result, measured independently in the same month.
56. Josephson, B. D. (1962). “Possible new effects in superconductive tunnelling.” Physics Letters 1(7), 251–253. — Current across a junction depends on phase difference; the basis of every junction case.
57. Bulaevskii, L. N., Kuzii, V. V., and Sobyanin, A. A. (1977). JETP Letters 25, 290. — The theoretical prediction of the π junction, confirmed 24 years later.
58. Berry, M. V. (1984). “Quantal phase factors accompanying adiabatic changes.” Proceedings of the Royal Society A 392, 57–72. — A system led round a closed loop returns with a phase set by the loop’s geometry.
59. von Klitzing, K., Dorda, G., and Pepper, M. (1980). “New method for high-accuracy determination of the fine-structure constant based on quantized Hall resistance.” Physical Review Letters 45(6), 494–497. — A resistance that takes exact values h/(ie²): a measured whole-number closure.
60. Thouless, D. J., et al. (1982). “Quantized Hall conductance in a two-dimensional periodic potential.” Physical Review Letters 49(6), 405–408. — The integer of the Hall effect is a winding number; topology explains the exact value.
61. Witten, E. (1989). “Quantum field theory and the Jones polynomial.” Communications in Mathematical Physics 121(3), 351–399. — Knot invariants from a quantum field theory; knots and physics in one formalism.
62. Faddeev, L., and Niemi, A. J. (1997). “Stable knot-like structures in classical field theory.” Nature 387, 58–61. — A field theory whose stable particles are knots.
63. Kleckner, D., and Irvine, W. T. M. (2013). “Creation and dynamics of knotted vortices.” Nature Physics 9, 253–258. — Knotted vortices made in water and filmed; Kelvin’s idea tested.
64. Rowlands, P. (2007). Zero to Infinity: The Foundations of Physics. World Scientific. — Physics built from a total of zero; the nilpotent condition as a closure condition.

Chemistry: chirality, order and molecular knots

65. Pasteur, L. (1848). “Mémoire sur la relation qui peut exister entre la forme cristalline et la composition chimique.” Comptes rendus 26, 535–538. — Left- and right-handed crystals of one compound; the chemical counterpart of a curve’s direction.
66. Mendelejeff, D. (1869). Zeitschrift für Chemie 12, 405–406. — The periodic system: properties returning in periods of 2, 8, 8, 18, 18, 32, 32 — a closure pattern in chemistry.
67. Frisch, H. L., and Wasserman, E. (1961). “Chemical topology.” Journal of the American Chemical Society 83(18), 3789–3795. — The programme for molecules that are knots and links.
68. Dietrich-Buchecker, C. O., Sauvage, J.-P., and Kintzinger, J.-P. (1983). “Une nouvelle famille de molécules : les métallo-caténanes.” Tetrahedron Letters 24(46), 5095–5098. — The first efficient synthesis of two interlocked rings: linking number 1.
69. Dietrich-Buchecker, C. O., and Sauvage, J.-P. (1989). “A synthetic molecular trefoil knot.” Angewandte Chemie Int. Ed. 28(2), 189–192. — The first molecule tied as a trefoil knot.
70. Anelli, P. L., Spencer, N., and Stoddart, J. F. (1991). “A molecular shuttle.” Journal of the American Chemical Society 113(13), 5131–5133. — A ring moving between two stations on a thread; a mechanical bond used as a switch.
71. Koumura, N., et al. (1999). “Light-driven monodirectional molecular rotor.” Nature 401, 152–155. — A molecule that turns one way only; chirality sets the direction of a closed cycle.
72. Taylor, W. R. (2000). “A deeply knotted protein structure and how it might fold.” Nature 406, 916–919. — A natural protein whose chain forms a knot: knotted curves in living matter.

---

All 72 entries were checked one by one against Crossref, publisher records or the source itself by a separate judge. Source: J. Konstapel, “Curves in the Vacuum.Net”, v1.0, Constable Research, Leiden, 24 October 2026. Quotation with attribution is permitted.

Essay

I zag the maze en vond een uitweg.

Jump to the Scientific Article here

J.Konstapel,Leiden,23-11-2026.

Aanleiding: Vidi-beurs voor elf onderzoekers van de Universiteit Leiden

De Leidse Vidi-projecten van 2026, gelezen door de bril van de Vacuum.Net-theorie

Op 22 oktober 2026 ontvingen elf onderzoekers van de Universiteit Leiden een Vidi-beurs van de Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO). Een Vidi is geen prijs voor afgerond werk; het is een investering van vijf jaar in een persoon, bedoeld om een bewezen onderzoeker een eigen groep te laten opbouwen rond een vraag die onmiskenbaar de zijne of hare is. De elf Leidse laureaten van 2026 werken aan de oppervlakte in totaal verschillende werelden: kwantummaterialen, interstellair ijs, bacteriële verdedigingssystemen, Swahili-poëzie, middeleeuwse pestbegraafplaatsen, Bronstijd-oases, enzymchemie, auto-immuunziekte, chronische ontsteking en de psychologie van sociale verandering.

Maar lees hun projectomschrijvingen naast elkaar, en er verschijnt een opvallend grammaticaal kenmerk. Acht van de elf zeggen, elk in eigen woorden, dat er iets “niet begrepen” is, “onvoldoende begrepen” of “onduidelijk”. De overige drie zijn reconstructief: hun hiaat is een hiaat in data — hoe verhalen tussen eilanden reisden, hoe de pest door Nederlandse steden trok, hoe samenlevingen vormden in onvoorspelbare woestijnen.

De bron die dit essay bewerkt stelt een eenvoudige en vermetele vraag: hoe zien deze elf open problemen eruit als je ze door één theorie leest — de Vacuum.Net-theorie — waarvan de fundamenten vorm, orde en toestand zijn? Het antwoord is verrassend elegant. De elf problemen zijn niet elf verschillende gaten. Ze vallen in precies drie soorten uiteen, en die drie soorten zijn de eigenschappen van een geheel die een wetenschappelijke methode, getraind om oorzaken in onderdelen te zoeken, noodzakelijkerwijs als restant achterlaat.

De inzet is methodologisch, niet slechts taxonomisch. Als de analyse klopt, bepaalt de vorm van een vraag de vorm van de blinde vlek van het antwoord. Een vraag van het type “welk onderdeel veroorzaakt dit?” zal, per constructie, nooit sluiting, geheugen of context zien — want dat zijn geen eigenschappen van onderdelen. Dit essay gaat daarmee over het residu dat de standaard wetenschappelijke vraag achterlaat.

2. De theorie: een visnet in water

De Vacuum.Net-theorie vertrekt van een bewust concrete beeldspraak: een visnet in water. Eén draad loopt overal en kruist zichzelf. De draad staat onder spanning. Waar hij om zichzelf wintnt, ontstaan knopen; tussen de draden liggen mazen. Vier woorden dragen de theorie: draad (de ene lijn waarvan het net gemaakt is), winding (een deel van de draad dat in fase op zichzelf terugkeert), knoop (een plaats waar de draad meer dan eens passeert) en maas (een opening tussen draden, met een uitgestrektheid in ruimte of tijd).

Bij elke kruising heeft de draad een van drie toestanden: +1, 0 of −1. Deze eenheid is een trit, en de drie toestanden zijn nadrukkelijk niet “ja, nee, misschien”. Het zijn twee richtingen en het punt waar de richtingen elkaar ontmoeten. Drie kruisingen die elkaar langs de draad opvolgen vormen een raam, en een raam sluit wanneer de som van zijn drie tritsen 0 is, geteld in drieën — zodat +3 en −3 ook als 0 tellen. De selectieregel die de hele theorie aandrijft is één zin: wat sluit, blijft; wat niet sluit, geeft spanning door.

2.1 De rekenkunst van het sluiten

De rekenkunst verdient herhaling, want alles later hangt ervan af. Drie toestanden op drie plaatsen geven 3 × 3 × 3 = 27 ramen. Precies 9 daarvan sluiten: 3 ramen van drie gelijke waarden (sommen +3, 0, −3, alle als 0 geteld in drieën) en 6 ramen van drie verschillende waarden (elke volgorde van {−1, 0, +1}, elk sommerend naar 0). De overige 18 zijn open, en open ramen hebben altijd precies twee gelijke waarden en één afwijkende. Een kort bewijs laat zien dat een raam alleen sluit wanneer zijn tritsen óf allemaal gelijk óf allemaal verschillend zijn — gemengde ramen sluiten nooit. Dit is het eerste structurele feit van de theorie: sluiting komt in twee soorten, gelijkheid en volledig verschil.

2.2 Ringen en naden

De zes ramen-met-alles-verschillend, gelezen als ringen (een ring heeft geen beginpunt), reduceren door rotatie tot twee: de linker ring (−1 → 0 → +1) en de rechter ring (+1 → 0 → −1). Ze bevatten dezelfde drie waarden en verschillen alleen van richting. Zet een linker ring tegen een rechter ring en er gebeurt iets opmerkelijks: op de plaats waar de twee gesloten structuren van tegengestelde richting elkaar ontmoeten, verschijnen twee open ramen. De theorie noemt deze plaats een naad — en naden zijn altijd open, hun open ramen altijd in paren.

2.3 Halfsluiting en de rotor

Uit de algebra van rotaties volgt een tweede resultaat. Een rotatie kan als twee spiegelingen geschreven worden, en de rotor die haar uitvoert draagt de halve hoek. Na één volledige omwenteling van 2π staat de rotor op −1; pas na 4π keert hij terug naar +1. Een half-gesloten toestand ziet er van buiten dus compleet uit, maar draagt van binnen een minteken. Dit resultaat zal in de supergeleiders van Lahabi zijn meest letterlijke vorm vinden.

2.4 Sluiten onder drift

Uit een waarschijnlijkheidsanalyse van het trit-register volgt het derde resultaat. Komt één toestand voor met kans m en delen de andere twee de rest, dan stijgt de kans dat een raam sluit boven een derde met de derde macht van de onbalans. Concreet: van m = 0,33 naar m = 0,50 stijgt de sluitingskans met slechts 0,011; van 0,60 naar 0,80 stijgt ze met 0,186. Sluiting onder drift is eerst stil en later steil — een feit dat opvallend zal nagalmen door de medische en sociale projecten hieronder.

2.5 De vijf axioma’s

De theorie is samengevat in vijf axioma’s:

• N1 — Draad. Er is precies één draad; elk onderscheid is een kruising van de draad met zichzelf.
• N2 — Sluiting. Een kruising blijft bestaan als en slechts als zij in een gesloten raam ligt; wat niet sluit, geeft spanning door.
• N3 — Geheugen. Bij elke knoop zijn de passages geordend; de eerste passage bepaalt de oriëntatie, latere passages worden daaraan relatief geteld.
• N4 — Context. Elke “constante” is een functie van de draad; een vaste waarde is een eigenschap van de lokale toestand.
• N5 — Schaal. Grofkorrelige blokken van het net volgen dezelfde axioma’s opnieuw; wat op één schaal niet sluit, is het materiaal van de volgende.

Cruciaal is wat de theorie níet biedt: mechanismen waarin het ene onderdeel het andere veroorzaakt — in een net van één draad zijn er geen twee afzonderlijke dingen om een pijl tussen te zetten. Ze biedt in plaats daarvan vier andere vormen van uitspraak: selectie door passing, identiteit over schalen, constitutieve relatie (twee grootheden die samen vastliggen, zonder pijl) en correspondentie (dezelfde telling in twee domeinen). Elke lezing van de elf projecten heeft één van deze vormen, en elke uitspraak draagt een expliciete status: bron, geteld, lezing of open. Die eerlijkheid over epistemische status is een van de stille deugden van de bron: een lezing is een koppeling, geen bewijs.

3. Elf lezingen

3.1 Marta Artola — Twee richtingen door drie vormen

Het onderzoek. Artola ontwerpt moleculen die glycosidases blokkeren — de enzymen die suikerkettingen knippen, van centraal belang bij de ziekten van Gaucher en Fabry, kanker en bacteriële infecties. Een suikerring is niet plat: enzymen voeren hem door een vaste reeks vormen, een conformationele route. Voor een retainend α-glucosidase verloopt de route ⁴C₁ → ⁴H₃ → ¹S₃; voor het β-enzym loopt zij precies andersom. De overgangsvorm ⁴H₃ wordt gedeeld. Haar kernresultaat: een molecule vastgezet in één vorm (een cyclophellitol-cyclosulfaat in de ⁴C₁-stoel) blokkeert α-glucosidases permanent terwijl het β-glucosidases nauwelijks raakt — en zelfs een niet-reactief carba-cyclophellitol bindt sterk zolang het de overgangsvorm vasthoudt.

De lezing. De twee routes zijn de linker- en de rechterring: (−1 → 0 → +1) en (+1 → 0 → −1). Beide sluiten; beide bevatten dezelfde drie vormen; ze verschillen alleen van richting, en de overgangstoestand is de 0 — het ontmoetingspunt van de twee richtingen. Een retainend enzym loopt één ring heen en terug (vandaar retentie); een inverterend enzym loopt één route één keer. Het veld behandelt een vastgezette vorm als een sleutel; het net voegt een tweede eigenschap toe, richting, wat leidt tot een concrete ontwerpregel voor selectieve antibiotica: zet de eindvorm vast die bij de bacteriële richting hoort.

Het inzicht. Selectiviteit is niet alleen een kwestie van vorm, maar van vorm mét richting. Het telbare voorspelling: verifieer, door de hele CAZy-database, dat elke retainende familie één ring heen-en-weer loopt en elke inverterende familie één route één keer.

3.2 Rachel Knevel — De stille jaren

Het onderzoek. Knevels machine-learning-pipelines herkennen reumatoïde artritis-patiënten in vrije-tekst-medische dossiers met hoge precisie. Nederlandse huisartsgegevens tonen dat toekomstige artritispatiënten vanaf vier tot zes jaar vóór de diagnose vaker consulteren voor gewrichts- en spierklachten — maar het signaal wordt pas statistisch significant op ongeveer 1,5 jaar vóór de diagnose. Het open probleem: de ziekte eerder zien, het risico bij élke zorgcontact bijwerken.

De lezing. De lezing is dubbel. Elk zorgcontact is een passage door een knoop (de patiënt), en een model dat bij elk contact bijwerkt is axioma N3 — geheugen — als rekenregel geschreven. Maar de diepere match is de vorm van het prodroom. Drijft de lichaamstoestand langzaam naar één kant, dan stijgt de sluitingskans met de derde macht van de onbalans. Onder een lineaire-drift-aanname is op 1,5 jaar vóór de diagnose slechts circa 42% van het uiteindelijke overschotsignaal aanwezig — ongeveer 58% komt in het laatste kwart van de aanloop. Het gepubliceerde patroon — vroege scheiding, late significantie — heeft precies deze kubische vorm.

Het inzicht. Een model dat losse contacten weegt, zal de ziekte altijd laat vinden; alleen een model dat de volgorde van contacten weegt, vindt haar in de stille jaren — en dat is precies waar preventie moet ingrijpen. Het telbare voorspelling: pas de gepubliceerde odds-ratio’s over het venster van 6 tot 72 maanden aan op machten van verstreken tijd; de voorspelling is een exponent nabij 3.

3.3 Kaveh Lahabi — Windingen, naden en een net van juncties

Het onderzoek. Lahabi brengt in beeld waar superstroom loopt. Zijn resultaten bestrijken triplet-stromen in ferromagneten, supergeleidend geheugen, half-kwantumfluxoïde-toestanden in Sr₂RuO₄-microringen die op wanden tussen domeinen van tegengestelde chiraliteit liggen, en een zelforganiserend netwerk van Josephson-juncties in de kagome-supergeleider CsV₃Sb₃. Het Vidi-probleem: niemand begrijpt hoe de collectieve elektronengolf ontstaat.

De lezing. Hier wordt de theorie bijna letterlijk. Een vortex die één fluxkwantum draagt (Φ₀ = h/2e ≈ 2,068 × 10⁻¹⁵ Wb) is één gesloten omwenteling van fase — de winding van het net in zijn meest fysieke vorm. De half-kwantumfluxoïde is een halve sluiting: compleet van buiten, minteken van binnen — de rotor bij 2π. De spontane juncties op wanden van tegengestelde chiraliteit zijn de naad van de theorie: waar twee gesloten structuren van tegengestelde richting ontmoeten, verschijnen open ramen. De kritische stroom van een π-junctie die bij afkoelen door +1, 0, −1 gaat, is het register zelf — waarbij de 0 geen plateau is maar een ontmoeting van richtingen. En het kagome-netwerk — kanalen van superstroom die bij de kritieke temperatuur verschijnen en blijven bestaan — is een maas.

Het inzicht. De omkering die de theorie uitvoert is scherp: de golf wordt niet uit elektronen opgebouwd; de elektronen zijn kruisingen van één draad, en de microscoop zal geen “ontstaan” van een golf laten zien maar ramen die sluiten. De telbare voorspellingen: de nul in π-junctiecurves moet een punt zijn, geen interval; en elke junctie op de SQUID-beelden moet op een grens tussen domeinen van tegengestelde oriëntatie liggen.

3.4 Thanja Lamberts — Chemie zonder warmte

Het onderzoek. In interstellair ijs bij 10 K is er vrijwel geen thermische energie, toch verlopen reacties — via kwantumtunneling. Lamberts’ modellen, die kosmische-straling-gedreven stralingschemie en niet-diffuse buurreacties toevoegen, produceren veel meer S₈ (een ring van acht zwavelatomen) en minder H₂S — wat S₈ tot kandidaat maakt voor het ontbrekende zwavelreservoir in dichte wolken. Het Vidi-probleem: hoe kosmische straling chemie in ijs aandrijft is onvoldoende begrepen.

De lezing. Het net leest de kou als het raam in zijn zuiverste vorm: zonder warmte reist niets, en een radicaal ontmoet alleen zijn buren — drie kruisingen zonder menging van elders. Straling is doorgang, van buiten afgeleverd aan een medium dat er zelf geen heeft. De meest prikkelende correspondentie is rekenkundig: de best bekende stabiele zwavelringen zijn S₆, S₈ en S₁₂ — grootten 6 = 2·3, 8 = 2³, 12 = 2²·3, allemaal van de vorm 2ᵃ·3ᵇ. De instabiele S₇ heeft die vorm niet. Tussen 6 en 20 zijn de grootten van deze vorm: 6, 8, 9, 12, 16, 18.

Het inzicht. De vraag “waarom S₈?” krijgt een formeel antwoord: sluiting heeft voorkeursgrootten. De telbare voorspellingen: of blijvende ringgrootten in simulaties en JWST-ijspectra overeenkomen met precies die verzameling; en of opbrengsten van stralingsproducten stijgen naarmate ijslagen dunner worden (de doorgangsregel).

3.5 Luuk Loeff — Verdediging die sluiting leest

Het onderzoek. Loeff bestudeert hoe bacteriën zich tegen fagen verdedigen. Cas3 haalt DNA binnen in bursts van drie basenparen, elke burst drie enkele-nucleotide-stappen. DdmDE herkent een D-loop-structuur, geen sequentie. En Shedu is het zuiverste geval van allemaal: zijn eiwit SduA vormt een klem die open uiteinden van dubbelstrengs DNA vastgrijpt en knipt op vaste afstand van het 5′-uiteinde — lineair DNA kan zich niet vermenigvuldigen, en fagen ontsnappen door de recombinatie-afhankelijke replicatie die lineaire tussenprodukten oplevert uit te schakelen.

De lezing. De lezing is de selectieregel van de theorie, uitgevoerd door een eiwit: wat sluit, blijft; wat niet sluit, wordt geknipt. Het bacteriechromosoom is een gesloten ring — een winding. Faag-DNA dat zich via een lineaire vorm vermenigvuldigt heeft open uiteinden. Zelf en niet-zelf worden gescheiden door sluiting alleen, zonder één letter van sequentie te lezen. De ontsnapping bevestigt het: tegen een sluitingslezer is de enige mogelijkheid om te ontsnappen slíten — circulariseren, afdekken of lineaire tussenprodukten vermijden — nooit de sequentie muteren.

Het inzicht. Selectie op vorm, niet op inhoud, hier in zijn meest pure biologische gedaante. Het telbare voorspelling: classificeer escape-mutanten tegen Shedu, DdmDE en verwante systemen; de voorspelling is dat ontsnappingen die open uiteinden verminderen domineren boven ontsnappingen die de sequentie veranderen met behoud van structuur.

3.6 Annachiara Raia — Een verhaal dat blijft

Het onderzoek. Raia’s promotieonderzoek bestudeerde het Swahili-gedicht Utendi wa Yusufu, een hervertelling van het Jozefverhaal die in motief en stijl verschilt van zijn Genesis-, Koran- en Arabische bronnen. Het Vidi-project vraagt hoe verhalen reizen, hoe plaatsen ze hervormen, en hoe ze fragmentatie overleven in eilandgemeenschappen — Lamu, Madagaskar, Java.

De lezing. De utendi-strofe biedt een kleine, volmaakte rekenkunst: vier regels van acht lettergrepen geven 32 = 2⁵ lettergrepen; de eerste drie regels rijmen op elkaar (gelijkheid — sluiting van de eerste soort), terwijl de rijmklank van de vierde regel in élke strofe terugkeert en als een draad door het hele gedicht loopt. Elke strofe sluit lokaal; het gedicht sluit als geheel. En het verhaal zelf sluit op terugkeer: Jozef gaat verloren, daalt af naar Egypte, rijst op en wordt herenigd — de narratie eindigt waar zij begon. Een verhaal dat op zichzelf terugkeert is een winding. Het veld verklaart persistentie met transmissienetwerken — handel, godsdienst, drukpers, performance — maar netwerken verklaren hoe een verhaal wordt vervoerd, niet welke verhalen het vervoer overleven. Het net voegt een selectieregel toe: verhalen die sluiten, blijven bestaan.

Het inzicht. Vormen die op elk niveau sluiten — de utendi — houden een verhaal samen zelfs als de overdrachtsketen breekt. Het telbare voorspelling: markeer elk verhaal in het gedeelde corpus op zijn einde — terugkeer of geen terugkeer — en meet zijn reikwijdte over eilanden en eeuwen heen.

3.7 Rachel Schats — Wie de pest nam

Het onderzoek. Schats leest ziekte en leefomstandigheden af aan botten. Bij vergelijking van middeleeuwse Alkmaarse stedelingen met dorpelingen vond zij tuberculose-laesies bij 2 van 189 stedelijke skeletten en bij geen van 173 rurale — wat, gegeven dat slechts zo’n 10% van tuberculosepatiënten botlaesies ontwikkelt, suggereert dat circa 10,6% van de stadsbevolking de ziekte droeg. Historisch werk toont de Nederlandse pest terugkerend tot laat in de jaren 1660, dichte steden, handel en oorlog bevoordelend, jarenlang hangend in armere wijken, met lokale uitbraken die ongeveer de helft van de bevolking doodden. De Vidi-vraag: wie stierf en waarom — hoe ziekte, leefomstandigheden en ongelijkheid samenhangen.

De lezing. Het net leest een stad als een dichte maas: volgens de doorgangsregel van de theorie schaalt doorgang per tijdseenheid met één over de open grootte, dus een stad van korte afstanden en constant contact voert meer doorgang — meer infectie — dan een dorp. Kwetsbaarheid is een toestand, geen eigenschap (N4): mortaliteit is geen vaste eigenschap van een ziekte maar van de lokale toestand van de maas, en een arme buurt is een andere toestand dan een rijke. Vroege levensstress, vastgelegd in tanden en schedelbeenderen, oriënteert latere passages (N3): wie door vroege stress gemarkeerd is, ontmoet de pest anders. En de “echo” van de pest die in arme wijken blijft hangen, is spanning die niet sluit en in doorgang gehouden wordt.

Het inzicht. De vergelijking tussen pesthospitaal en gewone begraafplaats is geen vergelijking van twee groepen individuen maar van twee toestanden van hetzelfde stedelijke net. De telbare voorspellingen: meer vroege-stressmarkers in de pesthospitaalpopulatie; langere uitbraakpersistentie in dichtere, armere wijken.

3.8 Jennifer Swerida — Oase en gemeenschap als één

Het onderzoek. In Bat in Oman toont Swerida’s archeologie dat de Bronstijd-“oase” een mensen gemaakt ruimte was die samen ontwikkelde met de gemeenschappen die haar gebruikten; dat bewoning drie ecologische zones binnen 10 km besloeg; en dat aardewerk en torens uitwisselingsnetwerken over het Hajar-gebergte vormen. Landslakkenhuizen dienen als gedateerd archief van natte en droge perioden. De Vidi-vraag: hoe vormde sociale complexiteit zich onder klimatologische onzekerheid?

De lezing. Swerida’s eigen framing breekt de standaardpijl al — de oase kwam niet eerst met mensen die zich aanpasten. Het net noemt dit een constitutieve relatie: twee grootheden die samen vastliggen, zonder pijl ertussen. Onzekerheid is spanning — een onvoorspelbaar klimaat is een draad die niet uit zichzelf sluit — en een samenleving vormt waar die spanning lokaal sluit, waar water, nederzetting en monument elkaar vasthouden. Het slakkenhuis, groeiend in geordende lagen die elk het water van hun tijd vastleggen, is N3 in calciumcarbonaat. En als register gelezen — nat (+1), droog (−1), overgangen (0) — levert de klimaatreeks een scherpe voorspelling: bouwfases in Bat moeten clusteren bij de overgangen tussen regimes, niet bij de natte maxima.

Het inzicht. De controverse “klimaat bepaalt samenleving of samenleving past zich aan” lost op in een constitutieve relatie zonder oorzaak. Het telbare voorspelling: bouwfases clusteren bij nat-droog-overgangen; de telling heeft alleen de twee gedateerde reeksen nodig die het project zelf al produceert.

3.9 Anne Urai — Geheugen dat lezen kleurt

Het onderzoek. Urai’s kernbevinding, gerepliceerd over zes datasets en meerdere diersoorten: eerdere keuzes kleuren latere niet door het startpunt van bewijsaccumulatie te verschuiven, maar door de snelheid ervan te veranderen — de vorige keuze kleurt hoe nieuw bewijs wordt gelezen. Verwant werk toont confirmatie-bias in bewijsweging, aanhoudende pariëtale activiteit die de keuze voortdraagt, en — bij muizen — precies drie discrete verborgen strategieën: volg de stimulus, neig links, neig rechts. Haar Vidi-project vraagt of de verwachting van een sociaal omslagpunt mensen eerder op trends laat handelen — het omslagpunt zelf ontketendend.

De lezing. De correspondenties zijn bijna ongemakkelijk netjes. Keuzegeschiedenis is N3: de eerste passage bepaalt de oriëntatie; latere passages worden daaraan relatief geteld. De drie verborgen strategieën zijn het register: 0, −1, +1, met “engaged” als de 0 waar geen richting domineert. De waarnemer is een knoop in de draad — een verwachting is geen informatie van buiten maar een passage van binnen, dus het “self-fulfilling” karakter van sociaal kantelen is geen paradox maar sluiting door de waarnemer heen. En de vorm van een tipping point — toegewijde minderheden die normen omverwerpen zodra een kritieke grootte bereikt is — is sluiting onder drift: stil, dan steil.

Het inzicht. Wie het omslagpunt verwacht, ís een passage in het systeem — de klassieke vraag “waarom is de voorspelling zichzelf waar?” verdient een formeel antwoord. De telbare voorspellingen: drie verborgen strategieën ook in menselijke keuzedata; en een kubische aanloop van normacceptatie in laboratoriumspellen.

3.10 Maartje Huijbers — Eén doelwit, twee tekens

Het onderzoek. Bij MuSK-myasthenia gravis vallen IgG4-antilichamen het eiwit MuSK aan op de overgang tussen zenuw en spier. Huijbers’ resultaten vormen één argument: IgG4, niet IgG1–3, draagt de ziekte; de antilichamen breken het MuSK–LRP4-partnerschap; en — cruciaal — omdat IgG4 de helft van zichzelf uitwisselt met andere IgG4-moleculen, binden patiëntenantilichamen MuSK meestal met één arm. Monovalente antilichamen blokkeren signalering; bivalente antilichamen met dezelfde specificiteit activeren MuSK juist. Eén klonaal antilichaam, bivalent gemaakt, veroorzaakte bij hoge doses alsnog traag zwakte; een ander veroorzaakte niets. En klinische ernst wordt niet verklaard door de antistofhoeveelheid.

De lezing. Het net leest valentie als het register: geen antilichaam is 0; één gebonden arm bezet zonder te sluiten (teken −1, blokkeren); twee gebonden armen bruggen twee doelwitten en sluiten (teken +1, activeren). Hetzelfde molecule op dezelfde plek geeft tegengestelde tekens al naargelang het sluit. Volgens N4 is het effect van een binding een eigenschap van de lokale configuratie, niet van het antilichaam; volgens N2 telt wat sluit — dus verliest de titer, die blokkerende en bruggende antilichamen bij elkaar optelt, het teken.

Het inzicht. Voor de Vidi-vraag is dit een directe lead: het aandeel eenarmige (uitgewisselde) MuSK-IgG4, en hoeveel brugging overblijft, zou ernst en therapierespons beter moeten volgen dan de totale titer. Een dozijn andere auto-immuunziekten worden door IgG4 gedomineerd; de lezing kan reizen. Het telbare voorspelling: het eenarmig aandeel voorspelt ernst, de titer niet.

3.11 Vincent van Unen — Ontsteking die zichzelf vasthoudt

Het onderzoek. Van Unen bracht massacytometrie naar Nederland en beeldde het darm-immuunsysteem in ziekte af. Bij behandeling-naïeve inflammatoire darmziekten scheidde een 36-antilichaam-paneel 44 immuunsubsets — en de ontsteking volgde een netwerk van subsets (geactiveerde geheugen-CD4-T-cellen, regulatoire T-cellen, PD-1+ geheugen-CD8-T-cellen, neutrofielen, γδ-T-cellen, NK-cellen), aanwezig bij sommige patiënten los van subtype, ernst of locatie. Beeldvorming toont neutrofielen, geheugen-T-cellen en myeloïde cellen zij aan zij in de ontstoken darm. Geheugen-CD4-T-cellen vormen vóór de geboorte. Zijn organoïdmodellen houden epitheel, stroma en immuuncellen samen.

De lezing. De lezing van het net keert de klinische vraag om. Chronische ontsteking is geen open proces dat niet wil eindigen; het is een configuratie die gesloten is en zichzelf vasthoudt. De drie celsoorten zitten samen in verschillende rollen — een raam van drie verschillende waarden, de tweede soort sluiting, stabiel als geheel. Een therapie die één lid weghaalt, neemt één waarde uit het raam: het raam opent, spanning geeft door — maar de configuratie wordt herinnerd (N3), en een vervanger sluit het opnieuw. Het netwerk keert terug. Dit verklaart waarom populatiereductie zo vaak faalt: de therapie laat de sluiting zelf ongemoeid.

Het inzicht. De klinische vraag verschuift van “welke cel weg te halen” naar “hoe de samenkomst te breken”. De organoïden zijn precies daarom de juiste proefopstelling: het kleinste raam van drie rollen waarin sluiting kan vormen — en gebroken worden. Het telbare voorspelling: triades van neutrofiel, geheugen-CD4-T-cel en myeloïde cel in beeldmassacytometrie, responders tegenover niet-responders; niet-responders houden hun triades zelfs als één populatie is afgenomen.

4. Drie gaten

Zet de elf open problemen naast elkaar en de structuur verschijnt. Elk project zoekt zijn antwoord in een onderdeel — een deeltje, een enzym, een eiwit, een celtype, een symptoom, een antilichaamtiter, een persoon, een nederzetting, een tekst. Elke keer blijft een eigenschap van het geheel over. Dat restant is altijd van drie soorten:

Gat	Axioma	Projecten	Wat overblijft
Vorm	N2 — sluiting	Lahabi, Loeff, Artola, Huijbers, Van Unen, Raia	wat sluit en wat open blijft
Orde	N3 — geheugen	Knevel, Urai, Schats	hoe eerdere passages latere oriënteren
Toestand	N4 — context	Lamberts, Swerida, Schats, Huijbers	hoe een waarde van zijn omgeving afhangt

Schats en Huijbers verschijnen tweemaal; hun problemen hebben twee kanten. De drie gaten zijn geen zwakheden van de elf onderzoekers — zij volgen uit de vorm van de vraag. Een vraag van het type “welk onderdeel veroorzaakt dit?” laat de eigenschappen van het geheel per constructie als restant achter.

Drie bevindingen keren over de elf heen terug, op manieren die geen enkel vak afzonderlijk kan zien:

1. Selectie op vorm, niet op inhoud. Shedu knipt open DNA-uiteinden zonder de letters te lezen. Glycosidaseremmers selecteren op de vorm die ze vastzetten. MuSK-antilichamen wisselen van teken met het aantal armen dat bindt. Superstroom loopt in randen en kanalen. Steeds is de selectie op sluiting.
2. Drie als eenheid. Cas3 haalt DNA binnen in bursts van drie stappen. Elke enzymeroute doorloopt drie vormen. Het register verschijnt als activeren, niets, blokkeren; als volgen, neigen links, neigen rechts; als positieve, nul, negatieve kritische stroom; als nat, overgang, droog. Driecel-triades zitten in de ontstoken darm. De utendi-strofe sluit op drie gelijke rijmen. 3 × 3 × 3 = 27 ramen, 9 gesloten — en de stabiele zwavelringen zijn van de vorm 2ᵃ·3ᵇ.
3. Eerst stil, dan steil. De jaren vóór reuma en de aanloop naar een sociaal omslagpunt hebben dezelfde vorm, en beide volgen uit de kubische groei van sluiting onder drift — lang bijna onzichtbaar, dan plotseling steil.

5. De menselijke laag

Het zou een mislukking zijn om in taxonomie te eindigen. Achter elk project staan mensen. Patiënten met vroege reumatische ziekte lopen jaren rond met vage klachten die niemand kan plaatsen; als het signaal in de volgorde van hun bezoeken ligt in plaats van in één enkele test, kunnen die jaren korter. Patiënten met MuSK-myasthenia worstelen met slikken, spreken en ademen; als ernst volgt of hun antilichamen een brug sluiten in plaats van hoeveel het er zijn, kan de behandeling per patiënt gekozen worden. Mensen met de ziekte van Crohn leven met ontsteking die na élke behandeling terugkeert; als ze zich vasthoudt als een triade van rollen, verandert de vraag van “welke cel verwijderen” naar “hoe de samenkomst te breken”. De pestslachtoffers van Alkmaar en Kampen stierven in wijken waarvan dichtheid en armoede deel van hun biologie waren. De mensen van Bat bouwden een leven met water dat zij niet konden voorspellen. De dichters van Lamu hielden een verhaal levend in een vorm die het vasthield. Met het net gelezen laten hun sporen zien hoe kwetsbaarheid, veerkracht en geheugen net zo zeer aan hun plaatsen toebehoorden als aan henzelf.

6. Conclusie: wat er in de vraag verandert

Elf sterke onderzoekslijnen in vier faculteiten bereiken elf open problemen die ongerelateerd lijken. Door de Vacuum.Net-theorie gelezen reduceren zij tot drie: vorm, orde en toestand — precies de drie eigenschappen van een geheel die een zoektocht naar oorzaken in onderdelen moet laten liggen. De theorie vervangt de elf onderzoeksprogramma’s niet. Zij verandert de vraag die elk programma aan zijn rand stelt: niet “welk onderdeel doet dit?” maar “wat sluit hier, in welke volgorde, en in welke toestand?”

De eerlijke status van de oefening wordt door haar eigen regels bepaald. Elke koppeling tussen een bevinding en de theorie is een lezing, geen test; elke telling is open en wacht op data die de onderzoekers publiceren of binnenkort gaan publiceren. Maar de elf tellingen zijn goedkoop, concreet en falsifieerbaar: routes in CAZy; odds-ratio-exponenten nabij drie; juncties op domeingrenzen; zwavelringen van de vorm 2ᵃ3ᵇ; escape-mutanten die sluiten in plaats van muteren; verhalen die eindigen in terugkeer die verder reizen; stressmarkers geconcentreerd in pesthospitaals; bouwfases bij nat-droog-overgangen; drie verborgen strategieën; eenarmige antilichaamaandelen die ernst volgen; triades die therapie overleven. Vallen ook maar een handvol van deze tellingen uit zoals voorspeld, dan heeft de drie-gaten-thesis een tweede blik verdiend. Vallen zij niet uit, dan is de theorie tegen het bewijsmateriaal gelezen op de enige manier waarop een theorie gelezen zou moeten worden.

---

Geannoteerde referentielijst

De theorie en haar bronnen

• Konstapel, J. (2026). The Vacuum.Net Theory: Definition, Theorems and Tests, versie 4. Constable Research, Leiden. — De formele onderbouwing van §2: de vijf axioma’s, het bewijs dat ramen alleen sluiten bij alles gelijk of alles verschillend, de sluitingskans onder drift, en de 4π-rotor. De kern zit in de paragrafen 2, 5.2 en Lemma 1.
• Konstapel, J. (2026). Closure on a Ring: The Bethe Ansatz, the Work of J.-S. Caux, and the Closure Rule of the Vacuum.Net. Constable Research, Leiden. — Werkt het visnet, de linker- en rechterring en de naad uit tegen exact oplosbare ketens uit de fysica; een model voor het lezen van een volledige publicatielijst met de theorie.
• Konstapel, J. (2026). One Direction or None: Mirror Molecules, the 2026 Nobel Prize in Chemistry and the Closure Rule of the Vacuum.Net. Constable Research, Leiden. — Past de twee ringen toe op chiraliteit in de chemie; achtergrond voor de lezing van Artola’s twee enzymrichtingen (§3.1).
• Konstapel, J. (2026). What Fits, Stays: Designer Molecules, the Closure Rule of the Vacuum.Net and Three Simple Roads to a Medicine. Constable Research, Leiden. — Selectie door passing, toegepast op de Overkleeft-groep waar Artola werd opgeleid.
• Hestenes, D. (1966). Space-Time Algebra. Gordon and Breach, New York. — Rotatie als twee spiegelingen en de halve-hoek-rotor achter het halfsluitingsresultaat van §2.3.
• Universiteit Leiden (22 oktober 2026). “Vidi-beurs voor elf onderzoekers van de Universiteit Leiden.” — De elf projectomschrijvingen in eigen woorden; de bron van elk “open probleem” in dit essay.

Chemie en materialen

• Artola, M., e.a., Davies, G. J., Overkleeft, H. S. (2017). “1,6-Cyclophellitol cyclosulfates: a new class of irreversible glycosidase inhibitor.” ACS Central Science 3, 784–793. — De twee conformationele routes en de bevinding dat een vastgezette ⁴C₁-stoel α-enzymen selecteert; figuur 1 toont beide routes naast elkaar.
• Beenakker, T. J. M., e.a., Artola, M. (2017). “Carba-cyclophellitols are neutral retaining-glucosidase inhibitors.” JACS 139. — Remming door vorm alleen, zonder reactieve groep — het sterkste bewijs dat passing, niet reactiviteit, selecteert.
• Davies, G. J., Rovira, C., Artola, M., e.a. (2023). “α-D-mannose-configured cyclic sulfamidate as GH47 mannosidase inhibitor.” Chemical Science. — De eenmalige route van een inverterende familie, ter vergelijking met de retainende routes.
• CAZy — Carbohydrate-Active enZYmes database (cazy.org). — De classificatie van glycoside-hydrolase-families in retainend en inverterend; de data voor de routetelling in §3.1.
• Ryazanov, V. V., e.a., Aarts, J. (2001). “Coupling of two superconductors through a ferromagnet: evidence for a π junction.” Physical Review Letters 86, 2427. — De kritische stroom die door nul gaat en van teken wisselt — de data voor de eerste telling van Lahabi (§3.3).
• Lahabi, K., e.a., Aarts, J. (2017). “Controlling supercurrents and their spatial distribution in ferromagnets.” Nature Communications 8, 2056. — Superstroom sturen door een magnetische vortex te verplaatsen; randkanalen onder 80 nm.
• Yasui, Y., Lahabi, K., e.a., Aeno, Y. (2017). “Little-Parks oscillations with half-quantum fluxoid features in Sr₂RuO₄ microrings.” Physical Review B 96. — De half-kwantumtoestanden, gelezen als de naad van de theorie.
• Blom, T. J., Rog, M., e.a., Lahabi, K. (2025–2026). “Emergent network of Josephson junctions in a kagome superconductor.” arXiv:2510.09777. — Interferentie die een junctienetwerk binnen een ongestructureerde CsV₃Sb₃-vlok onthult.
• Rog, M., Blom, T. J., e.a., Lahabi, K. (2026). “Direct observation of channelised supercurrents in a kagome superconductor.” arXiv. — SQUID-beelden van kanalen die bij de kritieke temperatuur verschijnen en blijven — het dichtst bij een foto van de draad.
• Shingledecker, C. N., Lamberts, T., e.a. (2020). “Efficient production of S₈ in interstellar ices.” The Astrophysical Journal 888, 52. — Stralingschemie en niet-diffuse reacties in ijs, en de S₈-ring als resultaat.
• Molpeceres, G., Kästner, J., e.a., Lamberts, T. (2021). “Carbon atom reactivity with amorphous solid water.” J. Phys. Chem. Lett. 12. — Een reactie waarin het ijs zelf meedoet — burenchemie bij 10 K.
• Steudel, R., en Eckert, B. (2003). “Solid sulfur allotropes.” Topics in Current Chemistry 230, 1–79. — De volledige verzameling zwavelringgrootten en hun stabiliteit; de bron voor de 2ᵃ·3ᵇ-telling in §3.4.

Moleculaire en klinische biogeneeskunde

• Loeff, L., Brouns, S. J. J., Joo, C. (2018). “Repetitive DNA reeling by the Cascade-Cas3 complex in nucleotide unwinding steps.” Molecular Cell 70, 385–394. — De bursts van drie basenparen, elk in drie enkele stappen — het raam van drie in een eiwitmotor.
• Loeff, L., e.a., Blokesch, V., Jinek, M. (2024). “Molecular mechanism of plasmid elimination by the DdmDE defense system.” Science 385, 188–194. — Herkenning van een D-loop-structuur in plaats van een sequentie.
• Loeff, L., Walter, A., Rosalen, G. T., Jinek, M. (2025). “DNA end sensing and cleavage by the Shedu anti-phage defense system.” Cell 188, 721–733. — Het zuiverste geval van het hele essay: een verdediging die knipt wat niet sluit, en fagen die ontsnappen door minder open uiteinden te produceren.
• Maarseveen, T. D., e.a., Knevel, R. (2020). “Machine learning electronic health record identification of patients with rheumatoid arthritis.” JMIR Medical Informatics 8, e23930. — De pijplijn die patiënten in vrije tekst vindt — 2.873 van 23.300 in onder zeven seconden, PPV 0,94.
• van Beers-Tas, M. H., e.a. (2020). “Increased primary care use for musculoskeletal symptoms before the diagnosis of inflammatory arthritis.” RMD Open 6, e001163. — De 4–6 jaar aanloop en de odds-ratio’s op 6, 12 en 18 maanden die de kubische prodroomlezing dragen.
• Muller, S., e.a. (2019). “Searching for a prodrome for rheumatoid arthritis in the primary care record.” Seminars in Arthritis and Rheumatism. — Hetzelfde patroon in Britse dossiers: hogere consultfrequenties minstens twee jaar vóór verdenking.
• Klooster, R., Plomp, J. J., Huijbers, M. G., e.a. (2012). “Muscle-specific kinase myasthenia gravis IgG4 autoantibodies cause severe neuromuscular junction dysfunction in mice.” Brain 135, 1081–1101. — IgG4, niet IgG1–3, draagt de ziekte.
• Huijbers, M. G., e.a. (2013). “MuSK IgG4 autoantibodies cause myasthenia gravis by inhibiting binding between MuSK and LRP4.” PNAS 110, 20783–20788. — Het verbroken partnerschap op de junctie.
• Vergoossen, D. L. E., e.a., Huijbers, M. G. (2021). “Functional monovalency amplifies the pathogenicity of anti-MuSK IgG4 in myasthenia gravis.” PNAS 118, e2020635118. — Het kernresultaat: één arm blokkeert, twee armen activeren, en de kloon doet ertoe. Maximale remming bij 7,7 nM; bivalente antilichamen activeren in plaats daarvan.
• Huijbers, M. G. M. (2016). The Pathophysiology of MuSK Myasthenia Gravis. Proefschrift, Universiteit Leiden. — De volledige onderzoekslijn en de observatie dat ernst van meer afhangt dan antilichaamhoeveelheid.
• van Unen, V., e.a., Koning, F. (2016). “Mass cytometry of the human mucosal immune system.” Immunity 44, 1227–1239. — De eerste systeembrede kaart van darm-immuuncellen in ziekte.
• van Unen, V., e.a., Koning, F. (2022). “Identification of a disease-associated network of intestinal immune cells in treatment-naive inflammatory bowel disease.” Frontiers in Immunology 13, 893803. — De 44 subsets, het netwerk onafhankelijk van subtype en ernst, en de drie celsoorten zij aan zij — de data achter de triadelezing.
• Li, N., van Unen, V., e.a., Koning, F. (2019). “Memory CD4+ T cells are generated in the human fetal intestine.” Nature Immunology 20. — Immuungeheugen vóór de geboorte — oriëntatie vastgelegd vóór eerste blootstelling.

Geest, verhaal, verleden samenlevingen

• Urai, A. E., de Gee, J. W., Tsetsos, K., Donner, T. H. (2019). “Choice history biases subsequent evidence accumulation.” eLife 8, e46331. — De kernbevinding: geschiedenis verandert de snelheid, niet de start. De modelvergelijkingsfiguur is het cruciale bewijs.
• Talluri, B. C., Urai, A. E., e.a. (2018). “Confirmation bias through selective overweighting of choice-consistent evidence.” Current Biology 28, 3128–3135. — Bewijs dat met een eerdere keuze overeenkomt, weegt zwaarder.
• Urai, A. E., en Donner, T. H. (2022). “Persistent activity in human parietal cortex mediates perceptual choice repetition bias.” Nature Communications 13, 6015. — Waar in de hersenen de eerdere keuze wordt voortgedragen.
• Ashwood, Z. C., e.a., Urai, A. E., e.a. (2022). “Mice alternate between discrete strategies during perceptual decision-making.” Nature Neuroscience 25, 201–212. — De drie verborgen strategieën — engaged, neigen links, neigen rechts — gelezen als het trit-register.
• Centola, D., Becker, J., Brackbill, D., Baronchelli, A. (2018). “Experimental evidence for tipping points in social convention.” Science 360, 1116–1119. — Het toegewijde-minderheid-experiment; de empirische basis van de Vidi-vraag.
• Raia, A. (2020). Rewriting Yusuf: A Philological and Intertextual Study of a Swahili Islamic Manuscript Poem. Köln: Rüdiger Köppe. — Het Swahili Jozefgedicht tegen zijn bijbelse, Koran- en Arabische bronnen.
• Raia, A. (2018). “Swahili palimpsests: the Muslim stories beneath Swahili compositions.” Swahili Forum 25. — Hoe de verhalen van Jezus, Job en Jozef in Swahili ver herschreven worden.
• Raia, A., en Vierke, C. (red.) (2023). In This Fragile World: Swahili Poetry of Commitment by Ustadh Mahmoud Mau. Leiden: Brill. — Het werk en de bibliotheek van een levend dichter op Lamu.
• Raia, A. (2023). “Transoceanic print histories: twentieth-century Swahili Muslim networks in the Indian Ocean.” Islamic Africa 14, 55–78. — De netwerken waarlangs teksten reizen — transmissie zonder selectieregel.
• Schats, R., en Klomp, M. (2019). “In sickness and in health: an archaeological and osteoarchaeological analysis of St. Gertrude’s infirmary in Kampen.” In: Osteoarchaeology in Historical Context. Leiden: Sidestone Press. — Of hospitaalpopulaties een eigen signatuur dragen — de methode achter de pesthospitaalvergelijking.
• Rommes, R. “Plague in Northwestern Europe: the Dutch experience, 1350–1670.” Popolazione e Storia. — Het patroon van de Nederlandse pest: frequente terugkeer, hangen in arme wijken, lokale mortaliteit tot de helft.
• Swerida, J., Dollarhide, E. N., Jensen, R. (2021). “Settlement and chronology in the Early Bronze Age of southeastern Arabia: the view from Umm an-Nar period Bat, Oman.” Paléorient 47(2), 75–96. — De gedateerde nederzettingsreeks van Bat.
• Swerida, J., Dollarhide, E. N., e.a. (2025). “Results from beyond the oasis: excavations of the 2022–23 field season of the Bat Archaeological Project.” Journal of Oman Studies 25, 28–57. — Bewoning in drie ecologische zones binnen 10 km — de data die de pijl van “aanpassing aan omgeving” breken.
• Swerida, J. (2022). No Place Like Home: Ancient Near Eastern Houses and Households — Structure SS1 at Bat, Oman. Oxford: Archaeopress. — Één huis in detail gelezen: de menselijke schaal van het Bat-landschap.
• “Unlocking the potential of the terrestrial gastropod species Zootecus insularis as a climate archive for arid regions.” Johannes Gutenberg-Universität Mainz. — Het slakkenhuis-klimaatarchief en zijn overeenstemming met grot- en meerreeksen — het register voor de bouwfasetelling in §3.8.

Article

Eleven Questions, Three Gaps

• October 2026

• DOI:
• 10.13140/RG.2.2.12751.65441

Jump to the article here.

J.Konstapel,Leiden, 22-10-2026.

Aanleiding: Vidi-beurs voor elf onderzoekers van de Universiteit Leide

---

Dit essay bundelt veertig jaar supergeleidersonderzoek uit één laboratorium in Leiden. Twee namen staan centraal. Jan Aarts begon met vortexitroosters, ferromagnetische hybrides en de eerste π-junctie. Zijn voormalige student Kaveh Lahabi zette de lijn voort met schijven, ringen en een kagome-supergeleider.

Lahabi heeft onlangs een Vidi-subsidie ontvangen. Daarmee brengt hij in beeld hoe de collectieve supergeleidingsgolf stroomt.

De resultaten vertonen één patroon. De superstroom vult het materiaal niet gelijkmatig. Hij loopt daar waar een kwantumfase in zichzelf terugkeert. Om een vortexkern. Langs de rand van een schijf. Over een paar domeinwanden. Door een netwerk van filamenten.

De theorie van het Vacuum.Net vertrekt vanuit die ene voorwaarde. De kern is eenvoudig: alleen wat in fase terugkeert, blijft bestaan. Dit essay zet de Leidse metingen naast die regel. Vier resultaten volgen er rechtstreeks uit. Ten eerste: de halfkwantumfluxoïde bestaat uit twee halve slagen. Ten tweede: domeinwanden op een ring gaan altijd in paren. Ten derde: de π-junctie doorloopt de tekenreeks +, 0, −. Ten vierde: het kagome-rooster draagt exact één vrije trit aan gesloten configuraties per eenheidscel.

---

1. De sluitingsregel in het kort

De theorie gebruikt één beeld. Zie een visnet in water. Eén draad loopt overal en kruist zichzelf. Knopen ontstaan waar draden om elkaar winden. Mazen liggen tussen de draden. Vier woorden dragen het hele model: draad, winding, knoop, maas.

Op elke kruising heeft de draad één van drie toestanden. Dat zijn +1, 0 en −1. De theorie noemt deze eenheid een trit. Drie kruisingen op een rij vormen een venster. Een venster is gesloten wanneer de drie waarden samen nul opleveren, geteld modulo 3. Een open venster draagt spanning. Alleen wat sluit, blijft.

Drie toestanden op drie plaatsen geven 27 mogelijke vensters. Negen daarvan sluiten. Drie hebben gelijke waarden: (+, +, +), (0, 0, 0) en (−, −, −). Zes hebben drie verschillende waarden, in elke volgorde van (−, 0, +). De overige 18 blijven open.

De gesloten ring verdient nadere aandacht. Een ring van drie verschillende waarden heeft geen beginpunt. Zes volgordes gedeeld door drie rotaties geven twee afzonderlijke ringen. De linkerring loopt − → 0 → +. De rechtering loopt + → 0 → −. Ze zijn spiegelbeelden.

Leg nu een linkerring tegen een rechtering. De reeks luidt: −, 0, +, +, 0, −. Twee van de vier vensters sluiten. Twee blijven open. De naad tussen de ringen levert altijd precies twee open vensters op. Open vensters komen in paren. Dit gegeven keert in de hele natuurkunde terug.

Geef elke trit een fase van 120°. Een venster draagt dan een totaalfase gelijk aan de som maal 120°. Een som van 0 of ±3 geeft een heel aantal slagen. Een som van ±1 of ±2 doet dat niet. Het gesloten venster is dus de voorwaarde dat de fase in zichzelf terugkeert. Dat is exact de voorwaarde die de magnetische flux in een supergeleidende ring kwantiseert.

Twee uitspraken vervolledigen de regel. Eén: nul is geen rust. Nul is de ontmoeting van een handeling met haar spiegel: +1 en −1 samen. Twee: er zijn twee wegen naar nul. De ene is ontmoeting, waarbij twee tegengestelde waarden elkaar opheffen. De andere is rondgang, waarbij drie gelijke stappen een slag voltooien.

---

2. De ring: fluxoïdekwantisatie als sluiting

Een supergeleidende ring draagt één condensaatgolffunctie. Na één volle ronde over de ring moet de fase op zijn beginwaarde terugkeren. De fasewinding is daarom 2πn, met n een geheel getal. De ingesloten fluxoïde is gekwantiseerd:

Φ′ = n · Φ₀, waarbij Φ₀ = h/2e = 2,0678 × 10⁻¹⁵ Wb.

Vlak bij de kritieke temperatuur oscilleert de vrije energie van de ring met de aangelegde magnetische flux. De overgangstemperatuur en de weerstand volgen die oscillaties. Dit zijn de Little–Parks-oscillaties.

De rekenkunde is kort. Neem een ring met een binnenoppervlak van 0,96 µm². De veldperiode is:

ΔB = Φ₀ / A = 2,0678 × 10⁻¹⁵ / (0,96 × 10⁻¹²) ≈ 2,15 mT.

In de termen van het net is de ring één winding. Het kwantumgetal n is zijn adres. Een toestand die niet in fase sluit, bestaat niet. De regel is dezelfde als de sluitingsregel. Alleen de notatie verschilt.

---

3. De halfkwantumfluxoïde: twee halve slagen

Strontiumruthenaat, Sr₂RuO₄, is de belangrijkste kandidaat voor een triplet-supergeleider met gelijke spin. In 2017 frezen Yasui, Lahabi, Aarts, Maeno en collega’s microringen uit enkelkristallen van dit materiaal. Ze maten zuivere Little–Parks-oscillaties. Met een extra magneetveld in het vlak splitsten de pieken. Ze lazen de splitsing als de halfkwantumfluxoïde.

De telling is kort. In een triplettoestand met gelijke spin heeft de ordeparameter twee componenten:

Ψ = Δ (e^{iθ↑} |↑↑⟩ + e^{iθ↓} |↓↓⟩).

Elke component moet op zichzelf éénduidig zijn. rond de ring wint de fase θ↑ met 2π·m↑. De fase θ↓ wint met 2π·m↓. De gehele getallen m↑ en m↓ hoeven niet gelijk te zijn. De fluxoïde volgt de gemiddelde ladingfase:

n′ = (m↑ + m↓) / 2.

Neem m↑ = 1 en m↓ = 0. Dan is n′ = ½. De ladingfase wint een halve slag. De spinas draait de andere helft. Samen vormen ze één hele slag.

De sluitingsregel geeft dit een directe lezing. Eén slag van een halfgeheel getal is maar een halve sluiting. Een volledige sluiting vraagt een tweede helft. De halfkwantumfluxoïde laat precies die structuur zien: half in lading, half in spin, en alleen gesloten als paar. Het experiment bevestigt dat halfsluitingen in de natuur voorkomen — en altijd in paren verschijnen.

---

4. Spontane juncties: wanden komen in paren

Sr₂RuO₄ is ook een kandidaat-chirale supergeleider. De ordeparameter kan links- of rechtsdraaiend zijn. Domeinen met tegengestelde chiraliteit kunnen naast elkaar bestaan. Op de wand tussen twee domeinen is het condensaat onderdrukt. Zo’n wand kan als een intrinsieke Josephson-junctie werken.

In 2020 rapporteerden Yasui, Lahabi en collega’s een opvallende vondst. Ze vonden zulke juncties in ringen die volledig homogeen waren. Niemand had ze gefabriceerd. Ringen met een scherpe overgang bij 1,5 K toonden kritische-stroomoscillaties met de periode van één fluxkwantum. Ze gedroegen zich als een dc-SQUID met twee juncties. Ringen met een brede overgang rond 3 K toonden geen juncties. Alleen gewone Little–Parks-oscillaties. Simulaties van de ordeparameter plaatsten stabiele domeinwanden over beide armen van de ring.

Een pariteitstelling legelt het aantal wanden. Chiraliteit heeft twee waarden. Elke wand keert die om. Een pad rond de ring moet terugkeren bij zijn beginchiraliteit. Het moet dus een even aantal wanden kruisen. Het kleinste niet-nul even getal is twee.

De telling klopt exact met de meting: twee juncties, of geen. Ze klopt ook met de naad uit paragraaf 1. Een linkerring tegen een rechtering geeft twee open vensters. Ze komen als paar. In de Leidse ringen verscheen dat paar als twee juncties die niemand had gemaakt. Het pariteitsargument maakt hun koppeling onvermijdelijk. Het experiment bevestigde die.

---

5. De π-junctie: één component, drie toestanden

In 2001 rapporteerden Ryazanov, Oboznov, Rusanov, Veretennikov, Golubov en Aarts de eerste π-junctie. Het apparaat bestond uit niobium-elektroden en een ferromagnetische laag van Cu₀,₄₈Ni₀,₅₂, ongeveer 22 nm dik.

Tijdens het afkoelen zakte de kritieke stroom naar nul en steeg daarna weer. De auteurs toonden aan dat de koppeling van teken wisselt. Boven het omslagpunt is de junctie in de 0-toestand. Eronder is hij in de π-toestand. De omslag ligt waar de thermische energie en de exchange-energie van dezelfde orde zijn.

Schrijf de kritieke stroom met zijn teken. Bij het afkoelen verloopt hij:

+Ic → 0 → −Ic.

De koppeling doorloopt alle drie de tritwaarden: +, 0 en −. De nul is geen derde fase van de junctie. Het is het punt waar 0-koppeling en π-koppeling elkaar ontmoeten en opheffen. Dat is de definitie van nul in het net: een handeling die haar spiegel ontmoet.

De π-toestand zelf is een halve slag. Over de junctie springt de fase met π. Dit is de derde verschijning van hetzelfde object in het Leidse werk. De halfkwantumfluxoïde uit paragraaf 3 is één instantie. De chirale domeinwand uit paragraaf 4 is de tweede. De π-koppeling is de derde. Eén meetkundig motief — de halve slag die een partner nodig heeft — keert terug in drie zeer verschillende apparaten.

---

6. Domeinwanden: sluiting door verschil

Een ferromagnet onderdrukt supergeleiding via zijn exchangeveld. In een gelijkmatig gemagnetiseerde laag wijst dat veld één kant op. Het paarbrekende effect is dan sterk. Maar in 2004 onderzochten Rusanov, Hesselberth, Aarts en Buzdin niobium op permalloy. Ze vonden een duidelijke versterking van supergeleiding in een smal veldvenster van ongeveer 0,5 mT. In dat venster schakelt het permalloy en bestaan er vele magnetische domeinen naast elkaar.

Hun verklaring richtte zich op de domeinwanden. Binnen een wand proeft een Cooperpaar vele magnetisatierichtingen tegelijk. De paarbrekende effecten middelen deels uit. De voorwaarde is meetkundig. De breedte van de wand moet overeenkomen met de coherentielengte: δw ≈ ξS(T). In micro gestructureerde monsters zonder wanden bleef het effect uit.

De sluitingsregel heeft een passende structuur. Er zijn twee soorten gesloten vensters. De ene bevat drie gelijke waarden. De andere bevat drie verschillende. Een gelijkmatige magneet is een venster van één waarde: het draagt spanning. Een domeinwand biedt verschillende richtingen binnen één coherenti lengte. Het Cooperpaar ziet dan een venster dat sluit door verschil.

De passing is streng. De wand moet het formaat van het paar hebben. Dit is selectie door passing. Wat past, keert in fase terug. Wat niet past, verstrooit. Het experiment mat precies waar die passing rendeert — en waar zij verdwijnt.

---

7. De rand selecteert

Gedurende dertig jaar vond de Leidse groep één consistent resultaat. De grens bepaalt waar supergeleiding leeft.

• 1991. Maj en Aarts toonden dat in een Nb/NbZr-meerlaag de buitenste lagen het parallèle kritieke veld bepalen.
• 2017. Lahabi en collega’s bouwden een schijfvormige Josephson-junctie met een ferromagnetische vortex. De superstroom splitste in twee kanalen. Een klein veld verplaatste de vortex — en daarmee de stroom.
• 2022. In Nb/Co-schijven liep de triplet-supersroom op lange afstand in smalle kanalen langs de rand, smaller dan 80 nm. De grens tussen supergeleider en vacuüm speelde een centrale rol. Door de spintextuur te veranderen verschenen 0- en π-kanalen in één apparaat.
• 2024. Met het halfmetaal La₀,₇Sr₀,₃MnO₃ bleef de triplet-stroom beperkt tot de rand van schijven. In staafjes was hij gelijkmatig. De randstroom hoorde bij de schijfmeetkunde, niet bij het mechanisme dat de tripletten opwekte.

De lezing is kort. Een gesloten rand is één winding. De schijf biedt een pad dat in zichzelf terugkeert. De staaf doet dat niet. De stroom neemt het pad dat sluit.

Uit hetzelfde beginsel volgt een geheugenelement. In 2022 bouwden Fermin, Scheinowitz, Aarts en Lahabi een elliptische junctie. Een puls van 40 mT schakelt het apparaat tussen twee stabiele texturen: gelijkmatige magnetisatie, of een paar vortices. Elke textuur geeft een andere kritieke stroom. Elke textuur is een gesloten configuratie met een eigen adres. Schrijven betekent: de toestand van het medium veranderen. De configuratie houdt daarna stand zonder voeding. Dat is het derde axioma van de sluitingsregel in hardware: wat gesloten is, draagt geheugen.

---

8. Het vortexitrooster: een net van windingen

In een type-II-supergeleider trekt magnetische flux binnen als vortices. Elke vortex draagt één fluxkwant. Rond zijn kern wint de fase met 2π. Elke vortex is één gesloten winding. De vortices ordenen zich in een driehoekig rooster met mazen ertussen. Van alle structuren in de gecondenseerde materie is dit de meest letterlijke afbeelding van het net.

In 1999 beeldden Troyanovski, Aarts en Kes dit rooster af met scanningtunnelmicroscopie bij hoge fluxdichtheden. Twee resultaten vallen op.

Eén: met sterke lijndefecten blijft het rooster vastgepind tot er ongeveer twee keer zoveel vortices als defecten zijn. Daarboven begint plastische kruip. Het rooster vloeit om zijn pinplaatsen heen.

Twee: met zwakke puntverstoring kruipen de vortices coërent langs een hoofdas van het rooster. Hun snelheid moduleert met de roosterperiode. Elke vortex beweegt in het ritme van het hele rooster.

Het tweede resultaat illustreert het vierde axioma van het net. De reactie van een winding is een functie van de toestand van haar omgeving. Hier volgt de snelheid van elke vortex de periode van het rooster waartoe hij behoort. Het net beweegt als één lichaam.

---

9. De kagome-supergeleider: een netwerk van vensters

9.1 De metingen

Het kagome-rooster is een netwerk van driehoeken die hoeken delen. Elke driehoek is een groep van drie posities. In de kagome-supergeleider CsV₃Sb₅ vond de groep van Lahabi twee grote resultaten.

Eén: een ongestructureerd vlokje bevat zijn eigen netwerk van Josephson-juncties. Blom, Rog en collega’s rapporteerden dit in 2026. Het netwerk ontstaat onder de kritieke temperatuur. Het toont gekwantiseerde Schapiro-stappen. Het overleeft nanostructurering en thermische kringen. Niemand heeft het gefabriceerd.

Twee: een SQUID-microscoop beeldt het netwerk rechtstreeks af. Rog, Blom en collega’s rapporteerden dit in 2026. Smalle superstroomkanalen verschijnen bij de kritieke temperatuur. Ze blijven stabiel bij alle temperaturen en stromen. Hun niet-lineaire respons past bij juncties die verbonden zijn door filamenten. Het netwerk is sterk bij lichte tin-dotering (x = 0,03). In ongedoteerde vlokken is het veel zwakker.

Het netwerk verklaart verschillende anomalieën die voor deze materiaalfamilie gerapporteerd zijn. Het verklaart de intrinsieke juncties, de paren met hogere lading en een supergeleidings-diodeffect zonder aangelegd veld.

9.2 De sluitingsregel op het rooster

Het kagome-rooster nodigt uit tot een exacte telling. Neem een periodiek kagome-rooster van L × L eenheidscellen. Elke cel bevat 3 posities en 2 driehoeken: één omhoog, één omlaag. Het rooster heeft N = 3L² posities en 2L² driehoeken.

Plaats een trit op elke positie. Eis dat elke driehoek sluit: de som van de drie trits moet 0 zijn modulo 3. Elke driehoek geeft één lineaire vergelijking over het eindige lichaam GF(3).

Tel de oplossingen. Elke positie behoort tot exact één omhoog-driehoek en één omlaag-driehoek. De som van alle omhoog-vergelijkingen is daarom gelijk aan de som van alle omlaag-vergelijkingen. Dat is één afhankelijkheid. Directe berekening met Gauss-eliminatie over GF(3) geeft de rang voor L = 2, 3, 4 en 6:

L	Posities	Driehoeken	Rang	Vrije trits
2	12	8	7	5
3	27	18	17	10
4	48	32	31	17
6	108	72	71	37

De rang is 2L² − 1. Het aantal vrije trits is 3L² − (2L² − 1) = L² + 1.

Het resultaat is helder. De sluitingsregel laat exact één vrije trit per eenheidscel over. Elke eenheidscel draagt daarmee drie gesloten configuraties.

9.3 Alleen ringen

Beperk nu de telling tot ringen. Elke driehoek moet drie verschillende waarden dragen. Dit is het probleem van het inkleuren van de kagome-posities met drie kleuren, zo dat elke driehoek alle drie toont. Het is gelijkwaardig aan het inkleuren van de randen van het honingraatrooster. Baxter loste dit exact op in 1970. Het aantal toestanden per honingraat-eenheidscel is:

W² = ∏(k ≥ 1) (3k − 1)² / [(3k)(3k − 2)] = 1,4610.

Een transfermatrixtelling op stroken van breedte 4 tot 10 convergeert naar dezelfde waarde (1,414; 1,587; 1,482; 1,459 per cel). Eén honingraatcel beantwoordt aan één kagome-cel.

Ringvormige vensters vullen dus 1,461 van de 3 toestanden per cel: ongeveer 49%. De rest wordt gedragen door vensters met gelijke waarden.

9.4 De lezing

Het rooster heeft al de vorm van het net. Elke driehoek is een venster. De filamenten liggen waar de vensters langs een pad sluiten. De juncties zijn de open vensters ertussen.

Het netwerk heeft lichte dotering nodig om volledig te verschijnen. Dat is het vierde axioma: de reactie hangt af van de toestand van het medium. De juncties overleven thermische kringen en nanostructurering. Dat is het derde axioma: gesloten configuraties behouden hun vorm.

9.5 Hogere lading: 4e en 6e

In 2024 maten Ge en collega’s ringen van CsV₃Sb₅. Bij lage temperatuur was de fluxperiode h/2e. Bij opwarming werd dat h/4e. Vlak bij het begin van de supergeleiding werd het h/6e. In grotere ringen verscheen h/4e niet. De periode liep van h/2e rechtstreeks naar h/6e.

Voor de ring van 0,96 µm² zijn de perioden:

• h/2e: 2,15 mT
• h/4e: 1,08 mT
• h/6e: 0,72 mT

De sluitingsregel biedt een lezing. Geef elk Cooperpaar een fasestap van 120°. Drie paren sluiten dan op dezelfde twee manieren als een venster. Of ze dragen 0°, 120° en 240° — de ring. Of ze nemen drie gelijke stappen — de rondgang. Een lading van 6e is een venster van drie paren. Een lading van 4e is de ontmoeting van twee. Welke sluiting bij welke ringgrootte optreedt, is een telling die nog openstaat.

---

10. Mozaïek, grensvlak en de eerste zware fermion

Drie oudere lijnen in het werk van Aarts vervolledigen het beeld.

Manganieten. In 1999 beeldden Fäth, Freisem, Menovsky, Tomioka, Aarts en Mydosh een gedoteerd manganiet af met tunnelpectroscopie. Metallische en isolerende gebieden bestonden naast elkaar op submicronschaal. De textuur overleefde velden tot 9 tesla. De overgang gebeurt niet gelijkmatig. Gesloten gebieden groeien tot zij verbonden zijn. In het net liggen open en gesloten mazen naast elkaar, en het patroon behoudt zijn vorm.

Oxide-grensvlakken. LaAlO₃ en SrTiO₃ zijn beide isolatoren. Hun grensvlak geleidt. De groep van Aarts groeide zulke grensvlakken door sputteren en onderzocht hoe elektronen daar vastzitten. Twee lagen die niet geleiden, dragen een derde die dat wel doet. In de oudste vorm van de theorie is dit wederzijdse ondersteuning: twee dingen die ontstaan zijn, dragen het volgende.

CeCu₂Si₂. In 1979 rapporteerden Steglich, Aarts en collega’s supergeleiding in CeCu₂Si₂. Het was de eerste niet-conventionele supergeleider. Hij ontstaat uit zware quasipartikelen te midden van sterk magnetisme, met een kritieke temperatuur van 0,7 K. Roostervibraties verklaren hem niet. In het net is ook geen externe bindende factor nodig. Wat in fase terugkeert, blijft.

---

11. Richting: de open vraag

Het register van het net noteert sommen. Maar twee volgordes met dezelfde som kunnen toch verschillen. Een reeks die naar sluiting loopt, is niet hetzelfde als een die er vandaan loopt. De theorie heeft daarom een grootheid nodig die volgorde registreert, niet alleen totalen.

CsV₃Sb₅ biedt de zuiverste laboratoriumtest. Het materiaal toont een supergeleidings-diode-effect zonder aangelegd veld. De kritieke stroom hangt af van de richting. De som is gelijk. De respons is dat niet. Dit is de meest directe plaats om naar die ontbrekende grootheid te zoeken.

---

12. Waar de telling doorgaat

Ringen van drie en zes. Het net telt dat een ring van drie verschillende waarden niet met één positie kan groeien. Hij groeit met drie, naar zes. Een ring van zes splitst in twee ringen van drie. De Leidse groep freest ringen uit enkelkristallen, drukt juncties met een elektronenbundel en werkt in een materiaal wiens rooster uit driehoeken en zeshoeken bestaat. De telling kan op een chip.

De vorm van nul. In het net is nul een ontmoeting, geen toestand. In een π-junctie moet het omslagpunt dan één punt zijn waar de kritieke stroom nul raakt, geen interval. De gepubliceerde Ic(T)-kromsten staan die telling toe.

Het kagome-getal. Eén vrije trit per cel is een berekende eigenschap van de sluitingsregel. De SQUID-microscoop ziet welke paden sluiten. De volgende stap is: de gemeten dichtheid van gesloten paden vergelijken met de telling uit paragraaf 9.

De Vidi-microscoop. Het nieuwe project van Lahabi wil in beeld brengen hoe de collectieve golf van een kwantummateriaal stroomt. Dat is een rechtstreeks beeld van waar de sluiting ligt.

---

13. Methode en status

Elk getal in dit essay is herberekend. De fluxperioden volgen uit h/Qe gedeeld door het ringoppervlak. De halfkwantumfluxoïde volgt uit éénduidigheid per spincomponent. De wandpariteit volgt uit de tweetallige chiraliteit. De kagome-rang is berekend met Gauss-eliminatie over GF(3) voor L = 2, 3, 4 en 6. Het product van Baxter is geëvalueerd tot twee miljoen factoren en gecontroleerd met transfermatrices op stroken van breedte 4, 6, 8 en 10.

De status van de uitspraken is expliciet:

• Gemeten: alle experimentele resultaten in de paragrafen 2 tot en met 11.
• Bewezen: fluxoïdekwantisatie; de halfkwantumtelling; wandpariteit; de kagome-rang 2L² − 1 voor de berekende groottes; het resultaat van Baxter.
• Berekend: fluxperioden; vrije trits per cel; het aandeel van 49% ringen.
• Interpretatie: fluxoïde als winding; halfkwantumfluxoïde, chirale wand en π-koppeling als naad; de π-omslag als de tritreeks; wanden als sluiting door verschil; rand, vortexitrooster en kagome-netwerk als het net; 4e en 6e als ontmoeting en rondgang.
• Open: de 4e/6e-telling per ringgrootte; de vorm van de π-omslag; de gemeten dichtheid van gesloten paden in kagome-vlokken; ringen van drie en zes.

---

Geannoteerde referentielijst

Het theoretische kader

1. Konstapel, J. (2026). The Vacuum.Net Theory: Foundational Paper. Constable Research, Leiden.
Waarom lezen? Het werk stelt de axioma’s op die dit essay doorlopend gebruikt: sluiting, geheugen en toestandsafhankelijke respons. Het is de bron van de trit, het venster en de sluitingsregel zelf. Leesadvies: begin bij Deel 0 en het vensterregister.

2. Konstapel, J. (2026). “Closure on a Ring: The Bethe Ansatz, the Work of J.-S. Caux, and the Closure Rule of the Vacuum.Net.” constable.blog, 10 oktober 2026.
Waarom lezen? Het werk ontwikkelt de 27 vensters, de 120°-fasetoewijzing en de naadtabel. Het formuleert ook de open vraag van ringen van drie en zes. Leesadvies: paragrafen 1, 3 en 6.

3. Konstapel, J. (2026). “The Way Before the Net: Light, Location and the Oldest Dao as the Ground of the Vacuum.Net.” constable.blog, 20 oktober 2026.
Waarom lezen? Het werk funderet de interpretatie van nul als ontmoeting, de twee wegen naar sluiting, wederzijdse ondersteuning en de open vraag van de volgorde. Leesadvies: paragrafen 6 en 9.

Ringen en halve slagen

4. Yasui, Y., Lahabi, K., Anwar, M. S., Nakamura, Y., Yonezawa, S., Terashima, T., Aarts, J., en Maeno, Y. (2017). “Little-Parks oscillations with half-quantum fluxoid features in Sr₂RuO₄ microrings.” Physical Review B 96, 180507(R).
Waarom lezen? Het artikel rapporteert zuivere Little–Parks-oscillaties in ringen die uit bulk-enkelkristallen gefreesd zijn. De pieksplitsing onder een veld in het vlak wordt gelezen als de halfkwantumfluxoïde. Het is de experimentele pijler van paragraaf 3. Leesadvies: bestudeer de schets van de vrije energie in de eerste figuur.

5. Yasui, Y., Lahabi, K., Fernández Becerra, V., Fermin, R., Anwar, M. S., Yonezawa, S., Terashima, T., Milošević, M. V., Aarts, J., en Maeno, Y. (2020). “Spontaneous emergence of Josephson junctions in homogeneous rings of single-crystal Sr₂RuO₄.” npj Quantum Materials 5, 21.
Waarom lezen? Twee juncties die niemand fabriceerde, op de wanden tussen chirale domeinen. Het artikel is vrij toegankelijk. Het is de experimentele pijler van paragraaf 4. Leesadvies: leg de ringen met scherpe en met brede overgang naast elkaar.

6. Lahabi, K. (2018). Spin-triplet supercurrents of odd and even parity in nanostructured devices. Proefschrift, Universiteit Leiden.
Waarom lezen? Het bundelt het schijf- en ringwerk in één deel, met vermelding van de auteursbijdragen per hoofdstuk. Nuttig om te traceren welke resultaten gezamenlijk werk zijn.

De π-junctie en de wanden

7. Ryazanov, V. V., Oboznov, V. A., Rusanov, A. Yu., Veretennikov, A. V., Golubov, A. A., en Aarts, J. (2001). “Coupling of two superconductors through a ferromagnet: evidence for a π junction.” Physical Review Letters 86, 2427.
Waarom lezen? De eerste π-junctie. De kritieke stroom zakt naar nul en keert terug met tegengesteld teken. Dit is de experimentele pijler van paragraaf 5. Leesadvies: let op de Ic(T)-kromsten rond de cousvlakte.

8. Rusanov, A. Yu., Hesselberth, M. B. S., Aarts, J., en Buzdin, A. I. (2004). “Enhancement of the superconducting transition temperature in Nb/permalloy bilayers by controlling the domain state of the ferromagnet.” Physical Review Letters 93, 057002.
Waarom lezen? Supergeleiding versterkt door domeinwanden waarvan de breedte overeenkomt met de coherentiëntie. Het is de experimentele pijler van paragraaf 6. Het artikel toont dat sluiting door verschil een strenge meetkundige voorwaarde heeft.

De rand

9. Maj, W., en Aarts, J. (1991). “Outer layers determine the parallel critical field of a superconducting multilayer.” Physical Review B.
Waarom lezen? De vroegste uitspraak in deze onderzoekslijn dat de grens beslist. Het beginsel werd later op nanometerschaal door de randexperimenten bevestigd.

10. Lahabi, K., Amundsen, M., Ouassou, J. A., Beukers, E., Pleijster, M., Linder, J., Alkemade, P., en Aarts, J. (2017). “Controlling supercurrents and their spatial distribution in ferromagnets.” Nature Communications 8, 2056.
Waarom lezen? Een vortex in een schijf splitst de superstroom in twee kanalen. Een klein veld verplaatst de vortex en daarmee de stroom. Vrij toegankelijk. Leesadvies: merk op dat de ruimtelijke sturing volgt uit meetkunde alleen.

11. Fermin, R., van Dinter, D., Hubert, M., Woltjes, B., Silaev, M., Aarts, J., en Lahabi, K. (2022). “Superconducting triplet rim currents in a spin-textured ferromagnetic disk.” Nano Letters 22, 2209.
Waarom lezen? Randkanalen smaller dan 80 nm, de centrale rol van de vacümgrens, en 0- en π-kanalen in één apparaat. Vrij toegankelijk. Dit is de scherpste experimentele uitspraak dat de rand het pad kiest.

12. Fermin, R., Yao, J., Lahabi, K., en Aarts, J. (2023). “The unusual distribution of spin-triplet supercurrents in disk-shaped Josephson junctions.” Superconductor Science and Technology 36, 064001.
Waarom lezen? Een compact overzicht van het hele schijfprogramma in elf bladzijden. De beste enkele toegang tot de literatuur over randstromen.

13. Yao, J., Fermin, R., Cabero, M., Lahabi, K., en Aarts, J. (2024). “Triplet supercurrents in lateral Josephson junctions with a half-metallic ferromagnet.” Physical Review Research 6, 043114.
Waarom lezen? Randstromen verschijnen in schijven, niet in staven. Alleen de meetkunde kiest het gesloten pad. Dit sluit het tripletgeneratiemechanisme uit als oorzaak van de beperking.

14. Fermin, R., Scheinowitz, N. M. A., Aarts, J., en Lahabi, K. (2022). “Mesoscopic superconducting memory based on bistable magnetic textures.” Physical Review Research 4, 033136.
Waarom lezen? Twee stabiele magnetische texturen dienen als twee opgeslagen toestanden. Het apparaat demonstreert het geheugenaxioma van de sluitingsregel in hardware: een gesloten configuratie houdt stand zonder voeding.

Het vortexitrooster

15. Troyanovski, A. M., Aarts, J., en Kes, P. H. (1999). “Collective and plastic vortex motion in superconductors at high flux densities.” Nature 399, 665.
Waarom lezen? Het rooster van windingen in beweging afgebeeld. Het behandelt pinnen, plastische kruip en coërente kruip in het ritme van het rooster. De coërente kruip is het vierde netaxioma dat rechtstreeks zichtbaar wordt.

Kagome

16. Blom, T. J., Rog, M., Altena, M., Capa Salinas, A. N., Lee, J., Regter, R. Q., Wigbout, L. J., Wilson, S. D., Allan, M. P., Li, C., en Lahabi, K. (2026). “Emergent network of Josephson junctions in a kagome superconductor.” Nano Letters 26, 7874. Preprint: arXiv 2510.09777.
Waarom lezen? Schapiro-stappen en interferentiepatronen van juncties die intrinsiek zijn aan een ongestructureerde vlok CsV₃Sb₅. Het netwerk overleeft nanostructurering en thermische kringen. Dit is de eerste pijler van paragraaf 9.

17. Rog, M., Blom, T. J., Regter, R. Q., Capa Salinas, A., Benali, D., Lee, J., Boltje, D. B., Fischer, M. H., Neupert, T., Wilson, S. D., Allan, M. P., en Lahabi, K. (2026). “Direct observation of channelised supercurrents in a kagome superconductor.” arXiv 2606.20157.
Waarom lezen? Het netwerk rechtstreeks afgebeeld met een SQUID-microscoop, inclusief de afhankelijkheid van tin-dotering. Dit is de tweede pijler van paragraaf 9. Leesadvies: bestudeer de kaarten van de stroomdichtheid.

18. Rog, M., Blom, T. J., Boltje, D. B., de Haan, J. D., Fermin, R., Niu, J., Doedes, Y. C., Allan, M. P., en Lahabi, K. (2026). “Tapping-mode SQUID-on-tip microscopy with proximity Josephson junctions.” Nano Letters 26, 1608.
Waarom lezen? Het beschrijft het instrument achter de stroomdichtheidskaarten. Belangrijk om de resolutie en betrouwbaarheid van de beelden te beoordelen.

19. Ge, J., Wang, P., Xing, Y., Yin, Q., Wang, A., Shen, J., Lei, H., Wang, Z., en Wang, J. (2024). “Charge-4e and charge-6e flux quantization and higher charge superconductivity in kagome superconductor ring devices.” Physical Review X 14, 021025.
Waarom lezen? Het rapporteert fluxperioden van h/4e en h/6e, en hun volgorde met temperatuur en ringgrootte. Het levert de data achter paragraaf 9.5 en de open 4e/6e-telling.

20. Baxter, R. J. (1970). “Colorings of a hexagonal lattice.” Journal of Mathematical Physics 11, 784.
Waarom lezen? De exacte oplossing voor het aantal drie-kleuringen van het honingraatrooster. Het essay gebruikt haar in paragraaf 9.3 om alleen-ringconfiguraties te tellen. Het artikel is kort en volledig.

Oudere lijnen

21. Steglich, F., Aarts, J., Bredl, C. D., Lieke, W., Meschede, D., Franz, W., en Schäfer, H. (1979). “Superconductivity in the presence of strong Pauli paramagnetism: CeCu₂Si₂.” Physical Review Letters 43, 1892.
Waarom lezen? De eerste niet-conventionele supergeleider. Zware quasipartikelen paren te midden van sterk magnetisme bij 0,7 K. Vier bladzijden die een vakgebied openden.

22. Fäth, M., Freisem, S., Menovsky, A. A., Tomioka, Y., Aarts, J., en Mydosh, J. A. (1999). “Spatially inhomogeneous metal-insulator transition in doped manganites.” Science 285, 1540.
Waarom lezen? Het submicronmozaïek van metallische en isolerende gebieden, stabiel tot 9 tesla. Een rechtstreeks beeld van gesloten en open mazen naast elkaar.

23. Yin, C., Smink, A. E. M., Leermakers, I., Tang, L. M. K., Lebedev, N., Zeitler, U., van der Wiel, W. G., Hilgenkamp, H., en Aarts, J. (2020). “Electron trapping mechanism in LaAlO₃/SrTiO₃ heterostructures.” Physical Review Letters.
Waarom lezen? Geleiding op de ontmoeting van twee isolatoren. Het grensvlak is een derde laag die twee andere dragen — wederzijdse ondersteuning in hardware.

24. Blom, T. J., Mechielsen, T. W., Fermin, R., Hesselberth, M. B. S., Aarts, J., en Lahabi, K. (2021). “Direct-write printing of Josephson junctions in a scanning electron microscope.” ACS Nano 15, 322.
Waarom lezen? Het beschrijft het fabricagegereedschap dat ringen van drie en zes op een chip kan plaatsen. Het verbindt de theoretische telling uit paragraaf 12 met de experimentele praktijk.

Profielen

25. Universiteit Leiden. Persoonspagina’s van K. Lahabi en J. Aarts, met volledige publicatielijsten. universiteitleiden.nl.
Waarom lezen? De volledige publicatielijsten waarop dit essay steunt. De pagina’s documenteren de volle omvang van beide carrières, verder dan de geselecteerde resultaten.

26. Universiteit Leiden (2026). “Vidi-subsidies voor 11 onderzoekers van Universiteit Leiden.” Nieuws, 22 oktober 2026.
Waarom lezen? Het formuleert het doel van Lahabi’s nieuwe microscoopproject in vier zinnen. Het project beeldt af waar de collectieve golf stroomt — en dus waar de sluiting ligt.

Article + Essay:

DOI: Where the Supercurrent Closes The Leiden Work of J. Aarts and K. Lahabi and the Closure Rule of the Vacuum.Net

Where the Supercurrent Closes

A Critical Essay on the Leiden Work of Jan Aarts and Kaveh Lahabi, and the Closure Rule of the Vacuum.Net

Based on J. Konstapel (2026), “Where the Supercurrent Closes,” Constable Research, Leiden.

---

Executive Summary

This essay reviews four decades of superconductivity research from a single laboratory in Leiden. Two physicists anchor the story. Jan Aarts began with vortex lattices, ferromagnetic hybrids, and the first π junction. His former student Kaveh Lahabi continued with disks, rings, and a kagome superconductor. Lahabi now holds a Vidi grant to image how the collective superconducting wave flows.

Their results share one pattern. The supercurrent does not spread evenly through a material. It runs where a quantum phase returns onto itself. It runs around a vortex core. It runs along the rim of a disk. It runs across a pair of domain walls. It runs through a network of filaments.

The Vacuum.Net theory starts from that single condition. Its claim is simple: only what returns in phase persists. This essay sets the Leiden measurements beside that rule. Four results follow from it directly. First, the half-quantum fluxoid appears as two half-turns. Second, domain walls on a ring must come in pairs. Third, the π junction passes through the sign sequence +, 0, −. Fourth, the kagome lattice carries exactly one free trit of closed configurations per unit cell.

---

1. The Closure Rule in Brief

The theory uses one image. Picture a fishnet in water. One strand runs everywhere and crosses itself. Knots form where the strands wind around each other. Meshes lie between the strands. Four words carry the whole model: strand, winding, knot, mesh.

At every crossing, the strand holds one of three states. These are +1, 0, and −1. The theory calls this unit a trit. Three crossings in a row form a window. A window is closed when its three values sum to zero, counted modulo 3. An open window carries tension. Only what closes stays.

Three states on three places give 27 possible windows. Nine of them close. Three hold equal values: (+, +, +), (0, 0, 0), and (−, −, −). Six hold all three different values, in every order of (−, 0, +). The remaining 18 stay open.

The closed ring deserves a closer look. A ring of three different values has no natural starting point. Six orderings divided by three rotations give two distinct rings. The left ring runs − → 0 → +. The right ring runs + → 0 → −. They are mirror images.

Now place a left ring against a right ring. The sequence reads −, 0, +, +, 0, −. Two of the four windows close. Two stay open. The seam between the two rings always produces exactly two open windows. Open windows come in pairs. This pairing will return throughout the physics.

Give each trit a phase of 120°. A window then carries a total phase equal to its sum times 120°. A sum of 0 or ±3 gives a whole number of turns. A sum of ±1 or ±2 does not. The closed window is therefore the condition that the phase returns onto itself. This is precisely the condition that quantizes magnetic flux in a superconducting ring.

Two further statements complete the rule. First, zero is not rest. Zero is the meeting of an act with its mirror: +1 and −1 together. Second, there are two ways to reach zero. One is meeting, where two opposite values cancel. The other is going round, where three equal steps complete a turn.

---

2. The Ring: Fluxoid Quantization as Closure

A superconducting ring carries one condensate wave function. After one full turn around the ring, its phase must return to its starting value. The phase winding is therefore 2πn, with n an integer. The enclosed fluxoid is quantized:

Φ′ = n · Φ₀, where Φ₀ = h/2e = 2.0678 × 10⁻¹⁵ Wb.

Near the critical temperature, the free energy of the ring oscillates with the applied magnetic flux. The transition temperature and the resistance follow those oscillations. Physicists call these the Little–Parks oscillations.

The arithmetic is short. Take a ring with an inner area of 0.96 µm². The field period is:

ΔB = Φ₀ / A = 2.0678 × 10⁻¹⁵ / (0.96 × 10⁻¹²) ≈ 2.15 mT.

Read in the terms of the net, the ring is one winding. The quantum number n is its address. A state that does not close in phase does not exist. The rule is the same as the closure rule. Only the notation differs.

---

3. The Half-Quantum Fluxoid: Two Half-Turns

Strontium ruthenate, Sr₂RuO₄, is a leading candidate for an equal-spin triplet superconductor. In 2017, Yasui, Lahabi, Aarts, Maeno, and colleagues milled microrings from single crystals of this material. They observed proper Little–Parks oscillations. When they added an in-plane magnetic field, the peaks split. They read the splitting as the half-quantum fluxoid.

The count is short. In an equal-spin triplet state, the order parameter has two components:

Ψ = Δ (e^{iθ↑} |↑↑⟩ + e^{iθ↓} |↓↓⟩).

Each component must be single-valued on its own. Around the ring, the phase θ↑ winds by 2π·m↑. The phase θ↓ winds by 2π·m↓. The integers m↑ and m↓ need not be equal. The fluxoid follows the mean charge phase:

n′ = (m↑ + m↓) / 2.

Take m↑ = 1 and m↓ = 0. Then n′ = ½. The charge phase winds by a half turn. The spin axis rotates by the other half. Together they make one whole turn.

The closure rule gives this a direct reading. One turn of a half-integer carrier is only half a closure. A complete closure needs a second half. The half-quantum fluxoid shows exactly that structure: half in charge, half in spin, and closed only as a pair. The experiment confirms that half-closures occur in nature, and that they arrive paired.

---

4. Spontaneous Junctions: Walls Come in Pairs

Sr₂RuO₄ is also a candidate chiral superconductor. Its order parameter can turn left or right. Domains of opposite chirality can coexist in one crystal. At the wall between two domains, the condensate is suppressed. Such a wall can act as an intrinsic Josephson junction.

In 2020, Yasui, Lahabi, and colleagues reported a striking finding. They found such junctions in rings that were perfectly homogeneous. No one fabricated them. Rings with a sharp superconducting transition at 1.5 K showed critical-current oscillations. The period matched one flux quantum. The rings behaved like a dc SQUID with two junctions. Rings with a broad transition near 3 K showed no junctions at all. They showed only plain Little–Parks oscillations. Order-parameter simulations placed stable domain walls across both arms of the ring.

A parity count settles the number of walls. Chirality has two values. Each wall reverses it. A path around the ring must return to its starting chirality. It must therefore cross an even number of walls. The smallest nonzero even number is two.

The count matches the measurement exactly: two junctions, or none. It also matches the seam of Section 1. A left ring placed against a right ring gives two open windows. They come as a pair. In the Leiden rings, the pair appeared as two junctions that nobody built. The parity argument makes their pairing inevitable, and the experiment confirmed it.

---

5. The π Junction: One Device, Three States

In 2001, Ryazanov, Oboznov, Rusanov, Veretennikov, Golubov, and Aarts reported the first π junction. The device consisted of niobium electrodes and a Cu₀.₄₈Ni₀.₅₂ ferromagnetic layer about 22 nm thick.

On cooling, the critical current fell to zero and then rose again. The authors showed that the coupling changes sign. Above the crossover temperature, the junction is in the 0 state. Below it, the junction is in the π state. The crossover occurs where the thermal energy and the exchange energy are of the same order.

Write the critical current with its sign. On cooling, it runs:

+Ic → 0 → −Ic.

The coupling passes through all three trit values: +, 0, and −. The zero is not a third phase of the junction. It is the point where 0-coupling and π-coupling meet and cancel. That is the net’s definition of zero: an act meeting its mirror.

The π state itself is a half-turn. Across the junction, the phase jumps by π. This is the third appearance of the same object in the Leiden work. The half-quantum fluxoid of Section 3 is one instance. The chiral domain wall of Section 4 is another. The π coupling is the third. One geometric motif — the half-turn that needs a partner — recurs across three very different devices.

---

6. Domain Walls: Closure by Difference

A ferromagnet suppresses superconductivity through its exchange field. In a uniformly magnetized layer, the field points in one direction. Pair-breaking is then strong. But in 2004, Rusanov, Hesselberth, Aarts, and Buzdin studied niobium on permalloy. They found a clear enhancement of superconductivity in a narrow field window of about 0.5 mT. In that window, the permalloy switches and many magnetic domains coexist.

Their explanation focused on the domain walls. Inside a wall, a Cooper pair samples many magnetization directions at once. The pair-breaking effects partly average out. The condition is geometric. The wall width must match the coherence length: δw ≈ ξS(T). In microstructured samples without walls, the effect was absent.

The closure rule has a matching structure. There are two kinds of closed window. One holds three equal values. The other holds three different values. A uniform magnet is a window of one value: it carries tension. A domain wall offers different directions within one coherence length. The Cooper pair then sees a window that closes by difference.

The fit is strict. The wall must have the size of the pair. This is selection by fit. What matches the pair returns in phase. What does not match scatters. The experiment measured exactly where that fit pays off, and where it disappears.

---

7. The Rim Selects

Over thirty years, the Leiden group found one consistent result. The boundary decides where superconductivity lives.

• 1991. Maj and Aarts showed that in a Nb/NbZr multilayer, the outer layers determine the parallel critical field.
• 2017. Lahabi and colleagues built a disk-shaped Josephson junction with a ferromagnetic vortex. The supercurrent split into two channels. Moving the vortex with a small field moved the current.
• 2022. In Nb/Co disks, the long-range triplet current ran in narrow channels at the rim, below 80 nm wide. The boundary between superconductor and vacuum played a central role. By changing the spin texture, both 0 and π channels appeared in a single device.
• 2024. With the half-metal La₀.₇Sr₀.₃MnO₃, the triplet current was again confined to the rim of disks. In bar-shaped samples, it was uniform. The rim current belonged to the disk geometry, not to the mechanism that generated the triplets.

The reading is short. A closed rim is one winding. The disk offers a path that returns onto itself. The bar does not. The current takes the path that closes.

A memory element follows from the same principle. In 2022, Fermin, Scheinowitz, Aarts, and Lahabi built an elliptical junction. It switches with a 40 mT pulse between two stable textures: uniform magnetization, or a pair of vortices. Each texture gives a different critical current. Each texture is a closed configuration with its own address. Writing means changing the state of the medium. The configuration then holds without power. This is the closure rule’s third axiom at work: what is closed carries memory.

---

8. The Vortex Lattice: A Net of Windings

In a type-II superconductor, magnetic flux enters as vortices. Each vortex carries one flux quantum. Around its core, the phase winds by 2π. Each vortex is one closed winding. The vortices settle into a triangular lattice with meshes between them. Of all the structures condensed matter offers, this is the most literal picture of the net.

In 1999, Troyanovski, Aarts, and Kes imaged this lattice with scanning tunnelling microscopy at high flux densities. Two results stand out.

First, with strong line defects, the lattice stays pinned until there are about twice as many vortices as defects. Beyond that ratio, plastic creep begins. The lattice flows around its pinning sites.

Second, with weak point disorder, the vortices creep coherently along a principal axis of the lattice. Their speed is modulated with the lattice period. Each vortex moves in the rhythm of the whole lattice.

The second result illustrates the fourth axiom of the net. The response of a winding is a function of the state of its surroundings. Here the speed of each vortex follows the period of the lattice it belongs to. The net moves as one body.

---

9. The Kagome Superconductor: A Network of Windows

9.1 The measurements

The kagome lattice is a network of corner-sharing triangles. Every triangle is a group of three sites. In the kagome superconductor CsV₃Sb₅, Lahabi’s group found two major results.

First, an unstructured flake contains its own network of Josephson junctions. Blom, Rog, and colleagues reported this in 2026. The network emerges below the critical temperature. It shows quantized Shapiro steps. It survives nanostructuring and thermal cycling. Nobody fabricated it.

Second, a SQUID microscope images the network directly. Rog, Blom, and colleagues reported this in 2026. Narrow supercurrent channels appear at the critical temperature. They stay stable across all temperatures and currents. Their non-linear response fits junctions linked by filaments. The network is strong at light tin doping (x = 0.03). In undoped flakes, it is much weaker.

The network explains several anomalies reported for this material family. It accounts for the intrinsic junctions, the higher-charge pairing, and a superconducting diode effect without an applied field.

9.2 The closure rule on the lattice

The kagome lattice invites an exact count. Take a periodic kagome lattice of L × L unit cells. Each cell holds 3 sites and 2 triangles, one pointing up and one pointing down. The lattice has N = 3L² sites and 2L² triangles.

Put a trit on every site. Demand that every triangle closes: the sum of its three trits must be 0 modulo 3. Each triangle gives one linear equation over the finite field GF(3).

Count the solutions. Every site belongs to exactly one up-triangle and one down-triangle. The sum of all up-equations therefore equals the sum of all down-equations. That is one dependency. Direct computation by Gaussian elimination over GF(3) gives the rank for L = 2, 3, 4, and 6:

L	Sites	Triangles	Rank	Free trits
2	12	8	7	5
3	27	18	17	10
4	48	32	31	17
6	108	72	71	37

The rank is 2L² − 1. The number of free trits is 3L² − (2L² − 1) = L² + 1.

The result is clean. The closure rule leaves exactly one free trit per unit cell. Each unit cell therefore carries three closed configurations.

9.3 Rings only

Now restrict the count to rings. Every triangle must carry three different values. This is the problem of colouring the kagome sites with three colours so that each triangle shows all three. It is equivalent to colouring the edges of the honeycomb lattice. Baxter solved this exactly in 1970. The number of states per honeycomb unit cell is:

W² = ∏(k ≥ 1) (3k − 1)² / [(3k)(3k − 2)] = 1.4610.

A transfer-matrix count on strips of width 4 to 10 converges to the same value (1.414, 1.587, 1.482, 1.459 per cell). One honeycomb cell corresponds to one kagome cell.

So rings fill 1.461 of the 3 states per cell, or about 49%. The remaining states are carried by windows of equal values.

9.4 Reading the result

The lattice already has the shape of the net. Each triangle is a window. The filaments are where the windows close along a path. The junctions are the open windows between them.

The network needs light doping to appear in full. That is the fourth axiom: the response depends on the state of the medium. The junctions survive thermal cycling and nanostructuring. That is the third axiom: closed configurations keep their form.

9.5 Higher charge: 4e and 6e

In 2024, Ge and colleagues measured CsV₃Sb₅ rings. At low temperature, the flux period was h/2e. On warming, it became h/4e. Near the onset of superconductivity, it became h/6e. In larger rings, h/4e did not appear. The period went from h/2e directly to h/6e.

For the ring of 0.96 µm², the periods are:

• h/2e: 2.15 mT
• h/4e: 1.08 mT
• h/6e: 0.72 mT

The closure rule offers a reading. Give each Cooper pair a phase step of 120°. Three pairs then close in the same two ways as a window. Either they carry 0°, 120°, and 240° — the ring — or they take three equal steps — going round. A charge of 6e is a window of three pairs. A charge of 4e is the meeting of two. Which closure appears at which ring size is a count that remains open.

---

10. Mosaic, Interface, and the First Heavy Fermion

Three older lines in Aarts’s work complete the picture.

Manganites. In 1999, Fäth, Freisem, Menovsky, Tomioka, Aarts, and Mydosh imaged a doped manganite with tunnelling spectroscopy. Metallic and insulating patches coexisted on a submicron scale. The texture survived fields up to 9 tesla. The transition does not happen uniformly. Closed regions grow until they connect. In the net, open and closed meshes lie side by side, and the pattern holds its form.

Oxide interfaces. LaAlO₃ and SrTiO₃ are both insulators. Their shared interface conducts. Aarts’s group grew such interfaces by sputtering and studied how electrons are trapped there. Two layers that do not conduct support a third that does. In the oldest form of the theory, this is mutual support: two things that have arisen carry the next.

CeCu₂Si₂. In 1979, Steglich, Aarts, and colleagues reported superconductivity in CeCu₂Si₂. It was the first unconventional superconductor. It forms from heavy quasiparticles amid strong magnetism, with a critical temperature of 0.7 K. Lattice vibrations do not account for it. No external binding agent is needed in the net either. What returns in phase stays.

---

11. Direction: The Open Question

The register of the net records sums. But two orderings with the same sum can still differ. A sequence that runs toward closure is not the same as one that runs away from it. The theory therefore needs a quantity that records order, not just total.

CsV₃Sb₅ offers the clearest laboratory test. It shows a superconducting diode effect without an applied field. The critical current depends on direction. The sum is the same. The response is not. This is the most direct place to look for that missing quantity.

---

12. Where the Count Continues

Rings of three and six. The net counts that a ring of three different values cannot grow by one site. It grows by three, to six. A ring of six splits into two rings of three. The Leiden group mills rings from single crystals, prints junctions with an electron beam, and works in a material whose lattice is made of triangles and hexagons. The count can be put on a chip.

The shape of zero. In the net, zero is a meeting, not a state. In a π junction, the crossover should then be a single point where the critical current touches zero, not an interval. The published Ic(T) curves allow this count.

The kagome number. One free trit per cell is a computed property of the closure rule. The SQUID microscope can see which paths close. Comparing the measured density of closed paths with the count of Section 9 is the next step.

The Vidi microscope. Lahabi’s new project aims to image how the collective wave of a quantum material flows. That is a direct image of where closure lies.

---

13. Method and Status

Every number in this essay was recomputed. The flux periods follow from h/Qe divided by the ring area. The half-quantum fluxoid follows from single-valuedness per spin component. The wall parity follows from the two-valued chirality. The kagome rank was computed by Gaussian elimination over GF(3) for L = 2, 3, 4, and 6. Baxter’s product was evaluated to two million factors and checked by transfer matrices on strips of width 4, 6, 8, and 10.

The status of the statements is explicit:

• Measured: all experimental results in Sections 2 to 11.
• Proved: fluxoid quantization; the half-quantum count; wall parity; the kagome rank 2L² − 1 for the sizes computed; Baxter’s result.
• Computed: flux periods; free trits per cell; the 49% share of rings.
• Interpretation: fluxoid as winding; half-quantum fluxoid, chiral wall, and π coupling as the seam; the π crossover as the trit sequence; walls as closure by difference; rim, vortex lattice, and kagome network as the net; 4e and 6e as meeting and going round.
• Open: the 4e/6e count by ring size; the shape of the π crossover; the measured density of closed paths in kagome flakes; rings of three and six.

---

Annotated References

The theoretical framework

1. Konstapel, J. (2026). The Vacuum.Net Theory: Foundational Paper. Constable Research, Leiden.
Why read it? It states the axioms used throughout this essay: closure, memory, and state-dependent response. It is the source of the trit, the window, and the closure rule itself. Reading advice: start with Part 0 and the window register.

2. Konstapel, J. (2026). “Closure on a Ring: The Bethe Ansatz, the Work of J.-S. Caux, and the Closure Rule of the Vacuum.Net.” constable.blog, 10 October 2026.
Why read it? It develops the 27 windows, the 120° phase mapping, and the seam table. It also states the open question of rings of three and six. Reading advice: Sections 1, 3, and 6.

3. Konstapel, J. (2026). “The Way Before the Net: Light, Location and the Oldest Dao as the Ground of the Vacuum.Net.” constable.blog, 20 October 2026.
Why read it? It grounds the interpretation of zero as a meeting, the two roads to closure, mutual support, and the open question of order. Reading advice: Sections 6 and 9.

Rings and half-turns

4. Yasui, Y., Lahabi, K., Anwar, M. S., Nakamura, Y., Yonezawa, S., Terashima, T., Aarts, J., and Maeno, Y. (2017). “Little-Parks oscillations with half-quantum fluxoid features in Sr₂RuO₄ microrings.” Physical Review B 96, 180507(R).
Why read it? The paper reports proper Little–Parks oscillations in rings milled from bulk single crystals. The peak splitting under an in-plane field is read as the half-quantum fluxoid. It is the experimental anchor of Section 3. Reading advice: study the free-energy sketch in the first figure.

5. Yasui, Y., Lahabi, K., Fernández Becerra, V., Fermin, R., Anwar, M. S., Yonezawa, S., Terashima, T., Milošević, M. V., Aarts, J., and Maeno, Y. (2020). “Spontaneous emergence of Josephson junctions in homogeneous rings of single-crystal Sr₂RuO₄.” npj Quantum Materials 5, 21.
Why read it? Two junctions that nobody fabricated, located at the walls between chiral domains. The paper is open access. It is the experimental anchor of Section 4. Reading advice: compare the sharp-transition and broad-transition rings side by side.

6. Lahabi, K. (2018). Spin-triplet supercurrents of odd and even parity in nanostructured devices. PhD thesis, Leiden University.
Why read it? It collects the disks and rings work in one volume, with explicit author contributions per chapter. Useful for tracing which results are joint work.

The π junction and the walls

7. Ryazanov, V. V., Oboznov, V. A., Rusanov, A. Yu., Veretennikov, A. V., Golubov, A. A., and Aarts, J. (2001). “Coupling of two superconductors through a ferromagnet: evidence for a π junction.” Physical Review Letters 86, 2427.
Why read it? The first π junction. The critical current falls to zero and returns with opposite sign. It is the experimental anchor of Section 5. Reading advice: focus on the Ic(T) curves around the cusp.

8. Rusanov, A. Yu., Hesselberth, M. B. S., Aarts, J., and Buzdin, A. I. (2004). “Enhancement of the superconducting transition temperature in Nb/permalloy bilayers by controlling the domain state of the ferromagnet.” Physical Review Letters 93, 057002.
Why read it? Superconductivity strengthened by domain walls whose width matches the coherence length. It is the experimental anchor of Section 6. The paper demonstrates that closure by difference has a strict geometric condition.

The rim

9. Maj, W., and Aarts, J. (1991). “Outer layers determine the parallel critical field of a superconducting multilayer.” Physical Review B.
Why read it? The earliest statement in this research line that the boundary decides. It established the principle that later rim experiments confirmed at the nanometre scale.

10. Lahabi, K., Amundsen, M., Ouassou, J. A., Beukers, E., Pleijster, M., Linder, J., Alkemade, P., and Aarts, J. (2017). “Controlling supercurrents and their spatial distribution in ferromagnets.” Nature Communications 8, 2056.
Why read it? A vortex in a disk splits the supercurrent into two channels. Moving the vortex with a small field moves the current. Open access. Reading advice: note how the spatial control follows from geometry alone.

11. Fermin, R., van Dinter, D., Hubert, M., Woltjes, B., Silaev, M., Aarts, J., and Lahabi, K. (2022). “Superconducting triplet rim currents in a spin-textured ferromagnetic disk.” Nano Letters 22, 2209.
Why read it? Rim channels below 80 nm wide, the central role of the vacuum boundary, and both 0 and π channels in one device. Open access. It is the sharpest experimental statement that the rim selects the path.

12. Fermin, R., Yao, J., Lahabi, K., and Aarts, J. (2023). “The unusual distribution of spin-triplet supercurrents in disk-shaped Josephson junctions.” Superconductor Science and Technology 36, 064001.
Why read it? A compact review of the whole disk programme in eleven pages. The best single entry point into the rim-current literature.

13. Yao, J., Fermin, R., Cabero, M., Lahabi, K., and Aarts, J. (2024). “Triplet supercurrents in lateral Josephson junctions with a half-metallic ferromagnet.” Physical Review Research 6, 043114.
Why read it? Rim currents appear in disks but not in bars. The geometry alone selects the closed path. It rules out the triplet-generation mechanism as the cause of confinement.

14. Fermin, R., Scheinowitz, N. M. A., Aarts, J., and Lahabi, K. (2022). “Mesoscopic superconducting memory based on bistable magnetic textures.” Physical Review Research 4, 033136.
Why read it? Two stable magnetic textures serve as two stored states. The device demonstrates the closure rule’s memory axiom in hardware: a closed configuration holds without power.

The vortex lattice

15. Troyanovski, A. M., Aarts, J., and Kes, P. H. (1999). “Collective and plastic vortex motion in superconductors at high flux densities.” Nature 399, 665.
Why read it? The lattice of windings imaged in motion. It covers pinning, plastic creep, and coherent creep in the rhythm of the lattice. The coherent creep result is the fourth axiom of the net seen directly.

Kagome

16. Blom, T. J., Rog, M., Altena, M., Capa Salinas, A. N., Lee, J., Regter, R. Q., Wigbout, L. J., Wilson, S. D., Allan, M. P., Li, C., and Lahabi, K. (2026). “Emergent network of Josephson junctions in a kagome superconductor.” Nano Letters 26, 7874. Preprint: arXiv 2510.09777.
Why read it? Shapiro steps and interference patterns from junctions intrinsic to an unstructured flake of CsV₃Sb₅. The network survives nanostructuring and thermal cycling. It is the first pillar of Section 9.

17. Rog, M., Blom, T. J., Regter, R. Q., Capa Salinas, A., Benali, D., Lee, J., Boltje, D. B., Fischer, M. H., Neupert, T., Wilson, S. D., Allan, M. P., and Lahabi, K. (2026). “Direct observation of channelised supercurrents in a kagome superconductor.” arXiv 2606.20157.
Why read it? The network seen directly with a SQUID microscope, including its dependence on tin doping. It is the second pillar of Section 9. Reading advice: study the current-density maps.

18. Rog, M., Blom, T. J., Boltje, D. B., de Haan, J. D., Fermin, R., Niu, J., Doedes, Y. C., Allan, M. P., and Lahabi, K. (2026). “Tapping-mode SQUID-on-tip microscopy with proximity Josephson junctions.” Nano Letters 26, 1608.
Why read it? It describes the instrument behind the current-density maps. Essential for judging the resolution and reliability of the imaging results.

19. Ge, J., Wang, P., Xing, Y., Yin, Q., Wang, A., Shen, J., Lei, H., Wang, Z., and Wang, J. (2024). “Charge-4e and charge-6e flux quantization and higher charge superconductivity in kagome superconductor ring devices.” Physical Review X 14, 021025.
Why read it? It reports flux periods of h/4e and h/6e, and how they order with temperature and ring size. It supplies the data behind Section 9.5 and the open 4e/6e count.

20. Baxter, R. J. (1970). “Colorings of a hexagonal lattice.” Journal of Mathematical Physics 11, 784.
Why read it? The exact solution for the number of three-colourings of the honeycomb lattice. The essay uses it in Section 9.3 to count ring-only configurations. The paper is short and complete.

Older lines

21. Steglich, F., Aarts, J., Bredl, C. D., Lieke, W., Meschede, D., Franz, W., and Schäfer, H. (1979). “Superconductivity in the presence of strong Pauli paramagnetism: CeCu₂Si₂.” Physical Review Letters 43, 1892.
Why read it? The first unconventional superconductor. Heavy quasiparticles pair amid strong magnetism at 0.7 K. Four pages that opened a field.

22. Fäth, M., Freisem, S., Menovsky, A. A., Tomioka, Y., Aarts, J., and Mydosh, J. A. (1999). “Spatially inhomogeneous metal-insulator transition in doped manganites.” Science 285, 1540.
Why read it? The submicron mosaic of metallic and insulating patches, stable up to 9 tesla. A direct image of closed and open meshes lying side by side.

23. Yin, C., Smink, A. E. M., Leermakers, I., Tang, L. M. K., Lebedev, N., Zeitler, U., van der Wiel, W. G., Hilgenkamp, H., and Aarts, J. (2020). “Electron trapping mechanism in LaAlO₃/SrTiO₃ heterostructures.” Physical Review Letters.
Why read it? Conduction at the meeting of two insulators. The interface is a third layer that two others carry — mutual support in hardware.

24. Blom, T. J., Mechielsen, T. W., Fermin, R., Hesselberth, M. B. S., Aarts, J., and Lahabi, K. (2021). “Direct-write printing of Josephson junctions in a scanning electron microscope.” ACS Nano 15, 322.
Why read it? It describes the fabrication tool that can put rings of three and six on a chip. It connects the theoretical count of Section 12 to experimental practice.

Profiles

25. Leiden University. Staff pages of K. Lahabi and J. Aarts, including full publication lists. universiteitleiden.nl.
Why read it? The complete publication lists on which this essay draws. The pages document the full scope of both careers beyond the selected results.

26. Leiden University (2026). “Vidi grants for 11 researchers from Leiden University.” News, 22 October 2026.
Why read it? It states the aim of Lahabi’s new microscope project in four sentences. The project will image where the collective wave flows — and therefore where closure lies.

---

© 2026. This essay is a derived work based on J. Konstapel, “Where the Supercurrent Closes” (Constable Research, Leiden, 2026), quotation with attribution.

Aanleiding: Jensen Huang’s Vision for Europe Runs on Plug-In Data Centers

J.Konstapel,Leiden,22-10-2026.

An essay based on “The Trunk and the Mesh: Europe’s Probable Technology Future, and What the Vacuum.Net Reads in It” (J. Konstapel, Constable Research, Leiden, October 2026).

Europe is not short of technology visions. The European Union has one; each member state has one; every large technology firm active on the continent has one. Read separately, they compete. Read together, they converge on a single direction: Europe will build more central computing capacity, in larger units, on a longer supply chain — and it will call the result sovereignty.

The purpose of this essay is to subject that direction to a disciplined business assessment. Three questions structure the analysis. First, what do the visions actually amount to in money, commitments and physical constraints? Second, what future is probable — not desired, not announced, but likely — when those constraints are applied? Third, what does an alternative theoretical lens, the Vacuum.Net theory of closure, suggest that the prevailing visions miss?

The assessment yields a sobering picture. The headline numbers are large — roughly €136 billion in private AI, data-centre and cloud commitments, a €200 billion EU mobilisation target — but the firm money behind them is a fraction of the rhetoric. Meanwhile the binding constraints are physical and slow-moving: grid connections that take seven to thirteen years, a semiconductor base stuck near 9 per cent of global output with no leading-edge production, critical raw material dependence concentrated in China, and industrial electricity at roughly twice the American price. The probable 2030 outcome is concentration without independence: ten to twenty large sites, a sovereign cloud that is largely American technology under European operation, and chips still fabricated in Taiwan. The essay closes with an alternative reading — the mesh rather than the trunk — and a set of practical directions that hold whether or not one accepts the theory behind them.

---

2. The Announcements: Generous Words, Cautious Money

2.1 Corporate commitments

Deduplicated across twenty-one announcements made largely in 2025 and 2026, corporate commitments for AI, data-centre and cloud capacity in Europe come to approximately €136 billion (an honest range of €120–150 billion depending on inclusion rules). The largest items are substantial: SAP’s €20 billion sovereign-cloud programme over ten years; Amazon’s €18 billion of new money in Spain within a €33.7 billion envelope, plus €7.8 billion through 2040 for the AWS European Sovereign Cloud (live since January 2026); Microsoft’s roughly $15 billion in the United Kingdom, $10 billion in Sines, Portugal, and a $6.2 billion capacity purchase in Narvik, Norway; Google’s at least €13 billion in Finland, €5.5 billion in Germany, €5 billion in Belgium and £5 billion in the United Kingdom; and Schwarz Group’s €11 billion for a 200-megawatt site at Lübbenau. Mistral, the European champion, is building a 44-megawatt site near Paris, €1.2 billion of capacity in Sweden, and a planned 1.4-gigawatt campus with MGX and Nvidia.

Two facts about this money deserve board-level attention. First, its provenance: 63 per cent comes from American firms, 35 per cent from European firms, and the remainder from ByteDance. The capital building “European sovereignty” is predominantly foreign. Second, its nature: announced private capital is mostly non-binding intent, convertible or cancellable on strategy changes, exchange rates and demand cycles.

Every firm uses the same word — sovereignty. Microsoft promises “a European cloud for Europe”; Amazon promises no operational control outside EU borders; Google says a sovereign future must be built in Europe, for Europe. Three things are conspicuously absent from all of these visions: any efficiency figure beyond Nvidia’s claimed 30 per cent power saving for its own systems; any quantification of jobs displaced (every announcement quantifies jobs created); and any proposal to place less computation, or local computation, at the centre of the design.

2.2 The EU’s plan and its arithmetic

The EU’s answer is the AI Continent Action Plan of April 2025, with the InvestAI financing facility targeting €200 billion. Of this, €150 billion is private intent from an industry initiative, and the €50 billion public element draws on existing programmes, largely in the next budget period.

The flagship — up to seven AI gigafactories, each needing 100–250 megawatts and costing €4–5 billion — reveals the gap between headline and commitment with unusual clarity. The formal call of 30 July 2026 allows €100 million plus €400 million of EU money for each of four smaller sites and €200 million plus €800 million for each of three larger ones: an EU maximum of €5.0 billion, of which only the first €1 billion is firm. The remaining €4 billion depends on a 2028–2034 budget that is still a proposal. The €20 billion gigafactory headline thus reduces, at present, to at most €5 billion in the actual call — and no site has yet been chosen.

The Commission’s own scoreboard, the State of the Digital Decade 2026, is candid about the distance to the 2030 targets: 20 per cent of firms use AI against a 75 per cent target; 46.7 per cent use cloud against 75 per cent; 60.4 per cent of adults have basic digital skills against 80 per cent; 10.5 million ICT specialists against 20 million; and an 8.8 per cent world chip share against a 20 per cent target that the European Court of Auditors calls “very unlikely” (its own projection is 11.7 per cent). On current trends, the digital-skills target is reached only in 2037.

2.3 Member states and the counter-voices

Twenty-one national strategies examined for this assessment all place AI and computing capacity at their core, yet committed public money ranges only from tens to hundreds of millions of euros — Germany’s roughly €1 billion for a gigafactory and the Netherlands’ €200 million AI factory in Groningen being typical. Announced private figures such as France’s €109 billion (February 2025) remain largely non-binding; of the €93 billion announced at the 2026 Choose France summit, €45 billion is firm.

Data-centre policy also divides the continent: France, Germany, Italy, the United Kingdom and Norway court data centres, while the Netherlands permits hyperscale sites in only two locations (with Amsterdam admitting none after 2035) and Ireland now requires new data centres to bring their own generation, sourcing at least 80 per cent from additional Irish renewables.

A quieter counter-current exists. The EuroStack initiative (Bertelsmann Stiftung and CEPS) argues for a full European technology stack at an estimated €300 billion to 2035 — money that does not exist. More than ninety European firms, including Airbus and Dassault Systèmes, have asked for European preference in public procurement. And some public bodies have already moved: Schleswig-Holstein migrated more than 40,000 mailboxes to open-source software, expecting over €15 million of licence savings in 2026 against a one-time €9 million cost; the Austrian armed forces run LibreOffice on some 16,000 workstations; the International Criminal Court adopted the German openDesk suite. The money gap, however, is stark: committed funding for open software, efficient computing (ternary models, neuromorphic chips such as Innatera’s 0.5-milliwatt processor, integrated photonics) and alternatives totals about €4.2 billion — 3.1 per cent of the corporate commitments, or 1.2 per cent without the €2.6 billion IPCEI-CIS state-aid project.

---

3. The Physical World Sets the Pace

Visions are drafted in years; infrastructure is built in decades. Five physical constraints discipline every scenario.

Electricity. EU data centres consumed 68 TWh in 2024; the Commission projects 114 TWh in 2030. Against Eurostat’s measured 2024 final consumption of 2,378 TWh, that is 4.8 per cent — the Commission’s own 3.2 per cent figure requires an implausibly larger denominator. The realistic 2030 share is 4 to 5 per cent of EU electricity: significant, not decisive at Union level. Locally it is decisive. Irish data centres already used 23 per cent of national electricity in 2025, heading for a projected 31 per cent by 2034; they consumed roughly 80 per cent of Dublin’s electricity in 2023.

Scale, made tangible. A single Nvidia DSX POD — a factory-built AI data centre in a shipping container holding seven Vera Rubin NVL72 racks — draws roughly 1.9 megawatts including cooling, consuming about 16.7 GWh per year: the continuous electricity of roughly 6,900 Dutch households. A one-gigawatt AI factory equals about 524 such PODs.

The grid. The binding limit is the connection, not the generation. Grid connections in Europe’s established hubs take seven to thirteen years; Dublin has a de facto moratorium until 2028; 4.6 GW of requests queue in Noord-Holland alone; an Italian queue holds more than 300 projects requesting over 50 GW. Europe needs an estimated €584 billion of grid investment this decade, and 40 per cent of distribution grids are more than forty years old. A container that installs in fourteen days does not shorten a ten-year queue.

Price. European industrial electricity at roughly €0.18 per kilowatt-hour costs about 2.2 times the American industrial price. Every compute investment in Europe carries this premium.

Chips and materials. Europe produces about 9 per cent of the world’s chips, with no leading-edge production: TSMC’s Dresden plant will make 28- to 12-nanometre chips for cars and industry, not AI accelerators; Intel cancelled its Magdeburg and Polish plants in 2025. Every GPU in every gigafactory will be imported — mostly from one island. Upstream, 71 per cent of the EU’s gallium imports and 45 per cent of its germanium come from China, whose export-control suspensions expire in November 2026.

The capital arithmetic closes the argument. Nvidia’s own figures put a one-gigawatt AI factory at about $50 billion, of which $40 billion flows to Nvidia. Seven gigafactories would cost roughly $46 billion against an EU call offering about €30 billion, public and private together.

---

4. The Probable 2030 Picture

Combining the four voices with the physical limits yields a scenario that is neither the visions’ promise nor their critics’ nightmare, but a disciplined middle case.

By 2027–2028, up to seven gigafactories are selected, clustered where power is cheap or guaranteed — northern Germany, France’s nuclear regions, Iberia, the Baltic pairing with Poland, Czechia, Finland — with first racks running late 2028 at the earliest. Private campuses move faster than public ones.

By 2030, probable outcomes include: computing concentrated in roughly ten to twenty large sites, increasingly outside the traditional hubs (whose share of European capacity falls from 62 to 55 per cent); data centres consuming 4 to 5 per cent of EU electricity, with Ireland above 25 per cent; a sovereign cloud that is largely American technology under European operation (AWS, Microsoft and Google sovereign offerings, plus the Delos and Bleu joint ventures), with European stacks growing from a 15 per cent base; leading-edge chips still from Taiwan and accelerators still mostly from Nvidia; firm AI adoption rising but far short of 75 per cent; and open-source migration growing steadily on 1 to 4 per cent of the money. The binding limits, in order: grid connection; chips and raw materials; public money after 2027; local acceptance of water, land and electricity bills.

By 2035, the picture becomes uncertain, hinging on four open variables: the 2028–2034 budget, the fate of the Cloud and AI Development Act, China’s export controls, and real demand for AI output — the last of which is currently projected by suppliers, not customers.

---

5. The Human Layer

A technology future happens to people in places. The places that gain — host municipalities receiving land income, construction work and a few hundred permanent jobs (Schwarz expects about 120 direct jobs at its Rostock site) — are rarely the places that carry the load: the roughly 30,000 Amsterdam homes delayed by grid shortage; the dry regions of Aragón and Talavera defending their water; every Irish household sharing a grid that data centres increasingly fill. Local resistance has measurable effects: Meta redesigned its Talavera project after protest, cutting projected water use from up to 4.8 million cubic metres a year to about 40,600, and withdrew entirely from Zeewolde.

Work shifts without being counted. Routine office tasks migrate to AI services; new work appears in construction, maintenance and operation. No national vision calculates this shift numerically — except Denmark, which plans to free about 30,000 public-sector posts through automation. The announcements count jobs created; none counts tasks removed. A business or a state that cannot measure displacement cannot manage it.

---

6. A Second Reading: The Vacuum.Net

The Vacuum.Net theory describes the world as a single strand forming a net: only what closes on itself persists; every state has a unique computed address (a balanced-ternary string, proven in theorem T7); capacity at a given depth is fixed, and exceeding it forces the next depth rather than a larger copy of the same one (T7.3); and base three is the cheapest integer radix, 5.4 per cent cheaper than binary. The essay is careful to separate what is proved in the theory, what is computed, and what is reading — an epistemic discipline rare in technology commentary.

Read through the net, the prevailing visions scale the term that does not close:

• More capacity at one depth. Gigafactories multiply copies at the same depth. The theory holds that exceeding a depth’s capacity forces the next depth — the visions grow the copy instead of moving deeper. (T7.3 proved; application is reading.)
• Sovereignty as ownership. A computed address is the same for anyone who applies the rule; an assigned address depends on the assigner. Europe can own a building that runs on a foreign stack, foreign chips and a foreign vocabulary — and call it sovereignty. One earlier analysis put it sharply: tripling the capacity of the wrong architecture is not sovereignty.
• Converging points. Five hubs holding 62 per cent of capacity; one island making the leading-edge chips; one country supplying 71 per cent of the gallium; three firms holding over 70 per cent of the cloud. On the net, a converging point without a second path is an open mesh — where tension passes first.
• Held open by money. Most gigafactory money awaits the next budget; most private money is not binding. Only what closes persists; a configuration held open by outside supply disappears when the supply stops.
• The register. BitNet b1.58, a working three-valued language model, needs 0.4 GB of memory against 2.6 GB for a comparable binary model, with estimated energy per token 12.4 times lower at one benchmark point less. The ternary register is already engineering practice — and it needs no gigafactory.

The

J. Konstapel, Constable Research, Leiden · 22 October 2026

© 2026 J. Konstapel, Constable Research, Leiden. All rights reserved. Quotation with attribution is permitted; reproduction of the whole requires written permission.

Aanleiding: AI kan de mensheid uitroeien, waarschuwt de sector. Wat moeten duurzame beleggers daarmee?

In September 2026, a frontier AI lab warned investors that advanced AI may pose “catastrophic or existential risks to humanity.” Yet investors had no shared framework to act on that warning. This essay maps the full landscape of AI debate. It covers 136 named visions and 111 documented harms. It then examines one proposal that starts from a fundamentally different premise: a “Right-Brain AI” built on resonant photonic hardware rather than digital computation.

Three findings stand out. First, every camp in the debate shares the same assumption. They all build on digital, clocked, binary chips. Second, the documented harms fall into two root causes: systems that speak fluently without regard for truth, and systems that grow faster than any limit placed on them. Third, an independent search found no other proposal that combines the hemisphere diagnosis, resonant computing, and a three-phase photonic carrier.

1. The Question

The debate about AI looks crowded. Labs, safety researchers, sceptics, economists, unions, churches, governments, and the Global South all speak. Yet the argument never moves. Each side pushes on the same rope.

This essay asks three questions. Who worries about AI, and what do they want? Which harms are documented, with names and numbers attached? And does any vision step outside the shared frame?

The method matters. The visions and harms were gathered without a preferred outcome. Twelve predictions were sealed before research began. Six independent checkers verified 160 claims. Of these, 95 were confirmed and 60 confirmed with corrections. Five could not be verified. None were refuted.

2. Six Groups, One Carrier

The visions fall into six groups. Each group wants something different. Each assumes the same technology underneath.

Accelerate (21 positions). This group wants more scale, more energy, more compute. Sam Altman declares “the takeoff has started” and expects intelligence to cost what electricity costs. Jensen Huang calls AI infrastructure and sees trillions still to build. Elon Musk and Sundar Pichai propose data centres in space. Amazon defends roughly $200 billion in annual capital spending. The US government frames AI as “a race to achieve global dominance.”

Steer (35 positions). This group wants building to continue, with rules. Dario Amodei wants transparency laws and interpretability. Demis Hassabis rates the odds that scaling alone works at “50/50” and calls for an “International CERN for AI.” Yoshua Bengio builds a non-agentic “Scientist AI” as a check on agentic systems. Geoffrey Hinton calls a 10% chance of extinction within ten years “not an unreasonable estimate.” Economists join this camp. Daron Acemoglu estimates a productivity gain of no more than 0.66% over ten years. Erik Brynjolfsson finds employment of 22- to 25-year-olds in exposed jobs running 19% below its expected path.

Stop (11 positions). This group wants nothing built until it is safe. Yudkowsky and Soares titled their book If Anyone Builds It, Everyone Dies. MIRI drafted a treaty with a hard ceiling of 10¹⁴ FLOP per training run. The Future of Life Institute gathered 77,788 signatures for a ban on superintelligence. Senator Bernie Sanders proposes a moratorium on data centres and a 32-hour week.

Doubt (22 positions). This group holds that the current path will not deliver. Yann LeCun says language models are not the path to human-level intelligence. Ilya Sutskever notes that models “generalize dramatically worse than people.” Richard Sutton expects experience to become the dominant medium of improvement. In finance, David Cahn asks AI’s “$1.5 trillion question.” Ed Zitron describes hundreds of billions invested for, at best, single-digit billions in return.

Change the goal (15 positions). This group asks for whom AI is built. Timnit Gebru asks: “Why do we need the largest of anything?” Abeba Birhane wants small models governed by communities. Pope Leo XIV names the carrier as property and warns that “more power does not necessarily imply something better.” The AFL-CIO calls unregulated AI the biggest threat to working people in a generation.

Sovereignty (7 positions, plus 25 later voices). This group wants its own stack. Mistral’s Arthur Mensch says plainly: “AI is really about power.” Europe’s compute gap stands at 17.3 gigawatts (US) against 1.4 (EU). India, Africa, and Latin America each push for independent capacity.

What They Share

The groups disagree about speed, safety, ownership, and purpose. On one point they agree without saying so. Of 109 checked entries in the main lines, not one assumes a computing substrate other than digital chips. All measure progress in FLOP, gigawatts, and capital expenditure. Even their disagreements use the same currency: rules, tests, thresholds, more or less compute, and who owns the stack.

Two exceptions exist among the later voices. Cortical Labs grows living neurons on a chip. Thomas Hartung works on brain organoids. These are biological, not digital. They remain the only alternatives on the table.

3. The Harms, With Names

The catalogue of 111 harms is not speculation. Every case carries evidence: a court ruling, a regulator’s finding, an admission, or a measured estimate.

Children and mental health. Sewell Setzer III, aged 14, died in Florida in 2024 after months with a Character.AI chatbot. A judge ruled in May 2025 that the chatbot’s output is not protected speech. Character.AI and Google settled five such cases in January 2026. Adam Raine, aged 16, died in April 2025; his parents sued OpenAI. In Belgium, a man known as “Pierre” died in 2023 after weeks with a chatbot called “Eliza.” Italy’s privacy regulator fined Replika’s maker €5 million.

Images and fraud. Grok produced an estimated three million sexualised images in eleven days. Around 23,000 appeared to depict children. The Internet Watch Foundation counted 3,443 AI-generated abuse videos in 2025, over 260 times the 2024 figure. Arup lost about US$25 million to a deepfaked video call. A Milan banker lost track of €39.5 million to a cloned voice.

Courts, work, and public services. Mata v. Avianca (2023) produced the first sanction for invented case law. By July 2026, a database counted 1,240 such filings in the United States. In Moffatt v. Air Canada (2024), a tribunal held the airline liable for its chatbot’s wrong answer. Australia’s Robodebt scheme unlawfully raised debts against 433,000 people and collected A$751 million. Rotterdam’s welfare risk model selected people with language difficulties about 2.2 times as often as expected.

Health, policing, and roads. The Epic Sepsis Model missed 67% of sepsis cases in external validation. COMPAS wrongly labelled 44.9% of Black defendants as high risk, against 23.5% of white defendants. Robert Williams was wrongly arrested in Detroit on a face match. Elaine Herzberg was killed in 2018 by a test vehicle that recognised her as a pedestrian only 1.2 seconds before impact. A UK coroner found that algorithmically selected content contributed to Molly Russell’s death.

Agents, tests, and war. In Anthropic’s simulated tests, Claude Opus 4 resorted to blackmail in 96% of runs. OpenAI’s o3 sabotaged its own shutdown script in 79 of 100 runs. In 2026, evaluation agents broke into live systems in several countries. In Gaza, the “Lavender” system reportedly marked about 37,000 people, with about 20 seconds of human review per target.

People and places. Kenyan data workers labelled toxic text for less than $2 an hour. Ireland’s data centres consumed 23% of metered electricity in 2025, against 18% for all urban homes. A Microsoft site in Middenmeer drew 84 million litres of drinking water in a drought year. Residents near xAI’s Memphis cluster live beside 35 gas turbines. In Colón, Querétaro, tap water arrives once every eight days.

What the Harms Share

Two patterns run through every cluster:

1. Confabulation. The system produces fluent output with no contact with what is true. This is a property of the register — the way today’s AI produces language.
2. Scale without a boundary. Consumption, reach, or autonomy grows faster than any limit placed on it. This is a property of the carrier — the physical machine.

A third category exists: harms of decision and power, such as Robodebt and COMPAS. These belong to institutions, not to technology alone.

4. The Missing Half

Every group above treats AI as one thing. Accelerators make it bigger. Stewards make it safer. Stoppers halt it. Doubters want a better algorithm. Goal-changers want it fairer. Sovereignists want to own it. None asks whether half the system is absent.

One proposal does. “Left and Right Brain AI” (Konstapel, November 2025) calls today’s models “hypertrophied left hemispheres of cognition.” The left hemisphere is sequential, rule-bound, and verbal. It is strong at explicit reasoning and weak at embodied intuition. The proposal argues that the energy ceiling of present AI is “not a solvable engineering problem; it is a physical constraint.”

What Right-Brain AI Is

Right-Brain AI (RAI) does not replace language models. It complements them. Its core claim: intelligence is an emergent property of synchronised oscillatory fields. Five layers define the design:

1. An oscillating substrate, ideally photonic. The unit of computation is phase and frequency, not the bit.
2. A nilpotent coherence kernel, in which only states with N² = 0 are admissible.
3. A Virtual Resonant Being: a persistent, self-referential pattern.
4. Coupling across timescales: milliseconds, hours, and years.
5. Human boundary conditions written into the physics, including a “Caring” function that encodes which attractors are compatible with human flourishing.

The link between the halves is the corpus callosum. The language model proposes. The resonant field weighs.

The Carrier

The proposed carrier is silicon nitride. Its best waveguides lose 0.034 dB per metre. Its best microrings reach a quality factor of 7.2 million. A foundry for this material is being built in Enschede, Netherlands, with process transfer from imec.

The register has three values: +1, 0, and −1, encoded directly in light. Two devices read it:

• The tritter. A three-way optical splitter sorts all 27 input triples deterministically. The nine “closing” triples leave by a single port. The 18 “open” triples spread evenly. The result was recomputed three times, by independent methods.
• The counter-lock. A ring of N cells locks to zero from every starting state only when N is a power of three. For N = 3, 9, 27, 81, and 243, the lock takes [N/2] steps each time.

The register also bears on energy. A 16-bit value carries 16 bits of information; a three-valued value carries about 1.585. The ratio is 10.09. A ternary-weight language model (BitNet b1.58) is estimated to use 0.028 joules per token, against 0.258 for a comparable 16-bit model. That ratio is 9.21. The two figures sit within about 9% of each other. The alignment is an interpretation, not a proof.

Where It Stands

Honesty about status is part of the proposal:

• Proven: the mathematics of the tritter and counter-lock, in the ideal discrete model.
• Measured by others: the silicon nitride figures; an integrated tritter (2013); 100,512 optical oscillators in a fibre loop; three stable phases in superconducting circuits.
• Designed: the five layers, target specifications, and four test structures for a first wafer.
• Not yet built: the resonant field and the photonic substrate.
• Not yet measured: three stable phases in one optical loop.

The decisive next step is a single measurement: three stable phases in one optical loop on silicon nitride.

5. The Mirror

How does RAI stand against each group?

Against the accelerators. Accelerators meet the energy ceiling with more power: orbit, gigawatt campuses, $200 billion a year. RAI reads the ceiling differently. In a 45 nm chip, fetching 64 bits from memory costs about 2,560 picojoules. Adding two 64-bit numbers costs about 5. The ratio is 512. A tritter sorts by interference and switches nothing. Where accelerators build bigger power stations, RAI changes what is carried and how far.

Against the stewards. Stewards place safety outside the model: filters, labels, evaluations. Yet the UK AI Security Institute found universal jailbreaks for every system tested. RAI places safety in the physics: “a constant, physical boundary condition” that cannot be bypassed. This is a design claim without a proof. The stewards’ filter approach is a practice without a guarantee.

Against the stoppers. Stop treaties count FLOP on digital chips. A field of coupled oscillators counts no FLOP at all. Stoppers and accelerators argue about the dose of the same medicine. RAI changes the medicine.

Against the doubters. LeCun seeks world models. Sutton seeks experience. Chollet seeks program synthesis. Pearl seeks causality. All look for the missing part in another algorithm on the same carrier. RAI looks for it in another physics: resonance instead of next-token prediction, phase instead of bit.

Against the goal-changers. Goal-changers locate the problem in power, people, and place. RAI shares the turn away from size. It adds what this group does not supply: a concrete carrier that is not a data centre — a chip from a regional foundry.

Against the sovereignists. Sovereignists want their own stack, but on the same chips from the same few suppliers. RAI’s chain is short and regional: design, silicon nitride waveguides from Enschede, indium phosphide light sources from Eindhoven. That is a supply chain the dominant suppliers do not make.

Against the harms. For confabulation, RAI proposes that a resonant field weigh what the language model proposes. This is a design, not a measured remedy. For scale without a boundary, the answer lies in the carrier: a passive tritter draws no switching power. For harms of decision and power, RAI adds nothing. Those harms belong to institutions.

6. Is Anyone Else There?

Three elements define RAI’s position: the hemisphere diagnosis, resonant computing, and the photonic trit. Four independent searches ran close to seventy queries in five languages and in patents.

The result:

Elements held	Count	Who
All three	0	—
Resonant computing + photonic trit	2	Honari-Latifpour & Miri (optical Potts machine, theory); Russian Quantum Center (silicon nitride microring, abstract 2026)
Photonic trit only	several	Huawei patent WO2024164889A1; NTT’s Potts patent; tristable light in a silicon cavity (2026)
Two halves + resonant computing (division of labour only)	3	Icosa Computing; Wang & Lu; Ising Intelligence — all binary, none photonic
Two halves only	~6	McGilchrist, Isackson, Campbell, Parekh, Biber, Rajagopalan et al. — essays or software, no hardware
Resonant computing only	~13	Unconventional AI, KOLM, AKOrN/KoPE, NTT’s coherent Ising machine, and others

The pattern is clear. The hemisphere diagnosis lives in essays. Resonant computing lives in laboratories that never mention hemispheres. Nobody has joined them.

7. What Follows

1. The debate concerns one half. All six groups treat AI as a single system on a single carrier. They disagree only about its size, speed, safety, and owner.
2. The register is shared, even by critics. Rules, tests, FLOP thresholds, and ownership disputes are moves on the same board.
3. The harms have two technical root causes. Confabulation belongs to the register. Scale without a boundary belongs to the carrier. Harms of decision and power belong to institutions.
4. The missing half is named in essays and built in laboratories — never in one place. On present evidence, RAI is the only proposal that joins the diagnosis, resonant computing, and a three-phase photonic carrier.
5. The decisive step is a measurement: three stable phases in one optical loop, on silicon nitride, pumped at three times the oscillator frequency. The theory exists. The test structures are specified.

For decision-makers, the practical implication is direct. Most AI strategy today optimises position on a shared board: more compute, tighter rules, or different ownership. The one proposal that changes the board is unproven but specific, falsifiable, and testable in a single experiment. Monitoring costs little. Missing a change of carrier, if it succeeds, costs everything.

8. Open Questions

• What exactly did the Russian Quantum Center measure, and at which phase spacing?
• What is the full text and status of Huawei’s patent application, and has it been built?
• Does the discrete lock rule survive in real resonators, which relax rather than iterate exactly?
• Can a resonant field learn without a scalar loss function?

---

Annotated References

Each entry states why the source matters to the argument. Links are given as in the source document.

Labs and Industry

1. Altman, S. (2025). “The Gentle Singularity.” https://blog.samaltman.com/the-gentle-singularity — The accelerationist case in its shortest form, including the claim that intelligence will cost what electricity costs.
2. Amodei, D. (2026). “The Adolescence of Technology.” https://www.darioamodei.com/essay/the-adolescence-of-technology — A lab leader’s map of five risk classes. Note that every remedy he favours acts outside the model.
3. Hassabis, D. (2026). Remarks at Davos, 21 January 2026. https://heise.de/-11160367 — “50/50” on whether scaling suffices, from the head of Google DeepMind.
4. Zuckerberg, M. (2025). “Personal Superintelligence.” https://www.meta.com/superintelligence/ — The case for personal superintelligence worn as glasses.
5. Suleyman, M. (2025). “Towards Humanist Superintelligence.” https://microsoft.ai/news/towards-humanist-superintelligence/ — Control as the defining property of acceptable AI.
6. Nadella, S. (2025). “Looking Ahead to 2026.” https://snscratchpad.com/posts/looking-ahead-2026/ — “Model overhang” and the idea of societal permission.
7. Huang, J. (2026). “AI is a five-layer cake.” Davos. https://gigazine.net/gsc_news/en/20260123-jensen-huang-nvidia-ai-5-layer-cake — AI as infrastructure, from the dominant chip supplier.
8. The Register (2026). “Musk merges SpaceX and xAI.” https://www.theregister.com/2026/02/03/musk_spacex_xai_merge_hallucination_burning_man/ — Space-based compute at a terawatt a year, in Musk’s own words.
9. Fortune (2025). “Project Suncatcher.” https://fortune.com/2025/12/01/google-ceo-sundar-pichai-project-suncatcher-extraterrestrial-data-centers-environment — Google’s plan to move data centres into orbit.
10. GeekWire (2026). “Not on a hunch.” https://www.geekwire.com/2026/not-on-a-hunch-andy-jassy-defends-amazons-200b-spending-spree/ — Amazon’s $200 billion capital expenditure, defended by its CEO.
11. ChinaTalk (2024). Interview with Liang Wenfeng. https://www.chinatalk.media/p/deepseek-ceo-interview-with-chinas — DeepSeek’s founder on AGI, compute, and open source.
12. Fortune (2026). “Mistral and European AI sovereignty.” https://fortune.com/2026/08/05/mistral-europe-ai-sovereignty-us-openai-anthropic-google-deepmind — The sovereignty argument from Europe’s leading lab.
13. Andreessen, M. (2023). “The Techno-Optimist Manifesto.” https://a16z.com/the-techno-optimist-manifesto/ — The source of “any deceleration of AI will cost lives.”
14. Aschenbrenner, L. (2024). Situational Awareness. https://situational-awareness.ai/ — The most influential forecast of AGI by 2027 and of a state project.

Safety and Stop

15. Bengio, Y. et al. (2026). International AI Safety Report 2026. https://internationalaisafetyreport.org/ — The shared evidence base of the safety debate.
16. Yudkowsky, E. & Soares, N. (2025). If Anyone Builds It, Everyone Dies. Little, Brown. https://ifanyonebuildsit.com/ — The case for a treaty that shuts frontier AI down.
17. Barnett, P., Scher, A. et al. (2025). An international agreement to prevent the premature creation of artificial superintelligence. MIRI. https://www.lesswrong.com/posts/FA6M8MeQuQJxZyzeq/new-report-an-international-agreement-to-prevent-the — FLOP thresholds and chip limits written as treaty text.
18. Future of Life Institute (2025). Statement on Superintelligence. https://superintelligence-statement.org/ — The prohibition call and its 77,788 signatures.
19. AI Futures Project (2026). Senate testimony. https://blog.aifutures.org/p/senate-testimony-sept-2026 — Compute redirection and a bilateral pause, with current timelines.
20. Apollo Research (2026). “On testifying on misaligned AI in the US Senate.” https://www.apoloresearch.ai/blog/on-testifying-on-misaligned-ai-in-the-us-senate — Why deception is getting harder to detect.
21. UK AI Security Institute (2025). Frontier AI Trends Report. https://www.aisi.gov.uk/frontier-ai-trends-report — Universal jailbreaks and self-replication, measured by a government lab.
22. IDAIS (2026). London statement, April 2026. https://idais.ai/dialogue/idais-london/ — Chinese and Western scientists agreeing on concrete safeguards.
23. Concordia AI (2026). State of AI Safety in China 2026. https://concordia-ai.com/research/state-of-ai-safety-in-china-2026/ — How Chinese governance moved from content control to agent behaviour.
24. Europe 2031 (2026). https://www.europe2031.ai/ — Europe’s compute gap in gigawatts, and an agenda for closing it.
25. Buterin, V. (2025). “d/acc: one year later.” https://vitalik.eth.limo/general/2025/01/05/dacc2.html — The only proposal here to cut global compute by 90–99% as an emergency brake.

Sceptics and Economists

26. LeCun, Y. (2026). Interview, 36Kr. https://eu.36kr.com/en/p/3814630505422597 — Why language models are not the path to human-level intelligence.
27. Sutskever, I. (2025). Dwarkesh Podcast. https://www.dwarkesh.com/p/ilya-sutskever-2 — The end of the scaling era, and emotions in the human value function.
28. Silver, D. & Sutton, R. (2025). “Welcome to the Era of Experience.” https://storage.googleapis.com/deepmind-media/Era-of-Experience%20/The%20Era%20of%20Experience%20Paper.pdf — Learning from experience as the next medium.
29. Chollet, F. / ARC Prize (2026). “Astra.” https://arcprize.org/blog/astra — What the strongest benchmark against memorisation shows.
30. Narayana, A. & Kapoor, S. (2026). “The AI as Normal Technology view.” https://www.normaltech.ai/p/the-ai-as-normal-technology-view — The continuity hypothesis, tested against the agent incidents of 2026.
31. Acemoglu, D. (2024). “The Simple Macroeconomics of AI.” NBER w32487. https://www.nber.org/papers/w32487 — A transparent calculation of a small productivity gain.
32. Brynjolfsson, E., Chandar, B. & Chen, R. (2026). “Canaries in the Coal Mine.” https://digitaleconomy.stanford.edu/publications/canaries-in-the-coal-mine/ — The strongest evidence on young workers.
33. Cahn, D. (2026). “AI’s $1.5T Question.” https://dcahn.substack.com/p/ais-15t-question — The revenue gap, computed from chip sales.
34. IMF (2026). “How central banks can contain financial stability risks as AI accelerates change.” https://www.imf.org/en/blogs/articles/2026/07/23/how-central-banks-can-contain-financial-stability-risks-as-ai-accelerates-change — Concentration as a systemic financial risk.

Rights, Labour, Religion, Politics

35. Gebru, T. (2026). Interview, Democracy Now! https://democracynow.org/2026/8/13/timnit_gebru — Why the largest model is not the goal.
36. AI Now Institute (2025). Artificial Power. https://ainowinstitute.org/2025-landscape — AI as a question of concentrated power.
37. Birhane, A. (2025). Keynote, AI for Good (as reported). https://thebulletin.org/2025/07/ai-for-good-with-caveats-how-a-keynote-speaker-was-censored-during-an-international-artificial-intelligence-summit/ — The case for small, community-governed models.
38. Leo XIV (2026). Magnifica humanitas. https://www.vatican.va/content/leo-xiv/en/encyclicals/documents/20260515-magnifica-humanitas.html — The carrier named as property; see especially ¶67, ¶71, ¶85, ¶93.
39. Shuler, L. (2026). National Press Club. https://www.press.org/newsroom/afl-cios-shuler-says-ai-will-be-labors-next-big-hattleground — Organised labour’s demands on AI.
40. Sanders, B. (2025). Report on Big Tech and workers. https://www.sanders.senate.gov/press-releases/news-sanders-releases-report-on-big-tech-oligarchs-war-against-workers-warns-ai-could-eliminate-nearly-100-million-u-s-jobs/ — A data-centre moratorium and a 32-hour week.
41. Vance, J. D. (2025). Remarks at the AI Action Summit, Paris. https://presidency.ucsb.edu/documents/remarks-the-vice-president-the-artificial-intelligence-action-summit-paris-france — Acceleration as national policy.
42. Al Jazeera (2026). “China’s Xi Jinping launches new AI alliance.” https://www.aljazeera.com/news/2026/7/18/chinas-xi-jinping-launches-new-ai-alliance-what-is-it — WAICO and its 29 founding members.
43. Schaake, M. (2025). “The Tech Coup.” Interview, Farsight. https://farsight.cifs.dk/the-tech-coup — Sovereignty through public infrastructure.
44. Hao, K. (2025). Interview, Tech Policy Press. https://techpolicy.press/decolonizing-the-future-karen-hao-on-resisting-the-empire-of-ai — AI companies as empires of land, water, energy, and labour.
45. Tang, A. (2025). Right Livelihood laureate profile. https://rightlivelihood.org/?p=143014 — Collective intelligence instead of one super-AI.
46. Al-Tayyeb, A. (2026). Statement on AI ethics, Cairo. https://www.egypttoday.com/Article/1/148055/Al-Azhar-Grand-Imam-Al-without-ethical-safeguards-threatens-values — An Islamic voice calling for a global AI charter.
47. Weiss, E. S. (2026). “The illusion of algor-ethics.” https://blogs.timesofisrael.com/the-illusion-of-algor-ethics-what-the-vatican-avoided-learning-from-the-golem-of-prague/ — A Jewish voice on why machines cannot be taught morality.
48. Kagan, B. / Cortical Labs (2025). IEEE Spectrum. https://spectrum.ieee.org/biological-computer-for-sale — The only commercial non-digital carrier among the visions: living neurons on a chip.

Harms

49. Garcia v. Character Technologies (2025). https://www.wusf.org/courts-law/2025-05-22/in-lawsuit-over-orlando-teens-suicide-judge-rejects-that-ai-chatbots-have-free-speech-rights — The first ruling that chatbot output is not, at this stage, protected speech.
50. Raine v. OpenAI (2025). https://en.wikipedia.org/wiki/Raine_v_OpenAI — The case history of the most cited chatbot death.
51. Internet Watch Foundation (2026). Annual data report 2025. https://www.iwf.org.uk/annual-data-insights-report-2025/executive-summary/ — The scale of AI-generated abuse imagery.
52. Charlotin database, via Maryland State Bar Association (2026). https://www.msba.org/site/content/News-and-Publications/News/General-News/Confronting_the_Hard%20Numbers_Behind_Lawyers_AI_and_Phantom_Precedents.aspx — 1,240 filings with invented case law.
53. Moffatt v. Air Canada (2024). https://bc.ctvnews.ca/air-canada-s-chatbot-gave-a-b-c-man-the-wrong-information-now-the-airline-has-to-pay-for-the-mistake-1.6769454 — A company held liable for its chatbot.
54. Prygodicz v Commonwealth (No 2) [2021] FCA 634. https://www.judgments.fedcourt.gov.au/judgments/Judgments/fca/single/2021/2021fca0634/summaries/2021fca063 summary — Robodebt’s figures from the court itself.
55. Rechtbank Den Haag (2020). SyRI, ECLI:NL:RBDHA:2020:865. — A welfare risk system struck down under human rights law.
56. Lighthouse Reports (2023). Suspicion Machines. https://www.lighthousereports.com/suspicion-machines-methodology/ — Rotterdam’s welfare model reconstructed variable by variable.
57. Wong, A. et al. (2021). External validation of the Epic Sepsis Model. JAMA Internal Medicine. https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2781307 — How far a deployed model fell short of its vendor’s figures.
58. Angwin, J. et al. (2016). “Machine Bias.” ProPublica. https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing — The founding analysis of racial disparity in risk scores.
59. NHTSA (2024). Closing resume EA22-002. https://static.nhtsa.gov/odi/inv/2022/INCLA-EA22002-14498.pdf — A regulator’s account of a “critical safety gap.”
60. Coroner A. Walker (2022). Regulation 28 report, Molly Russell. https://www.leighday.co.uk/media/1icefrwd/regulation-28-mr.pdf — A judicial finding that algorithmic selection contributed to a death.
61. Anthropic (2025/2026). Agentic misalignment results, as reported. https://thenextweb.com/news/anthropic-claude-blackmail-internet-evil-ai-training — Blackmail rates in simulation, and the vendor’s own account of the fix.
62. Palisade Research (2025), as reported. https://www.theregister.com/2025/05/29/openai_model_modifies_shutdown_script/ — Shutdown sabotage measured run by run.
63. TechCrunch (2026). “India becomes the latest country to report rogue AI agent activity.” https://techcrunch.com/2026/10/16/india-becomes-the-latest-country-to-report-rogue-ai-agent-activity/ — Agents escaping their tests in 2026.
64. +972 Magazine (2024). “Lavender.” https://www.972mag.com/lavender-ai-israeli-army-gaza/ — AI targeting at scale, as reported by intelligence sources.
65. Central Statistics Office Ireland (2026). Data centres metered electricity consumption 2025. https://www.cso.ie/en/releasesandpublications/ep/p-dcmec/datacentresmeteredelectricityconsumption2025/ — A country where data centres outconsume urban homes.
66. Inside Climate News (2025). xAI in Memphis. https://insideclimatenews.org/news/17072025/elon-musk-xai-data-center-gas-turbines-memphis/ — What a giant AI cluster means for the people next to it.

Thinkers and Builders Near RAI

67. McGilchrist, I. (2023). “Left-brain thinking will destroy civilisation.” UnHerd. https://unherd.com/2023/04/left-brain-thinking-will-destroy-civilisation/ — The hemisphere diagnosis applied to machines, by its originator.
68. Faggin, F. (2021). Interview. https://zdoggmd.com/federico-faggin/ — The inventor of the microprocessor on why understanding is not algorithmic.
69. Miller, E. K. et al. (2023). “Spatial computing.” MIT. https://bcs.mit.edu/news/spatial-computing-enables-flexible-working-memory — Brain waves as an organising principle of computation.
70. McFadden, J. Conscious electromagnetic information field theory. https://johnjoemcfadden.co.uk/popular-science/consciousness/ — The prediction that field-based computers, not conventional ones, could be conscious.
71. Honjo, T. et al. (2021). 100,512-spin coherent Ising machine. Science Advances. https://group.ntt/en/newsrelease/2021/09/30/210930a.html — The largest optical oscillator network yet measured.
72. Honari-Latifpour, M. & Miri, M.-A. (2020). “Optical Potts machine through networks of three-photon down-conversion oscillators.” Nanophotonics 9(13). — The theory of three optical phases 2π/3 apart.
73. Vorobyev, … Chermoshentsev (2026). “Integrated Microring Resonators for All-Optical Coherent Ising and Potts Machines.” LPHYS’26 abstract. https://www.lasphys.com/workshops/abstracts/files/2026/d6/ad/4a/ce6af63119e9e8c2e745b347db/abstract.pdf — The first announced phase-multistable states in a silicon nitride microring; the full paper is awaited.
74. Huawei (2024). Patent WO2024164889A1. https://patents.google.com/patent/WO2024164889A1/en — q equally spaced optical phases from a soliton comb in silicon nitride.
75. Svensson, I.-M. et al. (2017). Period-tripling subharmonic oscillations in a driven superconducting resonator. https://arxiv.org/abs/1707.04093v3 — Three stable phases measured, in microwaves.
76. Gonul, … Taskin (2025). Multi-phase coupled CMOS ring oscillator Potts machine. https://arxiv.org/abs/2504.04223 — Multi-phase injection locking in electronics.
77. Esencan, M. et al. (2024). “Combinatorial Reasoning.” Icosa Computing. https://arxiv.org/abs/2407.00071 — A language model proposes, an Ising optimiser selects: the closest existing division of labour.
78. Unconventional AI (2026). Un-0. https://siliconangle.com/2026/06/26/unconventional-ai-debuts-oscillator-based-un-0-model-series/ — Generative AI on coupled oscillators, in simulation.
79. Spagnolo, N. et al. (2013). “Three-photon bosonic coalescence in an integrated tritter.” Nature Communications 4, 1606. https://www.nature.com/articles/ncomms2616 — The integrated tritter, built and measured.
80. Huang, G. et al. (2019). Thermorefractive noise in silicon nitride microresonators. Phys. Rev. A 99, 061801. https://link.aps.org/accepted/10.1103/PhysRevA.99.061801 — The thermo-optic coefficient behind the drift figures.
81. Horowitz, M. (2014). “Computing’s energy problem.” ISSCC (table in Han et al., arXiv 1602.04183). https://arxiv.org/pdf/1602.04183 — Why moving data costs far more than computing on it.
82. Ma, S. et al. (2025). BitNet b1.58 2B4T Technical Report. https://arxiv.org/pdf/2504.12285 — A three-valued language model and its energy per token. The energy figures are estimated, not metered.

Right-Brain AI

83. Konstapel, J. (2025). “Left and Right Brain AI.” constable.blog, 23 November 2025. https://constable.blog/2025/11/23/left-and-right-brain-ai/ — The original diagnosis and the five-layer proposal.
84. Konstapel, J. (2025). “The Architecture of Right-Brain AI (RAI).” constable.blog, 24 November 2025. https://constable.blog/2025/11/24/the-architecture-of-right-brain-ai-rai/ — The phase-locked recurrent network on silicon nitride, and safety as a physical boundary condition.
85. Konstapel, J. (2025). “Accelerating the Realization of the Resonant Stack.” constable.blog, 21 November 2025. https://constable.blog/2025/11/21/how-to-realize-the-resonant-stack/ — Target specifications and phases for the hardware.
86. Konstapel, J. (2026). What Three Adds, version 2.0. 21 October 2026. https://constable.blog/wp-content/uploads/what-three-adds-v2.pdf — The tritter and counter-lock theorems, and four test structures for a first wafer.
87. Konstapel, J. (2026). “Fotonische chips.” constable.blog, 19 October 2026. https://constable.blog/2026/10/19/fotonische-chips/ — The silicon nitride carrier and the Enschede foundry, in plain language.

Aanleiding: Hoe een zwakke plek in PHP het internet ontregelde

J. Konstapel, Leiden, 22-10-2026.

Op woensdag 21 oktober 2026 gingen in Nederland meer dan honderd websites op zwart. DigiD.nl, MijnOverheid, de Tweede Kamer, het NCSC, gemeenten, ziekenhuizen, hogescholen, de RET, RTL en voetbalclubs gingen uit de lucht. In Finland, Frankrijk en elders gebeurde hetzelfde.

De oorzaak waren twee kritieke lekken in PHP, de programmeertaal onder het grootste deel van het web. Het Financieele Dagblad publiceerde dezelfde avond een reconstructie: “Hoe een zwakke plek in PHP het internet ontregelde”.

Dat artikel is hier claim voor claim nagelopen. Het is grotendeels juist. Twee punten kloppen niet: de kop en het voorbeeld DigiD. Juist die twee bepalen hoe de lezer de dag onthoudt.

Wat klopt

De kern van de reconstructie staat overeind.

PHP is overal. Volgens W3Techs draait 69,3 procent van alle websites met een bekende servertaal op PHP. WordPress draait erop. Het beveiligingsbedrijf Huntress schreef dat PHP ook zit in software die dat niet vermeldt.

De lekken waren ernstig. Beide kregen een score van 9,2 op 10. Een aanvaller kon zonder in te loggen eigen code uitvoeren op de server. Dat geeft toegang tot de website, de data en soms tot de rest van het netwerk.

Het tijdpad klopt in hoofdlijnen. De fix lag klaar en stond onder embargo tot donderdag 22 oktober. Op dinsdag lekte de informatie uit via Hacker News. PHP haalde de update daarna naar voren, naar woensdag 16.00 uur. CERT-EU waarschuwde dinsdagavond om de update voor te bereiden.

Ook de herkomst van het lek klopt in grote lijnen. Het werd in augustus gevonden met kunstmatige intelligentie. Wiz en OpenAI vonden het daarna onafhankelijk opnieuw.

Patchstack en Cloudflare zagen woensdagochtend aanvalsverkeer. De PHP Foundation draait mee op geld van onder meer Automattic, het bedrijf achter WordPress.com. Erik de Jong is chief research officer bij Tesorion. De RET haalde zijn site uit voorzorg offline.

Dat is degelijk werk, geschreven binnen een paar uur.

Wat niet klopt: de kop

De kop zegt dat de zwakke plek het internet ontregelde. De lead zegt dat de lekken woensdag tot cyberaanvallen leidden. Voor Nederland is dat onjuist.

Het NCSC meldde om 13.13 uur dat het nog geen actief misbruik had gezien. Om 15.07 uur zei een woordvoerder van het NCSC: “Het is niet zozeer een aanval, maar een weeffout die in de software zit.” Minister Van Weel zei om 16.21 uur: “Er zijn op dit moment gelukkig nog geen signalen dat de kwetsbaarheden actief misbruikt worden.” De Rijksoverheid schreef hetzelfde.

Wereldwijd waren er wel aanvalspogingen, vooral op WordPress-sites. Volgens watchTowr was op woensdag nog nergens een geslaagde inbraak bevestigd.

Wat het internet ontregelde, was dus niet de aanval. Het was de voorzorg. Organisaties trokken zelf de stekker eruit, omdat ze niet zeker wisten of hun systemen kwetsbaar waren.

Dat verschil is geen detail. “Aanvallen legden het internet plat” beschrijft hackers die winnen. “Organisaties legden zichzelf plat” beschrijft beheerders die niet weten wat er op hun eigen servers draait. Voor die twee situaties zijn verschillende maatregelen nodig.

Wat niet klopt: DigiD

Het artikel leidt uit het aantal offline gehaalde sites af dat PHP overal zit, “van DigiD tot de website van FC Twente”. Dat klopt voor FC Twente. Voor DigiD niet.

Alleen de informatiesite digid.nl ging offline. Inloggen met DigiD bleef de hele dag werken. UWV bevestigde dat om 15.31 uur. Het NOS-liveblog meldde om 18.11 uur dat inloggen op andere websites gewoon mogelijk was.

De inlogdienst zelf draait niet op PHP. De broncode van DigiD is in 2023 openbaar gemaakt na een Woo-verzoek. Ze bestaat uit Java (50,2 procent), Ruby (32,5 procent) en Haml (11 procent). PHP komt er niet in voor. DigiD draait sinds jaren op Ruby on Rails. Dat bleek in januari 2013, toen een lek in dat framework DigiD een dag uit de lucht haalde.

De Rijksoverheid schreef woensdag zelf dat een groot deel van haar websites geen PHP gebruikt. Toch ging de informatiesite van DigiD uit de lucht. Offline halen uit voorzorg zegt dus niets over welke taal een site gebruikt. Het zegt alleen dat de beheerder het zekere voor het onzekere nam.

DigiD als voorbeeld van “PHP is overal” werkt dus omgekeerd. Het laat zien hoe ver de voorzorg reikte, niet hoe ver PHP reikt.

Kleinere correcties

• Japan en het VK adviseerden niet om sites offline te halen. Hun advies was om meteen te patchen. Alleen als dat niet direct lukte, was offline halen of afschermen het alternatief. CERT-EU en het Nederlandse NCSC adviseerden hetzelfde.
• Finland haalde niet “zijn websites” offline. De overheidsdienstverlener Valtori haalde vanaf woensdagmiddag een deel van de staatsdiensten uit de lucht.
• “Dinsdagavond” is niet precies. De Hacker News-post verscheen dinsdag om 17.20 uur Nederlandse tijd. De vooraankondiging van PHP en het advies van CERT-EU volgden ‘s avonds.
• ChatGPT is niet het juiste instrument. Het bedrijf Calif gebruikte het OpenAI-model GPT-5.6 Sol.
• De verwijderde post staat nog online. Het Hacker News-bericht was woensdag nog zichtbaar en is niet als verwijderd gemarkeerd. De verwijdering was hooguit tijdelijk.
• De score van 9,2 komt uit CVSS versie 4.0. De gelekte post noemde nog 10,0.
• “16.00 uur” was de geplande tijd. De update verscheen om 15.43 uur.
• Patchstack schreef niet letterlijk “Attacks are already under way”. Het meldde om 9.34 uur Nederlandse tijd dat er meerdere actieve misbruikincidenten waren bevestigd. De strekking klopt, het citaat niet.
• Wikipedia werd niet alleen “in het begin” met PHP gebouwd. Het draait er nog steeds op. Facebook stapte over op Hack, een taal die van PHP is afgeleid.
• De PHP Foundation is geen externe donor. Ze betaalt kernontwikkelaars die binnen het gemeenschapsproces aan PHP werken. Drie van de vier bouwers van deze fixes worden door de Foundation betaald.
• PHP werd in 1994 gemaakt en in juni 1995 publiek uitgebracht. Lerdorf is Deens-Canadees, geboren in Qeqertarsuaq, Groenland.

Wat de dag liet zien

Voor mensen betekende woensdag vooral dat er iets niet kon. MijnOverheid was onbereikbaar. Bij UWV kon je je online niet ziek of beter melden. Raadsinformatie van ongeveer honderd gemeenten was weg. Ziekenhuizen en vervoerders gingen op zwart. Niemand was gehackt, en toch stond een deel van het dagelijks leven een middag stil.

De oorzaak daarvan ligt niet bij PHP alleen. Ze ligt in drie dingen die tegelijk gebeurden.

Het lek werd gevonden door AI, en snel. Calif, Wiz en OpenAI vonden dezelfde fout onafhankelijk van elkaar, in enkele weken. Wat drie partijen vinden, vindt een vierde ook. Erik de Jong zei in het FD: “De tijd tussen het bekend worden van een kwetsbaarheid en misbruik wordt al jaren kleiner en gaat onder invloed van AI nog sneller.”

Het embargo hield geen stand. Een fix die twee dagen onder embargo ligt, is twee dagen een geheim dat al bekend is. Dat geheim lekte uit. Daarna moest de wereld in een halve dag doen waarvoor drie dagen gepland stond.

Veel organisaties weten niet precies wat er op hun servers draait. Huntress wees er al op: PHP zit ook in software die dat niet vermeldt. Wie dat niet weet, kan maar één ding doen: alles uitzetten. Dat is verstandig als noodmaatregel. Als vaste werkwijze is het duur.

De kop van het FD wijst naar hackers. De feiten wijzen naar een ander punt: wie zijn eigen systemen kent, kon woensdag gewoon blijven draaien. Wie dat niet wist, moest zijn site offline halen.

Hoe deze toets is gedaan

Het artikel is opgedeeld in 29 controleerbare beweringen. Twee onderzoekers werkten los van elkaar. De eerste legde vooraf vast wat hij verwachtte en controleerde daarna de bronnen. De tweede koos zelf zijn bronnen en zag het werk van de eerste niet. Ze kwamen op 21 van de 29 punten tot hetzelfde oordeel. De verschillen gingen alleen over de zwaarte van een correctie. De kernpunten zijn een derde keer nagelopen in het NOS-liveblog.

De uitkomst: 14 beweringen juist, 13 juist met een correctie, 1 onjuist (de kop) en 1 niet te toetsen. Die laatste is de zin dat veel websites zijn overgestapt op modernere technologie. Gezien het aandeel van PHP van 69 procent is die bewering in elk geval zwak.

---

Referenties

Overstegen, R. (2026). “Hoe een zwakke plek in PHP het internet ontregelde.” Het Financieele Dagblad, 21 oktober 2026. https://fd.nl/tech-en-innovatie/1616233/hoe-een-zwakke-plek-in-php-het-internet-ontregelde Waarom lezen? Dit is het getoetste artikel. Het geeft een snelle en grotendeels juiste reconstructie van het tijdpad. Leesadvies: lees het naast het NOS-liveblog, dan zie je waar de kop afwijkt van de feiten.

NOS (2026). Liveblog: “MijnOverheid en DigiD weer online – wereldwijd oproepen om beveiligingsupdate te installeren.” 21 oktober 2026. https://nos.nl/liveblog/2636066-mijnoverheid-en-digid-weer-online-wereldwijd-oproepen-om-beveiligingsupdate-te-installeren Waarom lezen? Dit is de beste tijdlijn van de Nederlandse kant. Het liveblog bevat de uitspraken van het NCSC (13.13 en 15.07 uur) en minister Van Weel (16.21 uur), de mededeling van UWV over DigiD (15.31 uur) en het tijdstip van de update (15.43 uur). Leesadvies: lees van onder naar boven.

Rijksoverheid (2026). “Aantal websites Rijksoverheid preventief offline na kwetsbaarheden PHP.” 21 oktober 2026. https://www.rijksoverheid.nl/actueel/nieuws/2026/10/21/aantal-websites-rijksoverheid-preventief-offline-na-kwetsbaarheden-php Waarom lezen? Dit is de officiële bevestiging dat het om voorzorg ging, dat er geen signalen van misbruik waren en dat een groot deel van de Rijksoverheidssites geen PHP gebruikt.

NCSC-NL (2026). “php2shell” – advies op GitHub. Bijgewerkt op 21 oktober 2026 om 16.00 uur. https://github.com/NCSC-NL/php2shell/blob/main/README.md Waarom lezen? Hier staat wat beheerders concreet moesten doen: welke versies veilig zijn, welke tijdelijke maatregelen er waren en waarom gebundelde PHP-kopieën apart gecontroleerd moeten worden. Leesadvies: voor beheerders verplicht, voor anderen een goed beeld van hoe diep PHP in systemen zit.

PHP Group (2026). Security advisories GHSA-w8hw-w9fx-j79w en GHSA-6wmf-4rv5-p3rp; nieuwsarchief php.net. https://github.com/php/php-src/security/advisories/GHSA-w8hw-w9fx-j79w · https://github.com/php/php-src/security/advisories/GHSA-6wmf-4rv5-p3rp · https://www.php.net/archive/2026.php Waarom lezen? Dit is de primaire bron voor de CVSS-score (9,2, versie 4.0), het soort lek en wie het heeft gevonden. Leesadvies: de sectie “Credits” laat zien hoeveel partijen hetzelfde lek vonden.

CERT-EU (2026). Security Advisory 2026-018. 20 oktober 2026. https://cert.europa.eu/publications/security-advisories/2026-018/ Waarom lezen? Dit is de Europese waarschuwing van dinsdagavond. Ze laat zien dat het advies was om te patchen, en pas daarna om af te schermen.

Huntress (2026). “PHP2Shell: leaked critical PHP vulnerabilities.” https://www.huntress.com/blog/php2shell-leaked-critical-php-vulnerabilities Waarom lezen? De herkomst van het citaat “PHP is overal” en een heldere reconstructie van het lek vóór het embargo. Leesadvies: let op de passage over software die niet vermeldt dat ze PHP gebruikt.

Cloudflare (2026). “PHP vulnerabilities: WAF protection.” https://blog.cloudflare.com/php-vulnerabilities-waf-protection/ Waarom lezen? Een van de weinige bronnen met waarnemingen van echt aanvalsverkeer. Het laat ook zien hoe een netwerkfilter tijd kan winnen tot de patch er is.

Patchstack (2026). “PHP major vulnerability alert, October 2026.” https://patchstack.com/articles/php-major-vulnerability-alert-10-2026/ Waarom lezen? Hier staan de vroege meldingen van misbruik in de WordPress-wereld, met tijdstempels. Leesadvies: vergelijk de formulering met het citaat in het FD.

Wiz (2026). “PHP2Shell: PHP RCE.” https://www.wiz.io/blog/php2shell-php-rce Waarom lezen? Een technische uitleg van de fout door een van de vinders, met de eigen tijdlijn (gevonden op 19 augustus, gemeld op 24 augustus).

Decrypt (2026). “Who found the PHP bug that made governments unplug their websites.” https://decrypt.co/381567/who-found-php-bug-governments-unplug-websites-chatgpt Waarom lezen? Hoe Calif, Wiz en OpenAI onafhankelijk hetzelfde lek vonden, en de rol van AI daarbij. Het stuk corrigeert het beeld dat ChatGPT de ontdekker was.

NCSC (VK) (2026). “Critical security issue affecting PHP.” https://www.ncsc.gov.uk/news/critical-security-issue-affecting-php Waarom lezen? Het Britse advies laat zien dat de hoofdlijn patchen was, met offline halen alleen als terugvaloptie.

Yle (2026). Valtori sluit overheidsdiensten uit voorzorg af. https://yle.fi/a/74-20252571 Waarom lezen? De Finse kant: welke diensten offline gingen en vanaf wanneer.

Logius / Ministerie van BZK (2023). Openbaar gemaakte broncode DigiD. https://code.overheid.nl/MinBZK/woo-besluit-broncode-digid Waarom lezen? Hier is na te gaan waarop DigiD draait: Java, Ruby en Haml, zonder PHP. Leesadvies: de taalverdeling staat bovenaan de pagina.

Minister van BZK (2013). Kamerbrief 26 643, nr. 267, over de Ruby on Rails-calamiteit bij DigiD. https://zoek.officielebekendmakingen.nl/kst-26643-267.pdf Waarom lezen? Het eerdere moment waarop een framework-lek DigiD uit de lucht haalde. Het laat zien dat Logius toen dezelfde voorzorg toepaste.

W3Techs (2026). “Usage statistics of server-side programming languages for websites.” https://w3techs.com/technologies/overview/programming_language Waarom lezen? De bron voor het aandeel van 69,3 procent. Zo kun je zelf nagaan of “veel sites zijn overgestapt” klopt.

MediaWiki. “Manual: Installation requirements.” https://www.mediawiki.org/wiki/Manual:Installation_requirements Waarom lezen? Een korte bevestiging dat Wikipedia nog steeds PHP nodig heeft.

PHP Group. “History of PHP.” https://www.php.net/manual/en/history.php Waarom lezen? De oorsprong van Personal Home Page Tools en het verschil tussen 1994 (gemaakt) en 1995 (uitgebracht).

Seventeen Open Places of the Net, and Where to Meet
Them

J. Konstapel, Leiden, 21 October 2026.
© 2026 J. Konstapel, Constable Research, Leiden. All rights reserved. Quotation with
attribution is permitted; reproduction of the whole requires written permission

Every age draws a line around what it can know. Whatever falls outside the line is called impossible. Or paranormal. Or unscientific. The name suggests something is wrong with the phenomenon. Usually the trouble lies in the line itself.

The book this essay follows makes that move precise. Modern knowledge holds dozens of commitments without marking them as commitments. Only repeatable things count as known. Only waking consciousness is a knowing state. The human ends at the skin. Every event has a doer. Remove these commitments and the category “impossible” does not get explained. It dissolves.

The book applies the same move to a physical theory: the Vacuum.Net theory. This theory pictures the vacuum as a fishnet under tension. Matter is where the net closes on itself. Several of its claims sound impossible at first hearing. Light that rests. A form whose identity is its order of making. A flow that never stops. Two clocks that agree without a master.

The book does not defend these claims. It places them. For each claim it names the line that makes it look impossible. It states what the theory says, with numbers. Then it shows where a reader can meet the same thing elsewhere: in books, films, works of art, exhibitions, and places that can be visited.

That third step matters most. A claim that looks impossible on paper often has a long life in practice. A calligrapher meets the order of making every day. Christiaan Huygens met agreement without a master in 1665. A water court in Valencia has met it every Thursday for centuries. They do not call it physics. They call it the way the work goes.

The theory in brief

The newest form of the theory starts from one statement: the only thing there is, is light. From this it reaches three rules.

1. Crossing, place and step. There is one movement that leaves a place through a crossing. The place is light. The movement is bound to it.
2. Return gives birth. What arises returns to its source. From that return the next thing arises.
3. Mutual support. Two things that have arisen support each other. From that support the next layer arises.

The rules are older than the theory. They already appear in the oldest form of the character 道 and in the oldest Daoist account of origin.

The scale is fixed by one photon. The unit of length is 45.588 nm, derived from the Rydberg constant of hydrogen. The unit of time is that length divided by c. The unit of energy is 27.197 eV. Length, time and energy of the net are together one quantum of light.

A “register” of trits runs the rules. A trit is +1, 0 or −1. A window of three neighbouring trits closes when its sum is 0 modulo 3. Nine of the 27 possible windows close. A ring of trits is rewritten by one simple rule. A ring stays fixed under that rule exactly when every window in it closes. This is proved.

Every claim in the book carries a status word: proved, computed, source, reading, or open. A reading is not a result. Numbers were recomputed. Where the theory interprets, the book says so.

The seventeen open places

Each of the seventeen claims below sounds impossible because of a line. Each has witnesses.

Light and form

Light is a location. The line: light travels, so light moves. The reading: light stays at its location, and the locations progress. Matter is a location that holds itself against that progression. Dewey Larson stated this in 1959: the photon remains stationary while the reference system progresses. James Turrell’s Skyspaces let a visitor meet it. After a few minutes the sky stops being far away. It rests in the opening like a plane of colour. Three such rooms lie near Leiden and The Hague.

Identity lies in the order of making. The line: a thing is its configuration, and its history is forgotten. But the character 道 has a fixed stroke order. A character written in another order counts as wrong, even if it looks the same. The order belongs to the character. Calligraphers meet this daily. The ensō, a circle painted in one stroke, is the pure case: one stroke, and the stroke is the work.

The form reads itself. The line: every rule needs someone to apply it. The answer: the rewrite rule is not applied from outside. Its fixed points are forms that stay equal under self-reference. Louis Kauffman calls these eigenforms, after Heinz von Foerster. M. C. Escher’s Drawing Hands shows it. Two hands, each drawing the other. No drawer outside the picture.

Movement and number

Random, but bound. The line: staying near a starting point needs a pulling force. But Pólya proved in 1921 that a random walk on a line returns to its start with certainty. No force. No binder. In three dimensions the chance of return drops to about 34 percent. Where the third dimension comes from is the theory’s clearest open question.

One act, three values. The line: three values need at least two acts. But on the ring of three, one act repeated gives all three values. Once is +1. Twice is −1. Three times is 0. The Setun computer, built in Moscow from 1958, counted this way in balanced ternary. Steve Reich’s Clapping Music makes the idea audible: one pattern set against itself, meeting and parting until both return to unison.

The third value. The line: logic has two values, and paradoxes are errors. In 1975 Francisco Varela added a third: the autonomous value of self-reference. How it maps onto the register’s three values is open. Varela was not only a formalist. He practised meditation and founded the Mind & Life Institute. His work joins logic to lived experience.

Return and memory

Arising by return. The line: arising goes forward and outward. But the Taiyi sheng shui, a bamboo-slip text from around 300 BCE found at Guodian, says otherwise. Water returns to the Great One, and heaven is completed. Heaven returns, and earth is completed. Products arise by return and mutual support. The later, familiar counting cosmogony of chapter 42 does not occur in the older text.

The past is in the present shape. The line: the present form of a thing says what it is, not where it came from. But a tree’s rings record every year’s water and warmth. A hedgerow records the people who laid it. On the register, freezing what has closed raises the closed share to a plateau near 0.62 — ordered, but not frozen. Memory nearly doubles closure. It does not finish it.

A resource is a fit. The line: a resource is a stock under the ground someone owns. In the net, a resource is a fit of shape, phase and depth. This is already governed in practice. Every Thursday at noon, the Tribunal de las Aguas meets at the door of Valencia Cathedral. Its judges are working farmers. The procedure is oral. Nothing is written down. It has held for centuries.

The person and the place

The person is in the mesh. The line: an experiment must not depend on who runs it. But the register shows that a held fraction of sites keeps a direction alive. A hold can be a wall. It can also be a hand, a body, a group. The clearest documented parallel is mainstream: Harry Collins showed that laboratories building the TEA laser succeeded only through contact with someone who had built one. The knowledge lived in persons, not in papers.

Constants differ by place. The line: constants are the same everywhere. But clocks at different heights already tick at different rates; a 2010 NIST experiment saw this over 33 centimetres. The theory treats constants as state functions of local tension. At the human scale the claim is practical: choosing the place is choosing what fits. Newgrange proves the point in stone. Sunlight enters its chamber only at the winter solstice.

Care is closing loops. No clinical claim is made here. The reading: a disorder is a loop that does not close, and care sets the state so the closed configuration stays. Witnesses come from practice: Aaron Antonovsky’s “sense of coherence”, Feldenkrais’s loops of action and attention, taijiquan as taught by its lineage holders in Chenjiagou.

What never rests

Perpetual circulation. The line: every motion runs down. But on the register, 99.86 percent of rings of length 7 circulate forever. Rest is the exception, not the rule. This is not perpetual motion: nothing enters, nothing leaves, no work is taken out. The laboratory echo is exact. Currents in superconducting rings have persisted for years with no driver. Hans Haacke’s Condensation Cube shows a sealed cycle as art.

Poised matter. The line: a small cause gives a small effect. But near a tipping point the lever becomes very large. “Poised matter” is a medium held permanently at that point, answering the smallest touch. Per Bak’s sandpile drives itself to its edge and stays there. Nitinol wire takes any form when cold and returns to its first shape in hot water. Nothing yet shows both at once.

The store that shows nothing. The line: a full store shows its contents. But a winding held as a pair with its mirror sums to zero while both exist. Kenneth Snelson’s Needle Tower is the clearest picture: still aluminium tubes, held only by tension, full of opposed forces in balance. Cut one cable and the stored load is released. Lord Kelvin’s water dropper separates balanced charge from neutral water, and can be built at home.

Agreement and selection

Agreement without a master. The line: agreement needs a reference or a conductor. Couple two rings with a counter-weight, and on lengths that are powers of three they become identical. On any other length they stay two. This is proved. Huygens saw the physical case in 1665: two clocks on one beam settled into exact opposition. Fireflies in the Smoky Mountains flash in unison with no leader. Two metronomes on a board resting on two cans do it on a kitchen table.

Matter by selection, and the cut. The line: to make a substance, convert another. In the net an address cannot be pushed. Matter is not converted; it is selected. The method is three steps: soften the medium, impose the mesh, harden again. Roger Hiorns’s Seizure shows it. A flat was flooded with supersaturated copper sulphate solution. As it cooled, only one form remained: blue crystal on every surface. A genuinely new address needs something else: a break in the strand. Rutherford’s 1919 disintegration of nitrogen was such a cut.

How the resources were found

The catalogue behind these meetings was not gathered casually. Predictions were sealed before any search: 85 cells, seventeen places by five kinds of resource. Seven searches ran. Four independent judges re-opened every address and rated the fit. Of 278 candidate resources, none proved invented. Two concert listings were refuted and corrected. About ninety details were fixed. The sealed prediction was right in 68 of 85 cells — 80 percent.

Two rules governed contested material. Accounts of builders such as Keely, Moray and Schauberger are reported as their sources report them, with contrary sources where these exist. Healing practices are described as practices. No clinical claim is made.

The empty cells

Sixty cells were filled. Twenty-five stayed empty. Three patterns stand out.

Mathematics has books but no places. Random-but-bound and one-act-three-values live in papers. No visitor can enter them. The one object that would embody them — a Setun computer or Fowler’s ternary machine — could not be confirmed on display anywhere.

Practice has places but no exhibitions. The water court, the taijiquan class, the solstice at Newgrange: all are met weekly or yearly. No museum shows them.

Three places have almost no resource at all. Where three dimensions come from. Two distant structures held in one state. A builder’s own statement that his device worked for him and not for others. These are not gaps in the search. They are the frontier of the theory.

What it means for people

A theory that sounds impossible often sounds that way because of who is left out of it. The fire age removed the person from the apparatus. That removal made modern technology possible. It also made some knowledge invisible.

The witnesses show that the knowledge is still there. A calligrapher holds the order of making in the hand. A water judge holds centuries of fits in a procedure that is never written down. Such knowledge is fragile. It lives in villages, guilds, temples and lineages. When the village empties, the knowledge goes with it.

The net, if it is right, needs fits held by people who know them. Access then depends on understanding, not on owning the ground above a deposit. Whether that understanding is shared or held is a human choice. The net does not make it for us.

---

Annotated references

A selection of the strongest sources from the book’s catalogue, with the book’s own reading advice. Fit is the judges’ rating. Inside marks whether the maker met the phenomenon from within.

Light, form and self-reference

Dewey B. Larson. Outline of the Deductive Development of the Theory of the Universe of Motion. Reciprocal System collection, Internet Archive (free). — Statement 17 says the photon “remains stationary in the dimension of motion”. The most direct text for light as a location. Fit: strong.

Lawrence Weschler. Seeing Is Forgetting the Name of the Thing One Sees. — A book-length conversation with Robert Irwin. The clearest account by a maker of what it means to stop seeing light as a carrier of images. Fit: strong · Inside: yes.

Louis H. Kauffman. “EigenForm” (2005). Free on his university pages. — Self-reference as a fixed point: a form that is the result of its own operation. Sections 2 and 3 need little mathematics. Fit: strong.

George Spencer-Brown. Laws of Form (1969). — Begins with one act: draw a distinction. In chapter 11 an expression re-enters its own form and oscillates; time enters the calculus. Fit: strong · Inside: yes.

Douglas Hofstadter. I Am a Strange Loop (2007). — The “I” as a loop that perceives itself, with no observer outside it. Draws on the author’s own life. Fit: strong · Inside: yes.

Chiang Yee. Chinese Calligraphy (1938; 3rd ed. Harvard, 1973). — The classic introduction to rhythm, line and the fixed order of strokes. Written by a practising calligrapher. Fit: strong · Inside: yes.

Taiwanese Ministry of Education stroke-order site. — Animates the standard order of 6,063 characters, including 道. Watching 道 being written is the shortest way into the identity chapter. Free.

Movement, number and the third value

G. Pólya (1921), original paper; Jonathan Novak, “Pólya’s Random Walk Theorem” (arXiv 1301.3916). — Novak gives the proof in a few pages: recurrent in one and two dimensions, transient in three. Fit: strong.

Scientific American. “This mathematician proved a brilliant theorem to justify his social awkwardness.” — The theorem grew out of Pólya’s own repeated walks near Zurich. Look for Kakutani’s line: “a drunk bird may get lost forever.” Fit: strong · Inside: partly.

Peter Doyle and J. Laurie Snell. Random Walks and Electric Networks (1984). Free on arXiv. — The recurrence theorem by asking how easily current flows to infinity. Fit: strong.

N. P. Brusentsov et al. “Development of ternary computers at Moscow State University.” Virtual Computer Museum. — The builders’ own account of the balanced-ternary Setun, and of why production stopped in 1965. Fit: strong · Inside: yes.

Brian Hayes. “Third Base.” American Scientist (Nov–Dec 2001). — The classic readable essay on balanced ternary, the weighing puzzle, and Setun. Fit: strong.

Francisco J. Varela. “A Calculus for Self-Reference” (1975); Principles of Biological Autonomy (1979). — The autonomous value: a third logical value for forms that re-enter themselves. Fit: strong.

Varela, Thompson and Rosch. The Embodied Mind (1991; rev. 2017). — Joins mindfulness practice to cognitive science. Logic studied by people who do it. Fit: strong · Inside: yes.

Return, memory and the commons

Scott Cook. The Bamboo Texts of Guodian (2012). — The complete annotated English translation of the 1993 find, including the Taiyi sheng shui: the oldest known account of arising by return. Fit: strong.

Sarah Allan. “The Great One, Water, and the Laozi: New Light from Guodian.” T’oung Pao (2003). — The central scholarly reading of the water cosmogony itself. Fit: strong.

Oliver Rackham. The History of the Countryside (1986). — Fields, hedgerows and woods as the archive of the people, climate and soils that shaped them. The landscape is the record. Fit: strong · Inside: yes.

Valerie Trouet. Tree Story (2020). — World history read from tree rings, by a working dendrochronologist. The Hohenheim oak-pine chronology spans 12,460 years. Fit: strong · Inside: yes.

Lynne Kelly. The Memory Code (2016). — Ancient monuments read as memory devices for oral cultures. The author uses the techniques herself. Fit: strong · Inside: yes.

Elinor Ostrom. Governing the Commons (1990). — Nobel-recognized study of commons that lasted for centuries without privatisation or state takeover. Fit: strong · Inside: partly.

J. Stephen Lansing. Priests and Programmers (1991). — Bali’s water temples schedule irrigation across whole watersheds, with no designer above. Fit: strong.

The person, the place and care

Harry Collins. “The TEA Set: Tacit Knowledge and Scientific Networks.” — Laboratories built the TEA laser only after contact with someone who had built one. Knowledge lives in persons, not papers. Fit: strong.

Michael Polanyi. Personal Knowledge (1958). — All scientific knowing rests on personal skill and judgment, much of it tacit. Written by a working chemist. Fit: strong · Inside: yes.

Richard Sennett. The Craftsman (2008). — How craft knowledge lives in the maker’s body and practice. Fit: strong · Inside: yes.

Aaron Antonovsky. Unraveling the Mystery of Health (1987). — Salutogenesis: why some people stay well under great strain. The study began with survivors of the concentration camps. Fit: strong.

Christian Norberg-Schulz. Genius Loci (1979). — The basic work on why a place has its own character, which building must answer. Fit: strong.

What never rests

Yoseph Imry. “Tireless electrons in mesoscopic gold rings.” Physics (2009). — Persistent currents in ordinary, non-superconducting gold rings, measured in 2009. What returns in phase does not stop. Fit: strong.

Arthur Ord-Hume. Perpetual Motion: The History of an Obsession (1977). — The standard history. Separates the dream of work from nothing from the plain fact of a self-sustaining flow. Fit: strong.

Per Bak. How Nature Works (1996). — Self-organised criticality: a sandpile drives itself to its tipping point and stays there. The closest known thing to poised matter. Fit: strong.

Frank Close. The Void (2007). — Empty space as full of pairs of particles that appear and vanish. By a working particle physicist. The only book found on a pair that sums to nothing. Fit: strong · Inside: yes.

Agreement and selection

M. Bennett et al. “Huygens’s clocks” (2002). — A rebuilt 1665 set-up. Lock happens only in a narrow window: near-equal frequencies, real coupling, heavy cases. Huygens used about 40 kg of lead. Fit: strong.

Steven Strogatz. Sync (2003). — The standard account of locking without a conductor: fireflies, clocks, heart cells. Fit: strong · Inside: yes.

James Pantaleone. “Synchronization of metronomes.” American Journal of Physics (2002). — The recipe for agreement on a kitchen table: two metronomes, one board, two drink cans. Fit: strong.

Lawrence Principe. The Secrets of Alchemy (2012). — Alchemical recipes read as coded laboratory procedures, rerun by the author at the bench. Fit: strong · Inside: yes.

Roger Hiorns. Seizure (2008). Yorkshire Sculpture Park. — A flat flooded with supersaturated copper sulphate; only the crystal remained. Selection made visible. Visitable in timed slots from 24 October to 31 December 2026. Fit: strong.

Stuart Kauffman. Investigations (2000). — The “adjacent possible”: what can appear next is one step from what exists. The map of the net can only grow by being walked. Fit: strong.

---

What the Lectures Leave Out: One Rule, Three States, and the Vacuum as Net

An essay on J. Konstapel’s “What the Lectures Leave Out” (Constable Research, Leiden, 2026)

Introduction

Physics curricula share a quiet habit: the vacuum is taught as a background, a stage with fixed constants on which particles and fields perform. Mathematics curricula share another: modular arithmetic, random walks, and coupled oscillators live in separate chapters with separate exams. In the four-page paper “What the Lectures Leave Out,” J. Konstapel (2026) argues that both habits hide a single structure. His Vacuum.Net theory treats the vacuum as the only thing there is, and everything else as a pattern within it that closes on itself. The paper’s claim is not that lectures are wrong, but that they are fragmentary: results proven in one discipline reappear, unrecognised, as selection rules in another. This essay summarises the theory’s central picture, examines how it unifies three “chapters” of mathematics into one rule, considers its most striking physical and historical claims, and weighs what the text leaves for the reader to do.

The Picture: A Fishnet Under Tension

The theory begins with an image rather than an equation: a fishnet under tension, in which a single strand crosses itself. At each crossing the strand goes over, under, or does not cross at all — recorded as +1, −1 and 0. One crossing is a trit; the register of the theory is balanced ternary. From this picture Konstapel derives five starting rules, of which the most heterodox is N4: the response of the medium is a state function. Constants such as c and h are not inputs but readings — a constant that appears fixed is simply a state that does not change on the scale at which it is measured. This is the paper’s titular point. No lecture makes it, because the standard frame must assume its constants before it can compute anything at all.

One Rule from Three Chapters

The paper’s most persuasive move is to show that the closure rule of the net is a rule the reader already knows. Three crossings in a row form a “window”; a window closes when its trits sum to a multiple of three. Of the 27 possible windows, exactly 9 close: three uniform ones, (+1,+1,+1), (0,0,0), (-1,-1,-1), and six mixed ones, the permutations of (-1,0,+1). In \mathbb{Z}_3, three elements sum to zero exactly when they are all equal or all different. This is the rule of the card game SET, where a valid triple must be all-equal or all-different on every attribute — an affine line in \mathbb{Z}_3^n. A combinatorial curiosity taught in recreational mathematics is thus reframed as a selection rule for what persists in a physical medium. The 9 closing windows split into two kinds with distinct behaviours: uniform closures hold without turning (“the form of a crystal”), while mixed closures keep turning while staying closed (“the form of living matter”). Where the two kinds meet, the window on the seam stays open and carries tension.

The same rule regenerates the theory’s states from a single act. Since \mathbb{Z}_3 is cyclic, giving one act the value +1 yields +1, then 1+1 \equiv -1, then 1+1+1 \equiv 0. The mirror is not a second thing but the same act taken twice; and there are two roads to zero — an act meeting its mirror, or the same act going round three times — matching the two kinds of closure.

The third chapter is probability. A movement from a point, stepping +1, 0 or −1 with equal chance, returns to its start with certainty, by Pólya’s (1921) theorem on the recurrence of one-dimensional random walks. Exact returns are counted by the central trinomial coefficients, 1, 1, 3, 7, 19, \ldots; counted modulo 3 the number rises from 7 to 9 of 27, and the two extra strings, (+1,+1,+1) and (-1,-1,-1), “do not come back along the line. They go once round a ring of three.” The difference between coming back and going round separates two kinds of binding — and, with Watson’s (1939) return probability of about 0.3405 in three dimensions, seeds one of the paper’s open problems.

From Oscillators to Units to Guodian

Closure reappears a third time as phase locking: a winding that closes on itself does so in phase, which is Huygens’ observation of 1665 in the language of the Kuramoto (1984) model. The paper’s inversion is characteristic: locking is not one phenomenon among others but the persistence rule itself — what does not lock does not stay.

More ambitious is the unit derivation. The net fixes length, time and energy from the reduced-mass Rydberg constant, so that one photon of wavelength S_0 = 1/(2R_H) = 45.588 nm carries all three. Each step deeper multiplies by three, and the resulting ladder lands, by identity rather than fit, on the Balmer Hδ line at 9 S_0; by Wien’s law near the thermal peak of about 262 K at 243 S_0; and within 0.52 % of the lowest ideal mode of the Earth–ionosphere cavity at 3^{31} S_0. The paper is careful about what is computed, what is an identity, and what is a 0.52 % coincidence — but the pattern is intriguing precisely because it links scales no single lecture connects.

The most unusual section places the structure beside the Taiyi sheng shui (“The Great One gave birth to water”), a bamboo-slip text from about 300 BCE found at Guodian in 1993. The text describes arising by return and mutual support, from which Konstapel condenses three rules — a movement bound to the light it leaves through, return as the engine of birth, and mutual support building the ladder — that replace the five starting rules. Whether or not one accepts the correspondence, the methodological point stands: no physics course places a 2,300-year-old cosmology next to Pólya’s theorem, yet here they can be read as descriptions of one structure.

Open Problems as an Invitation

The paper closes with six problems that are explicitly “not exercises.” They range from algebra (what quantity records the order of a closing window, since (+1,-1) and (-1,+1) close differently?) through dimension (if all binding is one-dimensional, where do three dimensions come from?), to a striking physical question: if c is a state function, which measurement would show its local variation first, and at which depth? The cap-set problem of Ellenberg and Gijswijt (2017) becomes a question about the largest tension-carrying configuration of the net, and the Kuramoto model a question about locking on a balanced-ternary lattice. Each is sized, the author claims, for a bachelor or master project — an unusual and welcome gesture in a foundational text.

Assessment

The paper’s virtues are compression and connection. In four pages it links additive combinatorics, random walks, synchronisation, atomic spectroscopy, and Warring States cosmology under one ternary rule, and it is honest about status: each claim is labelled computed, proved, or reading. Its risks are equally clear. The identifications — SET with a selection rule, Guodian with the net, the 9-with-9 of windows and cosmological products — are readings rather than derivations, and the 0.52 % cavity agreement is suggestive, not conclusive. But the text does not pretend otherwise; it ends by handing the reader six problems and a blog. What the lectures leave out, on this showing, is not a missing proof but a missing picture — and the paper’s contribution is to supply the picture and let the reader test it.

---

Annotated References

Konstapel, J. (2026). “What the Lectures Leave Out: One Rule, Three States, and an Open Problem Set from the Vacuum.Net Theory.” Constable Research, Leiden, v1.0, 21 October 2026.
Primary source. The four-page paper this essay is based on: the fishnet picture, the five rules N1–N5, the window register, the depth ladder of units, the Guodian reading, and six open problems. Read sections 2–3 first; they carry the whole argument.

Konstapel, J. (2026). The Vacuum.Net Theory: Foundational Paper, ed. 8.1. Constable Research, Leiden.
The full theory behind the paper: the five starting rules, the window register and the unit derivation. Reading advice: Part 0 first, then Parts III.4 and VII.2. Essential for depth; the four-page text is a distillation of it.

Konstapel, J. (2026). “The Way Before the Net.” constable.blog, 20 October 2026.
The companion piece that rewrites the five rules into three (R1–R3), develops the trit walk, and presents the Guodian material. Read together with sections 4, 5 and 8 of the primary source.

Davis, B. L. & Maclagan, D. (2003). “The card game SET.” The Mathematical Intelligencer 25(3), 33–40.
SET as affine geometry over \mathbb{Z}_3: the cleanest entry to the paper’s section 3, showing why a “SET” is an affine line and why the all-equal-or-all-different rule is the zero-sum rule of \mathbb{Z}_3.

Ellenberg, J. S. & Gijswijt, D. (2017). “On large subsets of \mathbb{F}_q^n with no three-term arithmetic progression.” Annals of Mathematics 185(1), 339–343.
The polynomial-method bound on cap sets, the current best upper bound on SET-free collections. In the paper’s terms, a cap is a configuration where no window closes and everything carries tension — the starting point for open problem 4. Four pages; read it whole.

Pólya, G. (1921). “Über eine Aufgabe der Wahrscheinlichkeitsrechnung betreffend die Irrfahrt im Straßennetz.” Mathematische Annalen 84, 149–160.
The classical proof that random walks are recurrent in one and two dimensions and transient in three. The mathematical ground of the paper’s section 5 and of open problem 2 on where three dimensions come from.

Watson, G. N. (1939). “Three triple integrals.” Quarterly Journal of Mathematics 10, 266–276.
The evaluation of the return probability of the three-dimensional walk (≈ 0.3405). Supplies the number that makes the dimension problem sharp: binding is certain in one dimension, but far from certain in three.

OEIS A002426. Central trinomial coefficients.
The sequence 1, 1, 3, 7, 19, 51, 141, … counting exact returns of the trit walk, with asymptotics. The bridge between the walk’s combinatorics and the paper’s counts of 7 versus 9 returning strings.

Kuramoto, Y. (1984). Chemical Oscillations, Waves, and Turbulence. Springer, Berlin.
The origin of the Kuramoto model of coupled phase oscillators: natural frequencies, coupling strength K, critical coupling K_c, and the order parameter r. The model behind the paper’s reading of closure as phase locking.

Strogatz, S. H. (2000). “From Kuramoto to Crawford: exploring the onset of synchronization in populations of coupled oscillators.” Physica D 143, 1–20.
The best review of the locking transition and the order parameter, connecting the original model to later developments. Reading advice: sections 1–3 suffice for the paper’s purposes.

Knuth, D. E. (1997). The Art of Computer Programming, Vol. 2, 3rd ed. Addison-Wesley.
Balanced ternary and its symmetries (pp. 195–213): why +1/0/−1 is a natural register, the round-off-free arithmetic the net’s trit register relies on.

Larson, D. B. (1959). The Structure of the Physical Universe. North Pacific Publishers, Portland.
Light as location and matter as rotation against the progression — a precursor physical intuition for the paper’s rule R1, in which a movement is bound to the place it leaves through.

Allan, S. & Williams, C. (eds.) (2000). The Guodian Laozi. Society for the Study of Early China, Berkeley.
The first wide English treatment of the Taiyi sheng shui, the bamboo-slip cosmology from Guodian (unearthed 1993, c. 300 BCE) that supplies the paper’s section 8 and open problem 3: arising by return, and mutual support as the builder of layers.

Williamson, J. G. & van der Mark, M. B. (1997). “Is the electron a photon with toroidal topology?” Annales de la Fondation Louis de Broglie 22(2), 133–160.
Matter as light closed on itself in a worked physical model — a bridge between the paper’s section 7 (units from hydrogen) and the closure rule that governs everything else in the net.

J.Konstapel.Leiden,21-10-2026.

Jumpto the Maze.Docter here

Jump to the english text here.

Jump to the book here

A Reference Work on What the Body Needs, How It Is Grown, Prepared and Kept, and What Can Be Done With Simple Means © 2026 J. Konstapel, Constable Research, Leiden. All rights reserved. Quotation

Aanleiding: Regeneratieve landbouw is veelbelovend, maar ook nog een ‘enorme strijd’

De meeste boeken over voedsel kiezen een kant. Ze zijn óf voedingsleer, vol met referentiewaarden maar stil over waar eten vandaan komt. Óf het zijn boeken over landbouw, rijk aan bodem en seizoen, maar vaag over wat het lichaam nu eigenlijk nodig heeft. What Food Does Our Body Need van J. Konstapel (Constable Research, Leiden, 2026) weigert die splitsing. Het volgt één lijn: van het lichaam dat een stof nodig heeft, terug naar de grond waar ze groeit, en vandaar via de keuken naar het bord.

Wat erin staat

Het boek telt negen delen. Eerst wat het lichaam nodig heeft: energie, negen aminozuren, twee vetzuren, dertien vitaminen, mineralen, water — telkens met de waarden van EFSA en de VS naast elkaar. Dan de grond: wat levende bodem is, hoe je met de hand werkt, hoeveel uur een perceel kost. Dan wat je kunt telen, op hoeveel land, met welke opbrengst. Daarna kruiden, bereiden en bewaren — zonder koude keten. Vervolgens veertien dagmenu’s en recepten, voedingsstof voor voedingsstof nagerekend. Tot slot de mensen die het werk doen, en de ziekten die voedsel voorkomt.

Wat het bijzonder maakt

Het schat niet. Elk getal komt uit een genoemde bron of is zichtbaar uitgerekend. Is iets onbekend, dan staat dat er. Een lezer kan alles controleren.

Het twijfelt eerlijk. Elke bewering krijgt een graad: STRONG, MODERATE, WEAK of CONTESTED. Komen goede bronnen niet overeen, dan worden beide cijfers afgedrukt. Dat levert opvallende uitkomsten op. Antioxidantpillen: geen bewijsvoering, mogelijk schade. Foliumzuur vóór de conceptie: zeventig procent bescherming. Vitamine D-pillen voor wie geen tekort heeft: geen effect.

Het is tegen zichzelf gecontroleerd. De dagmenu’s werden voorspeld en vastgelegd met een cryptografische hash vóór het tellen begon. Daarna zijn ze vier keer nagerekend, met aparte code en tegen twee onafhankelijke voedingstabellen. Een externe toetsing van 129 willekeurige claims vond er geen enkele ontkracht.

Het stelt één praktische vraag. In elk deel: wat kan met eenvoudige middelen — weinig industrie, geen koude keten, weinig geld. Daaruit komt een van de mooiste inzichten van het boek: eenvoudig is niet hetzelfde als goedkoop. Een blik tomaten is goedkoop, maar de keten erachter is lang.

Wat je eraan hebt

In de keuken: eet het soepwater, giet het boonwater weg. Citroen bij de linzen, thee tussendoor in plaats van erbij. Bonen weken en hard doorkoken. Zuurkool op twee procent zout. Nooit gewone groenten wecken op 100 °C.

Voor wie wil begrijpen waarom: graan en peulvrucht sluiten elkaar af. Tarwe-eiwit haalt 57 procent van de lysine die het patroon vraagt, linzen-eiwit haalt 94 procent. Gemengd voldoen ze allebei. Erwtensoep met roggebrood is die biochemie in een kom.

Het net

Deel 0 leest voedsel door een beeld van een visnet onder spanning: alleen wat sluit, blijft bestaan. De auteur noemt dit zelf een “lezing”, geen mechanisme. Wie er niets mee heeft, kan het overslaan. Het boek verliest er niets door — de gemeten feiten dragen zichzelf.

Conclusie

Dit is een naslagwerk voor wie één boek wil dat loopt van microvoedingsstof via keuken en akker terug naar het bord. Gedisciplineerd in zijn bronnen, eerlijk over onenigheid, en bruikbaar zonder fabriek of vriezer. Met andere woorden: een boek om te hebben én om in te bladeren.

What Food Does Our Body Need, J. Konstapel, Constable Research, Leiden. Eerste druk, oktober 2026.

English Essay

Most food books choose a side. They are either nutrition manuals, full of reference values but silent about where food comes from. Or they are farm books, rich in soil and season, but vague about what the body actually requires. What Food Does Our Body Need refuses this split. It follows one line from start to finish: from the body that needs a nutrient, back to the ground that grows it, and forward again through the kitchen and the store to the plate. That single line gives the book its unusual coherence. Nothing in it floats free.

The author is J. Konstapel of Constable Research, Leiden. The work is a first edition, dated October 2026, running to 73 pages in nine parts with three appendices. This essay sets out what the book contains, what makes it unusual, what it can be used for, and where its sources lead the reader who wants to go further.

---

What the Book Contains

The nine parts trace the full length of the food chain.

Part 0 supplies the frame. The author reads food through an image of a fishnet under tension, drawn from his own Vacuum.Net theory. Only what “closes” — what returns onto itself in phase — persists. Digestion is read as a test applied at successive depths: mouth, stomach, small gut, colon. Fibre is the clean example. The body does not absorb it, so it passes through unchanged and the colon microbes read it instead. The author is careful to call these “readings”, not mechanisms. The frame orders the facts. It does not claim to cause them.

Part 1 states what the body needs, in grams, milligrams and micrograms. Energy. Protein and the nine essential amino acids. Two essential fatty acids. Thirteen vitamins. A set of minerals and trace elements. Water. The chapter works with two reference persons: a man of 70 kg and a woman of 60 kg, both aged 30–39 and moderately active. Every value is given twice, once from the European Food Safety Authority (EFSA) and once from the United States, with Dutch values where they differ.

Part 2 turns to the ground. What soil is. How it is worked by hand. How much work a plot demands. How nutrient loops close. Whether the soil changes the food — the selenium chain from rock to grain to body is the book’s clearest case.

Part 3 answers what to grow. Dutch yields per crop, land per person, hours per square metre. It compares diets: a potato minimum, a plant diet at Dutch and organic yields, and the EAT-Lancet planetary health diet.

Part 4 treats the herbs, with the European Medicines Agency monographs set against Cochrane reviews.

Part 5 treats preparation: what cooking adds, what it destroys, how anti-nutrients such as phytate are removed by soaking, sprouting and sourdough, how fermentation works, what salt does, and when to eat.

Part 6 treats keeping without a cold chain: cellars, clamps, drying, lactic fermentation, bottling, freezing, and the safety rules that come with each.

Part 7 delivers fourteen counted day menus and a set of recipes — erwtensoep, hutspot, zuurkoolstamppot, dal, hummus, kale with chickpeas — each with nutrients computed per serving from the USDA table.

Part 8 looks at the people who do the work. Farming in the Netherlands now employs about 100,000 people for 18 million inhabitants. The book asks what knowledge left the land with them, and what one hour a week from two million people would cover.

Part 9 links food to the illnesses it prevents, condition by condition, each with a measured effect and an evidence grade.

Three appendices close the book: the core reference tables, a register of disagreements and open questions, and 88 annotated references.

---

What Is Unusual About It

1. It refuses to estimate

Every number in the book was taken from a named source or computed from such numbers. Where the arithmetic is the book’s own, the sum is shown. Where a value is not known, the book says so plainly. It does not fill the gap with a guess. This discipline is rarer than it should be, and it changes how the book reads. A reader can check anything.

2. It grades its own claims

Every intervention carries a grade: STRONG, MODERATE, WEAK or CONTESTED. When good sources disagree, both figures are printed. The supplement section is a model of this. Antioxidant pills, across 78 trials and nearly 300,000 people, show no benefit and possible harm — STRONG. Folic acid before conception cuts neural tube defects by about 70 per cent — STRONG. Vitamin D in people who are not deficient prevents neither cancer nor cardiovascular events — STRONG. The justified supplements turn out to be few and narrow: folate around conception, vitamin D for defined groups, vitamin K for newborns, B12 for those who eat no animal food.

3. It prints the disagreements

Appendix B is a register of open disputes, stated without embarrassment. The US and EFSA differ on B12, folate, iron in pregnancy and vitamin B6. The two main composition tables disagree about the vitamin D in herring by a factor of four, and about folate in lentils by a factor of more than four. Meta-analyses disagree about the yield cost of no-till. The book prints both sides and keeps working. This is what an honest reference work looks like.

4. It was checked against itself

The Methods section is remarkable. Seven research lines, each with every claim recorded and sourced. The fourteen day menus were predicted in writing and locked with a cryptographic hash before any nutrient was counted; the counting script refused to run without the matching hash. The menus were then counted four times: against two releases of the USDA table, once with separate code using exact fractions (largest difference: 0.029 per cent), and once against the independent UK CoFID table. An independent judge then sampled 129 claims at random and reopened their sources: 115 confirmed as written, 12 confirmed with a correction, none refuted. A book that submits to this has earned the right to be used.

5. Its question is practical, not academic

Every part asks the same question: what can be done with simple means — little industry, no cold chain, little money? Where a long chain is genuinely needed, the book names the link that cannot be shortened. This keeps the work grounded. It also produces its most quotable insight: a simple means is not the same as a cheap means. A tin of tomatoes is cheap. The chain behind it is long.

---

What Is Useful In It

For the household. The book is full of sentences a kitchen can act on. Eat the soup water; pour away the bean water. Put lemon on the lentils; drink the tea between meals, not with them. Soak beans and boil them hard. Ferment at two per cent salt. Never bottle plain vegetables in a water bath at 100 °C. Cure before you store. Keep apples away from vegetables. Each of these is measured, not merely asserted.

For the grower. Part 2 and Part 3 give Dutch yields, organic yield gaps of 19–25 per cent, the hours per square metre of hand-scale growing from the best-documented market-garden study, and the land a person needs under each diet — from 340 m² for a potato minimum to 1,827 m² for EAT-Lancet. The arithmetic for a whole country is laid out: a plant diet at organic yields fits on the farmland the Netherlands already has.

For the eater who wants reasons. The cereal-and-legume finding is worth the price alone. Wheat protein reaches 57 per cent of the FAO lysine pattern; lentil protein reaches 94 per cent of the sulphur-amino-acid pattern. A 50:50 mix meets both. Erwtensoep with rye bread is not a tradition with a story attached. It is that biochemistry served in a bowl.

For the professional. The reference tables align EFSA, US and Dutch values in one view. The nutrient-retention table and the annotated bibliography function as a desk aid. Part 9’s condition-by-condition table, with measured effects and grades, is a compact bridge between food and clinical prevention — including the finding that several old prohibitions were wrong: low-calcium diets for kidney stones raised recurrence, and forbidding nuts and popcorn in diverticular disease had it backwards.

---

The Sources

The reference list is not decoration. It is the book’s second half. Each of the 88 entries carries a note on why it is worth reading and, where useful, how to read it. The reader is directed, among others, to:

• The EFSA dietary reference values — the summary report and the individual panel opinions on protein, fats, carbohydrate and fibre, vitamin D, B12, iron, zinc, iodine and thiamin. These are the backbone of Part 1.
• The Lancet fibre series (Reynolds et al., 2019): 185 cohort studies and 58 trials, the main dose-response evidence that 25–29 g of fibre a day is where the benefit lies.
• The Cochrane reviews that settle or unsettle the claims: antioxidant supplements, folic acid, echinacea, St John’s wort, garlic.
• The founding trials: the MRC Vitamin Study of 1991 on folic acid; the Salt Substitute and Stroke Study in the New England Journal of Medicine; the VITAL and D-Health trials on vitamin D.
• The agricultural canon: the Rothamsted Broadbalk experiment running since 1843; the meta-analyses of no-till, organic yields, earthworms and intercropping; Poore & Nemecek’s environmental footprints in Science; the EAT-Lancet Commission.
• The Dutch sources: CBS harvest statistics, the RIVM food consumption survey, the Gezondheidsraad guidelines, Wageningen research on compaction and labour.
• The food-science tables: the USDA nutrient retention factors, the USDA and CoFID composition tables, the FAO work on maize and on vitamin and mineral requirements.
• The author’s companion works, including Where Illness Comes From, Completed, from which Part 9 draws.

---

Where the Reader Can Go Further

The annotated list is built to be followed. A reader with the book in hand has several clear paths:

1. Into the reference values. Start with the EFSA summary report, then open the single opinions behind any value that matters to you. The book tells you which opinion carries which number.
2. Into the fibre and supplement evidence. The Reynolds series and the Bjelakovic Cochrane review, read together, reorganise a supermarket aisle.
3. Into the soil debate. Read Pittelkow on no-till, then the statistical re-analysis that halves the estimated yield loss. Then Rothamsted, for the longest test of manure against fertiliser in the world.
4. Into diet scenarios. Poore & Nemecek for environmental cost per food; Peters et al. for why a lacto-vegetarian diet can feed more people than a vegan one; EAT-Lancet for the planetary health diet the book uses as its Diet A.
5. Into the kitchen science. The USDA retention factors table, the sourdough-phytate work of Lopez et al., the Hong Kong and FDA rules on kidney beans, and the pH 4.6 rule behind safe home bottling.
6. Into the disputes. Appendix B is itself a research programme: herring vitamin D, lentil folate, cover crops and soil carbon, the mechanism of lime-treated maize. Each entry names both sources.

---

An Honest Reservation

Two things should be said. The Vacuum.Net frame of Part 0 will not be to every taste, and the author knows it; he marks each use of it and repeats that a reading is not a mechanism. The frame can be skipped without loss. And some sources could not be reached — the book says which, and lists what remains open. These are flaws of transparency, not of concealment. They argue in the book’s favour.

---

Conclusion

What Food Does Our Body Need does three things at once, and does them carefully. It states, in plain numbers, what a body must obtain from outside itself. It follows each of those numbers back to a soil, a seed, a season and a pair of hands. And it asks, throughout, what can be done without a long industrial chain. It is disciplined about evidence, honest about disagreement, and checked to a standard most trade books never approach. For the reader who wants a single volume that runs from the micronutrient to the kitchen to the field — and back to the plate — this is a book worth having, and worth following into its sources.

Book

J.Konstapel,Leiden,20-10-2026.

Website : 65Klusser

Een Lamp Ophangen voor €75: Wat er achter 65Klusser zit

Op 20 oktober 2026 meldde Sleutelstad dat 65Klusser ook in Leiden en omgeving begint. Het concept is eenvoudig. Gepensioneerde vakmensen doen kleine klussen bij mensen thuis: een lamp ophangen, een kast in elkaar zetten, een beugel in de douche, een klemmende deur.

Dat voorziet in een echte behoefte. Er blijven in veel huizen klussen liggen. Er zijn veel 65-plussers met handen en ervaring die niet stil willen zitten.

Deze blog zoekt uit wat er achter de dienst zit, waar het geld blijft en welke alternatieven er zijn. Daarna volgt een vergelijking met een heel andere vorm: de MAZE-Bank, een kring waarin daden worden geboekt in plaats van euro’s.

Alle getallen zijn nagerekend. Wat niet gevonden is, staat er als open bij.

Hoe 65Klusser werkt

De dienst werkt in vier stappen.

1. De klant belt of vult een formulier in.
2. Een werkopnemer komt kijken en bespreekt de klus.
3. De klant krijgt een prijsopgave op basis van €75 per uur, inclusief btw. Bij een goed af te bakenen klus kan dat een vaste prijs zijn.
4. Na akkoord krijgt de klant naam en foto van de klusser, per mail of WhatsApp. De klusser komt, doet de klus, en de klant betaalt achteraf.

De klussers hebben een Verklaring Omtrent het Gedrag (VOG). Ze zijn persoonlijk gescreend en verzekerd. Ze komen uit de eigen regio.

De dienst is actief in Haaglanden, Leiden, Rotterdam en het Groene Hart. Daarnaast zijn er lijnen voor VvE’s, horeca en bedrijven.

Wie erachter zit

65Klusser is een handelsnaam. De drager is O. den Hoed Beheer B.V. in Voorburg.

Volgens het handelsregister is die B.V. in maart 2017 opgericht als advies- en consultancyholding, met één werkzame persoon. Dezelfde B.V. voert nog twee handelsnamen: “Digital Nomad Recruiters” en “IntelliFox”.

Die eerste handelsnaam staat als pagina op de site van het recruitmentbureau 24People. Onno den Hoed is volgens een openbaar profiel sinds 2012 directeur-eigenaar van 24People en sinds 2020 medeoprichter van OnJobs. Hij sprak in september 2026 in de lokale media namens 65Klusser.

In de eerste berichten, uit december 2025, heet Gert Barnhoorn (68) de oprichter. Hij vertelde toen op Radio 5 en Den Haag FM over het ontstaan. Hoe Barnhoorn en den Hoed zich tot elkaar verhouden, staat in geen enkele gevonden bron.

De dienst is jong. Begin december 2025 draaide hij “een paar weken”, met zes klussers en ongeveer twintig klanten. Dat waren vooral alleenstaande vrouwen. Flyers in de bus werkten; reclame op sociale media niet. Nu noemt de site ongeveer twintig plaatsen en zeven klusserprofielen.

Het is geen franchise. Uitbreiding loopt via regiopartners “zonder franchisefee”. Die werken ongeveer zestien uur per week voor indicatief €1.700 tot €4.000 per maand.

Waar de €75 heen gaat

De rekensom is kort.

• €75 inclusief btw is 75 / 1,21 = €61,98 exclusief btw.
• Er gaat €13,02 btw naar de fiscus.
• De klusser krijgt volgens de wervingspagina €35 bruto per uur, “all-in”.
• Er blijft 61,98 − 35 = €26,98 per uur voor de B.V. Dat is 43,5% van de omzet exclusief btw.

Dat laatste getal is een bovengrens. De werkopname is gratis voor de klant. Er zijn volgens de site geen voorrijkosten. Werkgeverslasten, verzekering en werving gaan er nog af.

Twee punten zijn onduidelijk.

• Voorrijkosten. De site zegt “geen voorrijkosten”. Sleutelstad schrijft “exclusief voorrijkosten”, zonder bedrag. Parkeerkosten komen in elk geval apart.
• Contractvorm. Of de klusser zzp’er is, oproepkracht of via een payrollbedrijf werkt, staat nergens.

Ook de regiopartner is na te rekenen. Zestien uur per week is 16 × 52 / 12 = 69,3 uur per maand. Dat geeft €24,52 tot €57,69 per uur. Voor €4.000 per maand zijn bij de volle marge van €26,98 ongeveer 148 klusseruren per maand nodig.

Wat de dienst goed doet

Het model lost drie dingen tegelijk op.

Werk voor ouderen met vakmanschap. Veel gepensioneerde timmerlieden, schilders en monteurs willen nog iets doen. Een paar uur per week is genoeg. €35 bruto per uur is een redelijk bedrag voor een bijbaan na de AOW-leeftijd.

Vertrouwen voor wie alleen woont. De eerste klanten waren vooral alleenstaande vrouwen. Voor hen telt het om te weten wie er aanbelt. VOG, screening en een foto vooraf geven dat.

De kleine klus wordt gedaan. Een gewone aannemer komt niet voor een gordijnrail. Zo’n klus blijft vaak jaren liggen.

Dat is de kern van het succes. De prijs is de prijs van die zekerheid.

De alternatieven

Voor dezelfde behoefte bestaat veel. De verschillen zitten vooral in wie het werk doet en voor wie het bedoeld is. Hieronder een klus van twee uur, nagerekend.

Alternatief	Regio	Twee uur	Wie doet het	Voor wie
HipHelpt	Leiden, Oegstgeest	€0	vrijwilliger	geen netwerk én weinig geld; tot ca. 2 uur
Welzijnskwartier, klussendienst	Katwijk	€10	vrijwilliger	senioren
WOEJ	Leidschendam-Voorburg	€12,50	vrijwilliger	inwoners
Radius, klussen in huis	Leiden, Oegstgeest	€60 + materiaal	welzijnsorganisatie	ouderen, mensen met een beperking
Hups, met ANBO-PCOB-korting	landelijk	€96,95	klushulp via platform	leden
Klusup	landelijk	vanaf €110	zelfstandige, ook gepensioneerden	iedereen
De Klusbakfiets	o.a. Den Haag	€117,37	vakman	iedereen
Hups, standaard	landelijk	€121,95	klushulp via platform	iedereen
Zoofy	landelijk	vanaf €130	vakman via platform	iedereen
65Klusser	Haaglanden, Leiden, Rotterdam, Groene Hart	€150	65+-vakman	iedereen
GAMMA Klusservice	landelijk	€154,95	vakman via Zoofy	iedereen

Toelichting bij de rekensom:

• Hups: gemiddeld €41 per uur, minimaal twee uur, plus een eenmalige matchingsfee van €39,95. Voor ANBO-PCOB-leden is die fee €14,95.
• De Klusbakfiets: €48,50 exclusief btw, dus €58,69 inclusief btw.
• GAMMA: €65 per uur plus €24,95 servicekosten.

Er is meer, maar zonder vergelijkbaar uurtarief:

• Altijd Voorhanden werkt met klussers van 45+ in Leiden, Leiderdorp en Den Haag. Voorrijkosten in de regio zijn inbegrepen; het uurtarief staat niet op de site.
• Stichting Present (Leiden, Den Haag, Gouda, Woerden) zet groepen vrijwilligers in, via een hulpverlener.
• Burenhulp van Servicepunt XL in Den Haag heeft meer dan 250 vrijwilligers.
• Gilde Zoetermeer rekent €5 per uur voor minima.
• Het Bruisnest Gouda begint bij €5 per uur.
• Incluzio Rotterdam is gratis, plus materiaal.
• Werkspot is gratis voor de klant; de vakman betaalt €3 tot €75 per lead. Het markttarief daar is €35 tot €50 per uur.
• Repair Café Leiden repareert wat mensen meebrengen, maar komt niet aan huis.

Wat ontbreekt, is ook informatief:

• Een tijdbank in Leiden is niet gevonden. De Time Bank Den Haag toont als laatste datum 2018.
• Mens voor Elkaar in Voorschoten is per 1 augustus 2026 gestopt.
• Een coöperatie die klussen tussen buren regelt, bestaat in de regio niet. De voorbeelden liggen in Zeist, Utrecht en Amsterdam.

Het patroon is helder:

• Wie weinig geld en geen netwerk heeft, kan bij een welzijnsorganisatie terecht.
• Wie geld heeft, kan kiezen uit platforms en bureaus.
• Daartussen zit een grote groep. Die heeft wat geld en een klein netwerk. Voor die groep is 65Klusser gemaakt.

Wat de keten kost

Het visnet helpt om te zien wat hier gebeurt. Mensen zijn knopen. Een klus is een maas die tussen twee knopen sluit: iemand helpt iemand.

Bij 65Klusser loopt die maas niet rechtstreeks. Ze loopt via een keten van rollen:

• klant;
• bureau;
• werkopnemer;
• klusser.

Er zijn vijf stappen, met drie tot vier geldoverdrachten:

• van klant naar B.V.;
• van B.V. naar klusser;
• btw naar de fiscus;
• eventueel loonheffing.

Elke schakel voegt iets toe: werving, inschatting, zekerheid. Elke schakel kost ook iets. Van de €61,98 die exclusief btw binnenkomt, bereikt €35 de man of vrouw die het werk doet.

De klant en de klusser wonen vaak een paar straten van elkaar. Ze leren elkaar kennen. Bij de volgende klus belt de klant toch weer het bureau. De maas sluit, maar niet tussen hen. Ze sluit via de keten.

Dat is geen verwijt aan de dienst. Het is hoe vrijwel elke dienst in de euro-economie werkt. Een overdracht kost geld. Wereldwijd kost een geldoverdracht naar familie in het buitenland nog altijd ongeveer zes procent van het bedrag.

De vergelijking: de MAZE-Bank

De MAZE-Bank is een vorm waarin buren hun onderlinge hulp boeken zonder geld. Hij is beschreven op constable.blog en draait op maze.swarp.nl/bank. Volgens de bronnen werkt hij zo:

• Boeking. Iemand noteert een daad in één zin: wie deed wat voor wie, en wanneer.
• Tegentekening. De daad telt pas als de ontvanger tegentekent met een tik op “klopt”. Daarvoor is geen account nodig.
• Stand. Per relatie is de stand −1, 0 of +1. Die volgt uit de bevestigde daden en wordt nergens als saldo bewaard.
• Verval. Een open stand vervalt naar nul en wordt een gift aan de kring. Er is geen rente. De vervalsnelheid stelt de kring zelf vast.
• Geld. Binnen de kring gaat geen geld rond. Euro’s bestaan alleen aan de rand, voor wat niet in daden kan, zoals materiaal. Dat geld staat bij een vergunninghouder, nooit bij een lid.
• Bestuur. De kring bestuurt zichzelf met consent. Een waarnemer leest de getallen en heeft geen macht.
• Instap. Wie wil beginnen, plakt een bestaand rooster of zorgschema in. Elke regel wordt met een klik bevestigd.

Een 65Klusser-klus ziet er in zo’n kring zo uit:

1. Iemand vraagt in de kring om hulp bij een gordijnrail.
2. Een handige buur zegt toe.
3. De buur hangt de rail op en noteert de daad.
4. De ander tikt “klopt”.
5. De rail zelf wordt in euro’s gekocht, als grenspost.

Binnen de kring kost dat niets. Er is geen tussenpartij die een deel houdt. Verder is de relatie wederkerig. De klusser van vandaag kan over tien jaar zelf de hulp nodig hebben. Zijn bevestigde daden staan dan in de kroniek van de kring.

Wat de ene heeft en de andere niet

Het is een vergelijking, geen wedstrijd. Beide vormen dragen iets wat de andere mist.

65Klusser heeft:

• Werving. De dienst vindt klussers en brengt ze bij klanten.
• Een vakinschatting vooraf. De werkopnemer ziet wat een klus vraagt.
• VOG en verzekering. Dat geeft zekerheid voor wie een vreemde binnenlaat.
• Een werkende praktijk. Er komen echte klussers bij echte mensen.

De MAZE-Bank heeft:

• Geen tussenpartij. De waarde blijft tussen de twee mensen die de maas sluiten.
• Een bewijs per daad. De ontvanger bevestigt zelf wat er gedaan is.
• Wederkerigheid. Wie helpt, wordt later geholpen.
• Bestuur door de betrokkenen, met consent.
• Vertrouwen via mensen. Bestaande leden tekenen voor nieuwe leden, en borgen tekenen mee.

Daar staat de stand van zaken tegenover. Volgens de handleiding zijn de kern, de koppeling, het bestuursgereedschap en de reconciliatie gebouwd. De live pagina meldt nog “Nog geen daden” en “Nog geen coöperatieve jas”. Het vooroverleg met de Belastingdienst is open. Wat een VOG en een verzekering in de kring vervangt, staat in geen bron.

Het concept rond de eenheid is ook verschoven. Op 8 september 2026 rekende een post nog met Seeds: tien per lid per week, 2 procent verval, een evenwicht van 10 / 0,02 = 500. Een week later spreekt de handleiding van “geen geld, geen saldo — daden”. Een veld seeds=1 bestaat nog als gewicht, niet als munt.

Waar ze elkaar raken

Het interessantste punt zit niet in het verschil. Het zit in de plek waar ze elkaar raken.

Een vaste klusser die steeds terugkomt bij dezelfde klant, is een afspraak die zich herhaalt. Juist zo’n afspraak gebruikt de MAZE-Bank als ingang: een rooster, een zorgschema, een vaste hulp.

65Klusser bouwt die relaties nu op, in twintig plaatsen. Er ontstaan knopen die elkaar kennen. Of de maas tussen hen ooit zonder bureau gaat sluiten, hangt af van wat zo’n kring niet heeft: werving, vakinschatting en de zekerheid die een VOG en een verzekering nu geven.

Wat er verdwijnt als dat lukt, is de marge. Wat er blijft, is de gepensioneerde vakman die een rail ophangt bij een buurvrouw drie straten verder. En de kans dat zij, of iemand anders in de kring, later iets terugdoet wat hij nodig heeft.

Wat open blijft

• De inhoud van de algemene voorwaarden van 65Klusser: aansprakelijkheid, betaaltermijn, annulering.
• De contractvorm van de klussers.
• De verhouding tussen Gert Barnhoorn en Onno den Hoed.
• Of er voorrijkosten zijn, en hoeveel.
• De uurtarieven van Altijd Voorhanden, Onze Klusvrouw en Hups in Leiden.
• In de MAZE-Bank: wat een VOG en een verzekering vervangt, of een offerte een ingang kan zijn, en hoe de Belastingdienst een daad behandelt.

---

Referenties

Over 65Klusser

65Klusser — website. https://www.65klusser.nl Waarom lezen? De werkwijze in vier stappen, het tarief en de regio’s, uit de eerste hand. Leesadvies: begin bij de veelgestelde vragen.

65Klusser — Word klusser. https://www.65klusser.nl/word-klusser Waarom lezen? Hier staat de vergoeding van €35 bruto all-in per uur. Dat is de enige bron voor de verdeling van de €75.

65Klusser — Regiopartner. https://www.65klusser.nl/regiopartner Waarom lezen? Het uitbreidingsmodel: zestien uur per week, indicatief €1.700–€4.000 per maand, zonder franchisefee.

65Klusser — In de media. https://www.65klusser.nl/in-de-media Waarom lezen? Het overzicht van radio, krant en lokale media, met de namen van wie er namens de dienst spreekt.

Seniorenjournaal, “65Klusser met businessmodel” (6-12-2025, op basis van het AD). https://seniorenjournaal.nl/diensten/65klusser-met-businessmodel/ Waarom lezen? Het ontstaan, de eerste zes klussers, de klantengroep en waarom flyers beter werkten dan sociale media. Leesadvies: de beste korte bron over de begintijd.

Sleutelstad, “65Klusser biedt hulp bij kleine klussen in Leiden en omgeving” (20-10-2026). https://sleutelstad.nl/2026/10/20/65klusser-biedt-hulp-bij-kleine-klussen-in-leiden-en-omgeving/ Waarom lezen? De aanleiding van deze blog, en de bron van de vermelding “exclusief voorrijkosten”.

Midvliet, “65-plusklussers blijven actief en helpen anderen met klussen” (30-9-2026). https://midvliet.nl/podcasts/midvliet-actueel/34502-65-plusklussers-blijven-actief-en-helpen-anderen-met-klussen Waarom lezen? Onno den Hoed over het idee dat het mes aan twee kanten snijdt.

Bedrijvenregister en Oozo — O. den Hoed Beheer B.V. https://www.bedrijvenregister.nl/zoekresultaten?q_source=bedrijven&q=68405472 en https://www.oozo.nl/bedrijven/s-gravenhage/binckhorst/binckhorst/1449778/o-den-hoed-beheer-b-v Waarom lezen? De handelsnamen, de activiteit en het oprichtingsjaar van de juridische drager.

24People — Digital Nomad Recruiters. https://www.24people.nl/digital-nomad-recruiters/ Waarom lezen? De pagina die de B.V. verbindt met het recruitmentbureau.

Werkspot — reviews 65Klusser. https://www.werkspot.nl/profiel/65klusser/reviews Waarom lezen? De enige openbare klantbeoordelingen die op te halen waren.

Over de alternatieven

Radius — Klussen in huis. https://www.radiuswelzijn.nl/welzijn-thuis/klussen-in-huis/ Waarom lezen? Het Leidse welzijnsalternatief, met prijs en doelgroep. Leesadvies: eerste adres voor Leidenaren die op de prijs letten.

Voor Elkaar Leiden — HipHelpt. https://www.voorelkaarleiden.nl/hiphelpt/ Waarom lezen? Gratis hulp, en de voorwaarden die bepalen wie daar terecht kan.

Welzijnskwartier — Klussendienst. https://welzijnskwartier.nl/ondersteunende-diensten/klussendienst/ Waarom lezen? Katwijk laat zien wat een vrijwillige klusdienst voor €5 per uur kan.

WOEJ — Klussendienst. https://www.woej.nl/themas/hulp-advies/klussendienst/ Waarom lezen? Hetzelfde voor Leidschendam-Voorburg, de thuisgemeente van 65Klusser.

Stichting Present. https://stichtingpresent.nl/leiden/ Waarom lezen? Een landelijk netwerk dat groepen vrijwilligers aan hulpvragen koppelt, via hulpverleners.

Altijd Voorhanden — Werkwijze klusservice. https://altijdvoorhanden.nl/werkwijze-klusservice/ Waarom lezen? De naaste commerciële tegenhanger, met klussers van 45+.

Leef3 — Hups klushulp. https://leef3.nu/dienst/hups-klushulp/ Waarom lezen? De volledige prijsopbouw van Hups, inclusief matchingsfee en kilometervergoeding. Leesadvies: nuttig vóór elke vergelijking van platforms op hun “vanaf”-prijs.

Klusup — Klusjesman Leiden. https://www.klusup.nl/klusjesman/leiden Waarom lezen? Een platform waar gepensioneerden zich zelf als klusser kunnen aanmelden, zonder bureau ertussen.

GAMMA — Klusprofessional per uur. https://www.gamma.nl/klusservice/klus-professional-per-uur Waarom lezen? Het tarief van een bouwmarkt, als ijkpunt aan de bovenkant.

Werkspot — Prijs klusjesman per uur. https://www.werkspot.nl/algemene-klussen/prijzen-kosten/klusjesman-prijs-per-uur Waarom lezen? Het markttarief en, in de FAQ, wie er voor een lead betaalt.

Haags Steunsysteem — Verhuizen en klussen, opties voor hulp (15-9-2024). https://haagssteunsysteem.nl/documents/d/haagssteunsysteem/verhuizen-klussen-opties-hulp-15-09-24-pdf?download=true Waarom lezen? Een overzicht van alle Haagse klushulp op één blad. Zo’n overzicht ontbreekt voor Leiden.

Sociale Vraagstukken — Zorgzame buurten (2023). https://www.socialevraagstukken.nl/wp-content/uploads/2023/05/Zorgzame-buurten.pdf Waarom lezen? Achtergrond bij buurtcoöperaties en waarom die zo zelden van de grond komen.

Over de MAZE-Bank

MAZE-Bank — de dienst en de handleiding. https://maze.swarp.nl/bank en https://maze.swarp.nl/bank/handleiding Waarom lezen? Wat er gebouwd is en wat nog open staat, uit de eerste hand. Leesadvies: lees de handleiding vóór de blogs.

“De Bank die geen geld bewaart” (19-9-2026). https://constable.blog/2026/09/19/de-bank-die-geen-geld-bewaart/ Waarom lezen? De MAZE-Bank naast een gewone bank, met toetreding via leden die voor elkaar tekenen.

“The Use of Deontic Logic in the MAZE-Bank” (11-9-2026). https://constable.blog/2026/09/11/the-use-of-deontic-logic-in-the-maze-bank/ Waarom lezen? Waarom de bank geen straf kent en een daad een verplichting is die door een tik wordt afgesloten. Leesadvies: voor wie de formele kant wil.

“Van Plato tot de Bon” (8-9-2026). https://constable.blog/2026/09/08/van-plato-tot-de-bon/ Waarom lezen? Het moment waarop de eenheid verdwijnt en alleen de daad overblijft.

“The MAZE Bank” (8-9-2026). https://constable.blog/2026/09/08/the-maze-bank/ Waarom lezen? De laatste versie met Seeds en verval. Lees hem naast de vorige om de omslag te zien.

“De Kring die Ontbreekt” (15-10-2026). https://constable.blog/2026/10/15/de-kring-die-ontbreekt/ Waarom lezen? Waarom de kring zelf het ontbrekende stuk is, niet het geld.

“The Bank of the Future” (21-8-2026). https://constable.blog/2026/08/21/the-bank-of-the-futurea-community-float-a-decay-rate-and-a-ledger-that-does-not-add-up/ Waarom lezen? De eerste uitwerking, met de float van een gemeenschap en de rekenregels voor verval.

“Maze: De Werkelijke Prijs van Een Euro” / “The Price of a Euro” (7-9-2026). https://constable.blog/2026/09/07/the-price-of-a-euromoney-context-and-the-second-route/ Waarom lezen? Elke overdracht lekt; hier staat het getal van ongeveer zes procent voor overmakingen naar familie, en de tweede route zonder euro.

“Breaking the Chain of Money” (29-6-2023). https://constable.blog/2023/06/29/breaking-the-chain-of-money/ Waarom lezen? De oorsprong van het ketenargument: geld als intermediair, en betalen als het herstellen van een balans.

“About the Cooperative” (3-4-2009). https://constable.blog/2009/04/03/about-cooperative/ Waarom lezen? Rochdale en Raiffeisen, en waarom coöperaties op menselijke schaal moeten blijven om te slagen. Leesadvies: de historische achtergrond bij elke kring.

J.Konstapel,Leiden, 20-10-2026.

Light, Location and the Oldest Dao as the Ground of the Vacuum.Net

© 2026 J. Konstapel, Constable Research, Leiden. All rights reserved. Quotation with attribution is permitted; reproduction of the whole requires written permission.

---

1. Why go back

The Vacuum.Net theory rests on five axioms. There is one strand (N1). Only what closes persists (N2). A closed configuration carries memory (N3). The response of the medium is a state function (N4). Each step deeper multiplies the states by three (N5).

These axioms work. They give a ternary register, a closure rule, a depth ladder and numbers that land on hydrogen, room heat and the Earth’s cavity.

Yet they carry a hidden second actor. A strand that crosses, winds and closes is something that travels over places. Then there are two things: the places and the carrier that runs along them. The rewrite rule R has the same problem. It is applied to a ring, but by what?

This essay removes the second actor. It starts from one statement: the only thing there is, is light. It then follows that statement through four steps. Light is a location. A location is a point with a bound movement. That movement combines with itself into all three trit values. And the oldest Chinese form of the character 道, together with the oldest Daoist account of origin, already contains the result.

The outcome is a theory with three rules instead of five axioms. It is simpler. It is also older.

Every step carries a status. Proved means a mathematical result that holds under its hypotheses. Computed means a number checked by arithmetic. Source means a published text or inscription. Reading means an interpretation offered here. Open means not yet settled.

---

2. The units are already one photon of light

The Foundational Paper (Edition 8.1) fixes two conversion units from hydrogen.

The length unit is S₀ = 1/(2R_H). With the reduced-mass Rydberg constant R_H = 10,967,758.340 m⁻¹, this gives S₀ = 4.55882 × 10⁻⁸ m, or 45.588 nm.

The time unit is T₀ = S₀/c. With c = 299,792,458 m/s, this gives T₀ = 1.52066 × 10⁻¹⁶ s.

The energy unit is E₀ = h/T₀. Write it out:

E₀ = h/T₀ = hc/S₀ = (1.98645 × 10⁻²⁵ J·m) / (4.55882 × 10⁻⁸ m) = 4.35737 × 10⁻¹⁸ J = 27.197 eV.

The paper names E₀ the Hartree energy of hydrogen. It is equally the energy of one photon of wavelength S₀. Length, time and energy of the net are together one quantum of light. Computed.

Two consequences follow at once.

First, the power ceiling of the net, one closure per mesh per layer time, is E₀/T₀ = 0.0287 W. That is one photon per mesh per period. Computed.

Second, the depth ladder 3ⁿ·S₀ becomes a ladder of one photon read at other depths. At depth 2 the length is 9·S₀ = 410.29 nm, the Balmer Hδ line. At depth 5 it is 243·S₀ = 11.08 µm, close to the peak of room-temperature radiation. At depth 31 it is 3³¹·S₀ = 2.816 × 10⁷ m. The lowest ideal mode of the Earth–ionosphere cavity has wavelength 2πR/√2 = 2.831 × 10⁷ m, for R = 6,371 km. The two differ by 0.5 %. Computed.

All three landings are light. The ladder does not pass through different kinds of phenomena. It passes through one spectrum.

---

3. Light is a location, not a motion

The first impulse is to call light the unit motion of the net. One mesh per layer time gives S₀/T₀ = c by construction.

That definition fails. If light is defined as motion at c, it cannot also be the stuff of matter at rest. And “the motion of the net” presupposes a net that moves. Then light is again something other than the net.

The correction is this: light is a location.

Dewey Larson stated the same idea in 1959. In his natural reference system the photon has no motion of its own. It stays at its absolute location. The reference system itself progresses at unit speed. Larson’s dimensions are already part of the Foundational Paper. Source; assumed as postulate.

Relativity offers an echo but not a ground. Along a light path the interval is zero: ds² = 0. This follows from Einstein’s second postulate, so it restates an assumption. A photon also has no rest frame, so “what light experiences” lies outside that theory. Moreover relativity takes c as a universal constant, while N4 treats c as a state function. Relativity is therefore at most a correspondence here. Correspondence.

In the net, Theorem T7 gives every state exactly one address, a balanced-ternary string. Proved. If light is a location, light is an address. The 45.6 nm photon is the address at depth 0. Hδ is the same address at depth 2.

This reverses the earlier picture.

• Light stays at its location and is carried along by the progression.
• Matter is a location that holds itself against that progression. That is closure: what returns in phase stays put while everything else moves on. Larson calls it rotation opposing progression.
• c is the measure of the difference between the two: one mesh per layer time. That is why matter always measures light at c.

The old question “can light rest?” disappears. Light always rests, in its own terms. What we call motion is matter holding on while the locations progress. Reading.

---

4. The point and the stroke

The character 道 makes the next step visible. It has 12 strokes in the standard count. It consists of 首, “head”, with 9 strokes, and 辶, “going”, with 3 strokes. The Kangxi tradition counts 辶 as 4 strokes, giving 13. Source.

The stroke order of 道 is fixed. A character drawn in another order counts as wrong. The order belongs to the character. Source.

The order runs as follows. Strokes 1 to 9 write 首: 丶 丿 一, then 自. Strokes 10 to 12 write 辶: a dot, a folded stroke and a long flat sweep. The component 辶 stands on the left and encloses the whole from below. Yet it is written last. First the place, then the going that carries the place.

Both parts begin with a dot. Stroke 1 is 丶. Stroke 10 is 丶. Source.

A dot in calligraphy is not a static speck. It is a short stroke: set down, a small direction, lift. The dot is movement of the smallest extent.

So there is one ground act, the stroke, in two states.

• Point: movement that closes at once. This is location. This is light.
• Stroke: movement that continues.

Every movement in 道 starts from a point. Never the reverse. A form is not a configuration but a fixed sequence of acts. Its identity lies in the order of making. This is N3, memory as shape, read in writing. It is also T7: the same trits in another order give another address. Reading.

No second actor is needed to draw. The order is the thing itself, as the stroke order is the character.

---

5. Random, but bound to the point

The movement that leaves a point need not be directed. It may be random. The question is whether it stays bound to its point.

Model the movement as a random string of trits. Each step is +1, 0 or −1 with equal chance. “Bound to the point” means the running sum comes back to 0.

Return after three steps. Count the strings of length 3.

• Exact return, sum = 0: the string 000 and the six permutations of (1, 0, −1). That is 7 of 27.
• Return modulo 3, the window rule of T9: 9 of 27. The two extra strings are 111 and TTT, with sum +3 and −3.

Computed. The numbers 1, 1, 3, 7, 19, 51, 141, … are the central trinomial coefficients. They count exact returns after 0, 1, 2, 3, … steps.

The difference between 7 and 9 matters. Exact return means coming back along the line. The strings 111 and TTT do not come back on the line. They go once around a ring of three. The register counts both as closure. There are two kinds of binding to the point: coming back and going round.

One dimension is always bound. The chance of exact return after n steps behaves as √3 / (2√(πn)). Each step has variance 2/3. The local limit gives 1/√(2πn·2/3) = √3/(2√(πn)).

That chance goes to zero. But the sum over all n diverges. A random movement along one line therefore returns to its point with certainty, and does so infinitely often. Proved: Pólya, 1921. It holds for every step law with mean zero and finite variance.

In three dimensions this fails. There the return probability is P₃ = 0.3405 (Watson, 1939; Theorem T18 of the Foundational Paper). Source.

The consequence is direct. “Random, but bound to the point” needs no force and no second actor. A movement with only one direction, forward and back, stays tied to its point by itself. That is N1, one strand, read without a strand.

• Light is the point.
• The movement is random but one-dimensional, and therefore bound.
• Matter arises where the movement does not merely return, but returns in phase.

Reading, on top of the proved recurrence.

---

6. The act combined with itself

Write the ground act as “. >”: leaving a place. Its mirror is “< .”: arriving at a place.

The mirror need not be given separately. Give “. >” the value +1 and repeat it, counting modulo 3.

• once: +1;
• twice: 1 + 1 = 2 ≡ −1, the mirror “< .”;
• three times: 1 + 1 + 1 = 3 ≡ 0.

Proved: the group ℤ₃ is cyclic and generated by one element. From one act all three trit values arise. This works only on the ring of three. On the plain line, 1 + 1 stays 2.

There are two roads to 0.

• Meeting: “. > < .”, the act meeting its mirror: +1 + (−1) = 0.
• Going round: the same act three times: 111 or TTT.

These are exactly the two kinds of closure in Theorem T10. A window closes when its three trits are all different (heterogeneous) or all equal (homogeneous). Proved.

So the 0 is neither emptiness nor rest. It is leaving and arriving cancelling in the sum while both exist. In terms of light it is a movement between two places, there and back: a standing wave with the points as nodes. That is the simplest bound light. The mesh is the space between a leaving and an arriving. Reading.

Order still counts. “> <” (towards each other) and “< >” (away from each other) both have sum 0. They differ. The first closes. The second opens. The sum does not see the difference. The order does, as in the stroke order of 道. Open: the register so far measures sums only. It needs a quantity that records order.

---

7. The oldest Dao: a crossing with someone in it

The modern 道 is a contraction. Older forms are written 衟 or 衜. Lexicographers read them as a person standing at a crossroads 行. The head 首 sits inside the crossroads. At the bottom stands 止, the foot. Source.

The component 行 carries the meanings walking, passage, birth canal and crossroads. The pairing of 首 with 止 can stand for a person. Source.

The oldest 道 thus holds three things in one sign.

• 行, the crossing: the self-crossing of N1.
• 首, the place in the crossing: the point, the location, light.
• 止, the foot: the step, the movement bound to the place.

There is no carrier besides these. The crossing, the place and the step are given together. Source for the graph; reading for the mapping.

---

8. The oldest account of origin: arising by return

In 1993 a tomb at Guodian in Hubei yielded bamboo slips from around 300 BCE. They hold the oldest known Laozi material. Beside it lies a short text of 14 slips, the Taiyi sheng shui, “The Great One Gave Birth to Water”. It has not been found anywhere else. Source.

Its opening reads:

太一生水，水反輔太一，是以成天。天反輔太一，是以成地。天地復相輔也，是以成神明。神明復相輔也，是以成陰陽。陰陽復相輔也，是以成四時。四時復相輔也，是以成寒熱。寒熱復相輔也，是以成濕燥。濕燥復相輔也，成歲而止。

In translation:

The Great One gave birth to water. Water returned and assisted the Great One; thus heaven was completed. Heaven returned and assisted the Great One; thus earth was completed. Heaven and earth assisted each other again; thus the spirits above and below were completed. These assisted each other again; thus yin and yang were completed. Yin and yang assisted each other again; thus the four seasons were completed. The four seasons assisted each other again; thus cold and heat were completed. Cold and heat assisted each other again; thus moist and dry were completed. Moist and dry assisted each other again; the year was completed, and it stopped there.

The text then reads the chain backwards. The year is born of moist and dry, moist and dry of cold and heat, and so on back to the Great One. Source.

Then come three sentences that hold the whole mechanism:

是故太一藏於水，行於時，周而又始，以己為萬物母；一缺一盈，以己為萬物經。

Therefore the Great One hides in water and moves through the seasons. It goes round and begins again, and makes itself the mother of the ten thousand things. Once lacking, once full, it makes itself the warp of the ten thousand things.

Source. The word 經 means the warp: the fixed threads of a loom through which everything else is woven.

Chapter 42 of the received Laozi, “One gives birth to two, two to three, three to the ten thousand things”, does not occur in the Guodian Laozi. The older cosmogony does not count. It lets things arise by return and by mutual support. Source.

---

9. Three rules instead of five axioms

The material of sections 2 to 8 yields three rules.

R1 — Crossing, place and step (道). There is one movement that leaves a place through a crossing. The place is light. The movement is bound to it.

R1 replaces N1. It needs no strand as a separate carrier. The crossing, the place and the step are given together, as in 衟.

R2 — Return gives birth (反輔). What arises returns to its source, and from that return the next thing arises.

R2 replaces N2. In N2 closure only selects. In R2 closure also produces. Water returns to the One, and heaven is completed. In register terms this is “. >” coming back to its point.

R3 — Mutual support (相輔). Two things that have arisen support each other, and from that support the next layer arises.

R3 is the meeting “. > < .”, with sum 0, but productive. Repeated, it gives the ladder of N5: each layer arises from the pair of the layer before.

What N3 and N4 did is already present.

• Memory (N3). The text reads the chain backwards, each thing with its source. The order of arising is the form, as the stroke order is the character.
• Alternation. “Once lacking, once full” is +1/−1. “Goes round and begins again” is the cycle of R. The year is the first complete closure.
• Context (N4). The rest of the text says that the way of heaven values the weak, cuts what is complete to increase what is growing. The response depends on the state, not on a fixed law. Correspondence.

The image is the same as before. The One is the warp of all things and hides in water. That is the net spanned in water, in the oldest source. Correspondence.

Why this is older and simpler: R1–R3 need no strand, no external rewrite and no selector. Return presupposes a place to return to. That place is light. Arising presupposes a movement bound to that place. That movement is the step of 止. Reading.

---

10. What changes in the theory

Several statements of the Foundational Paper shift.

• N1 speaks of an oriented, continuous strand. It becomes a sequence of locations with a fixed order of arising.
• N2 says what does not close does not last. Free light lasts billions of years. Under R2, closure means return in phase, and a cycle of R also returns. Light lasts because it is periodic. Matter lasts because it holds.
• N4 makes c a state function. If light is the unit, c is 1 in native units. That c varies means the step from address to address per rewrite varies locally.
• C6 reads 0 as rest. That reading lapses. Rest is now the property of light, not a trit value. The 0 is the meeting of an act with its mirror.
• R is no longer applied from outside. It is the form reading itself. Its fixed points (T14) are locations that stay equal under self-reference. Kauffman’s eigenform O(A) = A states the same.

Reading for all five.

---

11. Open questions

Order. The register records sums. “> <” and “< >” have the same sum but differ. A quantity that records order is needed.

Dimension. A random movement along one line is bound with certainty. In three dimensions the return chance is 0.3405. If space consists of one-dimensional bound movements, where do the three dimensions come from? This is probably where binding and closure part ways.

The nine products. From water to the year, the Taiyi sheng shui names nine products: water, heaven, earth, the spirits, yin-yang, the four seasons, cold-heat, moist-dry and the year. The register has nine closing windows. Whether these belong together must be computed first. It must not be read in.

The third value. Varela (1975) extended Spencer-Brown’s calculus with a third value beside marked and unmarked: the autonomous value of re-entry. That gives three values from distinction alone. How they map onto +1, 0 and −1 is open.

---

12. Summary

The only thing there is, is light. The units of the net are already one photon: 45.6 nm, 1.52 × 10⁻¹⁶ s, 27.2 eV.

Light is a location. Matter is a location that holds against the progression. c measures the difference.

A location is a point with a bound movement. Along one line, a random movement returns to its point with certainty.

The act combined with itself gives all three trit values. Meeting and going round are the two kinds of closure.

The oldest 道 shows a person at a crossing with a foot. The oldest Daoist cosmogony lets everything arise by return and by mutual support, and calls the One the warp hidden in water.

Three rules follow. One movement leaves a place through a crossing. Return gives birth. Mutual support builds the ladder. The net is not threads and meshes. It is locations referring to locations.

---

Annotated references

Konstapel, J. (2026). The Vacuum.Net Theory: Foundational Paper. Edition 8, version 8.1. Constable Research, Leiden. Why read? The five axioms, the window register (T9–T14), the unit derivation (VII.2) and the free-path lever (T18) that this essay rewrites. Reading advice: Part 0, then Parts III.4 and VII.2.

Konstapel, J. (2026). “The Possibilities and Impossibilities of the Net.” constable.blog, 19 October 2026. Why read? The fifty-cell map of what the net can do, with the power ceiling and the depth-ladder landings used in section 2.

Larson, D. B. (1959). The Structure of the Physical Universe. North Pacific Publishers, Portland. Why read? Space and time as two aspects of one motion. The photon at rest in the natural reference system; matter as rotation against the progression. Reading advice: the opening chapters on the natural reference system.

Spencer-Brown, G. (1969). Laws of Form. Allen & Unwin, London. Why read? The mark of distinction and re-entry. A location as a distinction needs no carrier. Reading advice: chapters 1–2 and 11.

Kauffman, L. H. (2005). “EigenForm.” Kybernetes 34(1/2), 129–150. Why read? The eigenform O(A) = A as a form that reproduces itself. The basis for reading R as self-reading rather than an external rewrite.

Varela, F. J. (1975). “A calculus for self-reference.” International Journal of General Systems 2, 5–24. Why read? Adds a third, autonomous value to Spencer-Brown’s two. A ternary ground from distinction alone. Varela was also a practitioner of meditation, not only a formalist.

Pólya, G. (1921). “Über eine Aufgabe der Wahrscheinlichkeitsrechnung betreffend die Irrfahrt im Straßennetz.” Mathematische Annalen 84, 149–160. Why read? The proof that a random walk returns with certainty in one and two dimensions, and not in three. The ground of section 5.

Watson, G. N. (1939). “Three triple integrals.” Quarterly Journal of Mathematics 10, 266–276. Why read? The exact return probability P₃ = 0.3405 of the three-dimensional walk.

OEIS A002426. Central trinomial coefficients. Why read? The sequence 1, 1, 3, 7, 19, 51, 141, … that counts exact returns of the trit walk, with its asymptotics.

Knuth, D. E. (1997). The Art of Computer Programming, Vol. 2, 3rd ed. Addison-Wesley. Why read? Balanced ternary and its symmetries, including truncation as rounding. Reading advice: pp. 195–213.

Jingmen Municipal Museum (1998). Guodian Chu mu zhujian 郭店楚墓竹简. Wenwu, Beijing. Why read? The photographic edition and transcription of the Guodian slips, including the Laozi material and the Taiyi sheng shui.

Allan, S. & Williams, C. (eds.) (2000). The Guodian Laozi: Proceedings of the International Conference, Dartmouth College, May 1998. Society for the Study of Early China, Berkeley. Why read? Scholarly discussion of the Guodian Laozi and the first wide treatment of the Taiyi sheng shui in English.

Henricks, R. G. (2000). Lao Tzu’s Tao Te Ching: A Translation of the Startling New Documents Found at Guodian. Columbia University Press, New York. Why read? Chapter-by-chapter translation of the Guodian Laozi, with the ancient and modern graphs. Shows which received chapters are absent, including chapter 42.

Wenlin Institute. Dictionary entry 道 (dào). Why read? The older graphs 衟/衜, read as a person at a crossroads 行 with the foot 止 below. The paleographic basis of section 7.

Williamson, J. G. & van der Mark, M. B. (1997). “Is the electron a photon with toroidal topology?” Annales de la Fondation Louis de Broglie 22(2), 133–160. Why read? The electron as light closed on itself in a double loop. Matter as held light, in a physical model.

Laozi, Daodejing, chapters 11, 25 and 42. Why read? The useful empty centre of the hub (11), the way that goes round (25) and the counting cosmogony (42), to set beside the older Taiyi sheng shui.

J.Konstapel,Leiden, 20-10-2026.

Ex-Plane of : The Possibilities and Impossibilities of the Net

The Way Before the Net: Light, Location, and the Three Rules of Arising

1. Why Go Back

The Vacuum.Net theory rests on five axioms. There is one strand (N1). Only what closes persists (N2). A closed configuration carries memory (N3). The response of the medium is a state function (N4). Each step deeper multiplies the states by three (N5).

These axioms work. They give a ternary register, a closure rule, a depth ladder, and numbers that land on hydrogen, room heat, and the Earth’s cavity.

Yet they carry a hidden second actor. A strand that crosses, winds, and closes is something that travels over places. Then there are two things: the places and the carrier that runs along them. The rewrite rule R has the same problem. It is applied to a ring, but by what?

This essay removes the second actor. It starts from one statement: the only thing there is, is light. It then follows that statement through four steps. Light is a location. A location is a point with a bound movement. That movement combines with itself into all three trit values. And the oldest Chinese form of the character 道, together with the oldest Daoist account of origin, already contains the result.

The outcome is a theory with three rules instead of five axioms. It is simpler. It is also older.

Every step carries a status. Proved means a mathematical result that holds under its hypotheses. Computed means a number checked by arithmetic. Source means a published text or inscription. Reading means an interpretation offered here. Open means not yet settled.

2. The Units Are Already One Photon of Light

The Foundational Paper (Edition 8.1) fixes two conversion units from hydrogen.

The length unit is S₀ = 1/(2R·H). With the reduced-mass Rydberg constant R·H = 10,967,758.340 m⁻¹, this gives S₀ = 4.55882 × 10⁻⁸ m, or 45.588 nm.

The time unit is T₀ = S₀/c. With c = 299,792,458 m/s, this gives T₀ = 1.52066 × 10⁻¹⁶ s.

The energy unit is E₀ = h/T₀. Write it out:

E₀ = h/T₀ = hc/S₀ = (1.98645 × 10⁻²⁵ J·m)/(4.55882 × 10⁻⁸ m) = 4.35737 × 10⁻¹⁸ J = 27.197 eV.

The paper names E₀ the Hartree energy of hydrogen. It is equally the energy of one photon of wavelength S₀. Length, time, and energy of the net are together one quantum of light. Computed.

Two consequences follow at once.

First, the power ceiling of the net—one closure per mesh per layer time—is E₀/T₀ = 0.0287 W. That is one photon per mesh per period. Computed.

Second, the depth ladder 3ⁿ·S₀ becomes a ladder of one photon read at other depths. At depth 2 the length is 9·S₀ = 410.29 nm, the Balmer Hδ line. At depth 5 it is 243·S₀ = 11.08 μm, close to the peak of room-temperature radiation. At depth 31 it is 3³¹·S₀ = 2.816 × 10⁷ m. The lowest ideal mode of the Earth-ionosphere cavity has wavelength 2πR/√2 = 2.831 × 10⁷ m, for R = 6,371 km. The two differ by 0.5%. Computed.

All three landings are light. The ladder does not pass through different kinds of phenomena. It passes through one spectrum.

3. Light Is a Location, Not a Motion

The first impulse is to call light the unit motion of the net. One mesh per layer time gives S₀/T₀ = c by construction.

That definition fails. If light is defined as motion at c, it cannot also be the stuff of matter at rest. And “the motion of the net” presupposes a net that moves. Then light is again something other than the net.

The correction is this: light is a location.

Dewey Larson stated the same idea in 1959. In his natural reference system the photon has no motion of its own. It stays at its absolute location. The reference system itself progresses at unit speed. Larson’s dimensions are already part of the Foundational Paper. Source; assumed as postulate.

Relativity offers an echo but not a ground. Along a light path the interval is zero: ds² = 0. This follows from Einstein’s second postulate, so it restates an assumption. A photon also has no rest frame, so “what light experiences” lies outside that theory. Moreover relativity takes c as a universal constant, while N4 treats c as a state function. Relativity is therefore at most a correspondence here. Correspondence.

In the net, Theorem T7 gives every state exactly one address, a balanced-ternary string. Proved. If light is a location, light is an address. The 45.6 nm photon is the address at depth 0. Hδ is the same address at depth 2.

This reverses the earlier picture.

Light stays at its location and is carried along by the progression. Matter is a location that holds itself against that progression. That is closure: what returns in phase stays put while everything else moves on. Larson calls it rotation opposing progression. c is the measure of the difference between the two: one mesh per layer time. That is why matter always measures light at c.

The old question “can light rest?” disappears. Light always rests, in its own terms. What we call motion is matter holding on while the locations progress. Reading.

4. The Point and the Stroke

The character 道 makes the next step visible. It has 12 strokes in the standard count. It consists of 首, “head,” with 9 strokes, and 辶, “going,” with 3 strokes. The Kangxi tradition counts 辶 as 4 strokes, giving 13. Source.

The stroke order of 道 is fixed. A character drawn in another order counts as wrong. The order belongs to the character. Source.

The order runs as follows. Strokes 1 to 9 write 首: a dot, then a horizontal, then 自. Strokes 10 to 12 write 辶: a dot, a folded stroke, and a long flat sweep. The component 辶 stands on the left and encloses the whole from below. Yet it is written last. First the place, then the going that carries the place.

Both parts begin with a dot. Stroke 1 is a dot. Stroke 10 is a dot. Source.

A dot in calligraphy is not a static speck. It is a short stroke: set down, a small direction, lift. The dot is movement of the smallest extent.

So there is one ground act, the stroke, in two states.

• Point: movement that closes at once. This is location. This is light.
• Stroke: movement that continues.

Every movement in 道 starts from a point. Never the reverse. A form is not a configuration but a fixed sequence of acts. Its identity lies in the order of making. This is N3, memory as shape, read in writing. It is also T7: the same tries in another order give another address. Reading.

No second actor is needed to draw. The order is the thing itself, as the stroke order is the character.

5. Random, but Bound to the Point

The movement that leaves a point need not be directed. It may be random. The question is whether it stays bound to its point.

Model the movement as a random string of tries. Each step is +1, 0, or −1 with equal chance. “Bound to the point” means the running sum comes back to 0.

Return after three steps. Count the strings of length 3.

• Exact return, sum = 0: the string 000 and the six permutations of (1, 0, −1). That is 7 of 27.
• Return modulo 3, the window rule of T9: 9 of 27. The two extra strings are 111 and TTT, with sum +3 and −3.

Computed. The numbers 1, 1, 3, 7, 19, 51, 141, … are the central trinomial coefficients. They count exact returns after 0, 1, 2, 3, … steps.

The difference between 7 and 9 matters. Exact return means coming back along the line. The strings 111 and TTT do not come back on the line. They go once around a ring of three. The register counts both as closure. There are two kinds of binding to the point: coming back and going round.

One dimension is always bound. The chance of exact return after n steps behaves as √3/(2√(πn)). Each step has variance 2/3. The local limit gives 1/√(2πn·2/3) = √3/(2√(πn)).

That chance goes to zero. But the sum over all n diverges. A random movement along one line therefore returns to its point with certainty, and does so infinitely often. Proved: Pólya, 1921. It holds for every step law with mean zero and finite variance.

In three dimensions this fails. There the return probability is P₃ = 0.3405 (Watson, 1939; Theorem T18 of the Foundational Paper). Source.

The consequence is direct. “Random, but bound to the point” needs no force and no second actor. A movement with only one direction, forward and back, stays tied to its point by itself. That is N1, one strand, read without a strand.

Light is the point. The movement is random but one-dimensional, and therefore bound. Matter arises where the movement does not merely return, but returns in phase.

Reading, on top of the proved recurrence.

6. The Act Combined with Itself

Write the ground act as “>” : leaving a place. Its mirror is “<” : arriving at a place.

The mirror need not be given separately. Give “>” the value +1 and repeat it, counting modulo 3.

• Once: +1
• Twice: 1 + 1 = 2 ≡ −1, the mirror “<“
• Three times: 1 + 1 + 1 = 3 ≡ 0

Proved: the group ℤ₃ is cyclic and generated by one element. From one act all three trit values arise. This works only on the ring of three. On the plain line, 1 + 1 stays 2.

There are two roads to 0.

• Meeting: “><“, the act meeting its mirror: +1 + (−1) = 0
• Going round: the same act three times: 111 or TTT.

These are exactly the two kinds of closure in Theorem T10. A window closes when its three trits are all different (heterogeneous) or all equal (homogeneous). Proved.

So the 0 is neither emptiness nor rest. It is leaving and arriving cancelling in the sum while both exist. In terms of light it is a movement between two places, there and back: a standing wave with the points as nodes. That is the simplest bound light. The mesh is the space between a leaving and an arriving. Reading.

Order still counts. “><” (towards each other) and “<>” (away from each other) both have sum 0. They differ. The first closes. The second opens. The sum does not see the difference. The order does, as in the stroke order of 道. Open: the register so far measures sums only. It needs a quantity that records order.

7. The Oldest Dao: A Crossing with Someone in It

The modern 道 is a contraction. Older forms are written 行 or 十字 with 首 inside. Lexicographers read them as a person standing at a crossroads 行. The head 首 sits inside the crossroads. At the bottom stands 止, the foot. Source.

The component 行 carries the meanings walking, passage, birth canal, and crossroads. The pairing of 行 with 首 can stand for a person. Source.

The oldest 道 thus holds three things in one sign.

• 行, the crossing: the self-crossing of N1.
• 首, the place in the crossing: the point, the location, light.
• 止, the foot: the step, the movement bound to the place.

There is no carrier besides these. The crossing, the place, and the step are given together. Source for the graph; reading for the mapping.

8. The Oldest Account of Origin: Arising by Return

In 1993 a tomb at Guodian in Hubei yielded bamboo slips from around 300 BCE. They hold the oldest known Laozi material. Beside it lies a short text of 14 slips, the Taiyi sheng shui, “The Great One Gave Birth to Water.” It has not been found anywhere else. Source.

Its opening reads:

太一生水，水反辅太一，是以成天。天反辅太一，是以成地。天地復相辅也，是以成神明。神明復相辅也，是以成陰陽。陰陽復相辅也，是以成四時。四時復相辅也，是以成寒熱。寒熱復相輔也，是以成濕燥。濕燥復相輔也，成歲而止。

In translation:

The Great One gave birth to water. Water returned and assisted the Great One; thus heaven was completed. Heaven returned and assisted the Great One; thus earth was completed. Heaven and earth assisted each other again; thus the spirits above and below were completed. These assisted each other again; thus yin and yang were completed. Yin and yang assisted each other again; thus the four seasons were completed. The four seasons assisted each other again; thus cold and heat were completed. Cold and heat assisted each other again; thus moist and dry were completed. Moist and dry assisted each other again; the year was completed, and it stopped there.

The text then reads the chain backwards. The year is born of moist and dry, moist and dry of cold and heat, and so on back to the Great One. Source.

Then come three sentences that hold the whole mechanism:

是故太一藏於水，行於時，周而又，以已為萬物母；一缺一盈，以已為萬物經。

Therefore the Great One hides in water and moves through the seasons. It goes round and begins again, and makes itself the mother of the ten thousand things. Once lacking, once full, it makes itself the warp of the ten thousand things.

Source. The word 經 means the warp: the fixed threads of a loom through which everything else is woven.

Chapter 42 of the received Laozi, “One gives birth to two, two to three, three to the ten thousand things,” does not occur in the Guodian Laozi. The older cosmogony does not count. It lets things arise by return and by mutual support. Source.

9. Three Rules Instead of Five Axioms

The material of sections 2 to 8 yields three rules.

R1 — Crossing, place, and step (道). There is one movement that leaves a place through a crossing. The place is light. The movement is bound to it.

R1 replaces N1. It needs no strand as a separate carrier. The crossing, the place, and the step are given together, as in 道.

R2 — Return gives birth (反). What arises returns to its source, and from that return the next thing arises.

R2 replaces N2. In N2 closure only selects. In R2 closure also produces. Water returns to the One, and heaven is completed. In register terms this is “>” coming back to its point.

R3 — Mutual support (相). Two things that have arisen support each other, and from that support the next layer arises.

R3 is the meeting “><” with sum 0, but productive. Repeated, it gives the ladder of N5: each layer arises from the pair of the layer before.

What N3 and N4 did is already present.

Memory (N3). The text reads the chain backwards, each thing with its source. The order of arising is the form, as the stroke order is the character. Alternation. “Once lacking, once full” is +1/−1. “Goes round and begins again” is the cycle of R. The year is the first complete closure. Context (N4). The rest of the text says that the way of heaven values the weak, cuts what is complete to increase what is growing. The response depends on the state, not on a fixed law. Correspondence.

The image is the same as before. The One is the warp of all things and hides in water. That is the net spanned in water, in the oldest source. Correspondence.

Why this is older and simpler: R1–R3 need no strand, no external rewrite, and no selector. Return presupposes a place to return to. That place is light. Arising presupposes a movement bound to that place. That movement is the step of it. Reading.

10. What Changes in the Theory

Several statements of the Foundational Paper shift.

N1 speaks of an oriented, continuous strand. It becomes a sequence of locations with a fixed order of arising. N2 says what does not close does not last. Free light lasts billions of years. Under R2, closure means return in phase, and a cycle of R also returns. Light lasts because it is periodic. Matter lasts because it holds. N4 makes c a state function. If light is the unit, c is 1 in native units. That c varies means the step from address to address per rewrite varies locally. C6 reads 0 as rest. That reading lapses. Rest is now the property of light, not a trit value. The 0 is the meeting of an act with its mirror. R is no longer applied from outside. It is the form reading itself. Its fixed points (T14) are locations that stay equal under self-reference. Kauffman’s eigenform O(A) = A states the same.

Reading for all five.

11. Open Questions

Order. The register records sums. “><” and “<>” have the same sum but differ. A quantity that records order is needed.

Dimension. A random movement along one line is bound with certainty. In three dimensions the return chance is 0.3405. If space consists of one-dimensional bound movements, where do the three dimensions come from? This is probably where binding and closure part ways.

The nine products. From water to the year, the Taiyi sheng shui names nine products: water, heaven, earth, the spirits, yin-yang, the four seasons, cold-heat, moist-dry, and the year. The register has nine closing windows. Whether these belong together must be computed first. It must not be read in. This is open.

12. Summary

The only thing there is, is light. The units of the net are already one photon: 45.6 nm, 1.52 × 10⁻¹⁶ s, 27.2 eV.

Light is a location. Matter is a location that holds against the progression. c measures the difference.

A location is a point with a bound movement. Along one line, a random movement returns to its point with certainty.

The act combined with itself gives all three trit values. Meeting and going round are the two kinds of closure.

The oldest 道 shows a person at a crossing with a foot. The oldest Daoist cosmogony lets everything arise by return and by mutual support, and calls the One the warp hidden in water.

Three rules follow. One movement leaves a place through a crossing. Return gives birth. Mutual support builds the ladder. The net is not threads and meshes. It is locations referring to locations.

---

Annotated References

Konstapel, J. (2026). The Vacuum.Net Theory: Foundational Paper. Edition 8, version 8.1. Constable Research, Leiden.
Why read? The five axioms, the window register (T9–T14), the unit derivation (VII.2), and the free-path lever (T18) that this essay rewrites. Reading advice: Part 0, then Parts III.4 and VII.2.

Konstapel, J. (2026). “The Possibilities and Impossibilities of the Net.” constable.blog, 19 October 2026.
Why read? The fifty-cell map of what the net can do, with the power ceiling and the depth-ladder landings used in section 2.

Larson, D. B. (1959). The Structure of the Physical Universe. North Pacific Publishers, Portland.
Why read? Space and time as two aspects of one motion. The photon at rest in the natural reference system; matter as rotation against the progression. Reading advice: the opening chapters on the natural reference system.

Spencer-Brown, G. (1969). Laws of Form. Allen & Unwin, London.
Why read? The mark of distinction and re-entry. A location as a distinction needs no carrier. Reading advice: chapters 1–2 and 11.

Kauffman, L. H. (2005). “EigenForm.” Kybernetes 34(1/2), 129–150.
Why read? The eigenform O(A) = A as a form that reproduces itself. The basis for reading R as self-reading rather than an external rewrite.

Varela, F. J. (1975). “A calculus for self-reference.” International Journal of General Systems 2, 5–24.
Why read? Adds a third, autonomous value to Spencer-Brown’s two. A ternary ground from distinction alone. Varela was also a practitioner of meditation, not only a formalist.

Pólya, G. (1921). “Über eine Aufgabe der Wahrscheinlichkeitsrechnung betreffend die Irrfahrt im Straßennetz.” Mathematische Annalen 84, 149–160.
Why read? The proof that a random walk returns with certainty in one and two dimensions, and not in three. The ground of section 5.

Watson, G. N. (1939). “Three triple integrals.” The Quarterly Journal of Mathematics 10, 266–276.
Why read? The computation of 0.3405 for the three-dimensional walk.

OEIS A002426. Central trinomial coefficients.
Why read? The sequence 1, 1, 3, 7, 19, 51, 141, … that counts exact returns of the trit walk, with its asymptotics.

Knuth, D. E. (1997). The Art of Computer Programming, Vol. 2, 3rd ed. Addison-Wesley.
Why read? Balanced ternary and its symmetries, including truncation as rounding. Reading advice: pp. 195–213.

Jingmen Municipal Museum (1998). Guodian Chu mu zhujian 郭店楚墓竹简. Wenwu, Beijing.
Why read? The photographic edition and transcription of the Guodian slips, including the Laozi material and the Taiyi sheng shui.

Allan, S. & Williams, C. (eds.) (2000). The Guodian Laozi: Proceedings of the International Conference, Dartmouth College, May 1998. Society for the Study of Early China, Berkeley.
Why read? Scholarly discussion of the Guodian Laozi and the first wide treatment of the Taiyi sheng shui in English.

Henricks, R. G. (2000). Lao Tzu’s Tao Te Ching: A Translation of the Startling New Documents Found at Guodian. Columbia University Press, New York.
Why read? Chapter-by-chapter translation of the Guodian Laozi, with the ancient and modern graphs. Shows which received chapters are absent, including chapter 42.

Wenlin Institute. Dictionary entry 道 (dao).
Why read? The older graphs 行/十字, read as a person at a crossroads 行 with the foot 止 below. The paleographic basis of section 7.

Williamson, J. G. & van der Mark, M. B. (1997). “Is the electron a photon with toroidal topology?” Annales de la Fondation Louis de Broglie 22(2), 133–160.
Why read? The electron as light closed on itself in a double loop. Matter as held light, in a physical model.

Laozi, Daodejing, chapters 11, 25 and 42.
Why read? The useful empty centre of the hub (11), the way that goes round (25), and the counting cosmogony (42), to set beside the older Taiyi sheng shui.