Company / Research
The invariant across this work is not a number — it is a rule for choosing one. 1 × 3 × 7 for simple things. 3 × 3 × 7 for most things. Fractal recursion when the thing is genuinely complex. Global crisis management was the scope chosen to work that rule out in, because it is the hardest case available and it had to be picked somewhere.
Where it starts
The premise was that the compounding crises of the next few decades — climate, overpopulation, food and water, autonomous labour, pandemics, migration, energy — are not separate problems, and that the people who will inherit them are not being equipped to reason across them. No degree programme was doing that.
So the thesis had to build its own instrument: a way to score how any one of those domains affects any other, at any scale, without the analysis collapsing into either a spreadsheet or a slogan.
Crisis management is not where the sizing rule came from — it is where it had to survive its hardest case. A rule you only ever apply to easy problems is not a rule. Twenty-one interacting crisis domains, at every scale from a neighbourhood to a hemisphere, is the test that made it worth keeping.
The earliest thesis files carry internal document timestamps of 8–9 September 2024, with a changelog trail through 14, 20, 21, 26 and 28 September. A letter to the university's president asking whether such a discipline existed is dated the same 48 hours.
Stated precisely: that is document metadata, which establishes a date well and cannot establish authorship at all. It is corroborating evidence across dozens of files, not a cryptographic timestamp, and we are not going to call it proof.
The rule, at the middle setting
Crisis management is a "most things" problem, so it takes 3 × 3 × 7. Three lenses, each subdivided into the same seven domains, over three tiers of scale — so any topic has a coordinate before anything is scored. Written down in September 2024.
What is empirically the case, and what technology can actually do about it — climate science, demographics, agriculture and water, AI and biotech, epidemiology, migration science, energy technology.
How people actually respond — behavioural adaptation, mental-health impact, response to technological disruption, urbanisation stress, crisis psychology, displacement stress, social adaptation to energy transition.
What we owe each other about it — environmental ethics, ethics of population and allocation, food and water rights, the moral problems of AI and biotech, global health ethics, migration ethics, responsibility in energy transition.
Seven is where the domains stopped being separable. Past that they stop being distinguishable categories and start being restatements of each other — the same ceiling the Arc Lab now measures empirically when it reports a basis as entangled. The number was chosen in a thesis about floods and pandemics; it survived into a system that scores rhetoric, and it is the same argument in both places.
The second constant
A crisis does not have one scale. The same analysis has to work for a neighbourhood and for a hemisphere, and the thesis solved that by making every tier the same shape as the tier above it:
| Weighted Node | A collection of relevant factors, given weights relevant to its parent |
|---|---|
| Parent Cluster | A collection of Weighted Nodes, with its own modifiers, which then becomes a factor |
| Super Node | The total of the Weighted Nodes and Parent Clusters in its domain — and can itself have factors and modifiers |
Each tier is a complete instance of the tier beneath it, and the whole structure scales up or down — city to county, national to international — without changing form. That is the descent rule of the address space, written for disaster scoring two years before it addressed memory.
The instrument
The thesis needed numbers that stayed comparable across scales and crisis types. Two formulas do that work, and both are in the September 2024 material.
It is a weighted projection with a scope multiplier: two impact terms, a signed compounding term, and a factor for how many people it reaches. The shape recurs — a magnitude, an alignment that can be negative, and a weight — in the arc scoring the cognition engine runs today. Same instrument, different subject.
The groundwork
Written before the scoring system, and explicitly grounded in existing literature — systems thinking, and Jay Forrester's system dynamics from the 1950s. The thesis does not claim to have invented the field it is standing in.
| State variables | The measurable properties of a system at a moment — sea level, infrastructure stability, population density |
|---|---|
| Interaction rules | The cause-and-effect dynamics between them — rising sea levels increase coastal erosion |
| Rate of change | Speed and direction over time, from internal dynamics and external forces |
| Feedback loops | Cycles where a change accelerates or damps another, producing compounding effects |
| Degree of impact | The quantified consequence for human or ecological interests |
| Proportional weighting | How much each variable matters to the whole, scaled by economy or population |
The applied instance
The thesis method pointed at one domain and taken all the way down: a 228-page manual — roughly 69,500 words — on smart-city urban resilience and emergency deployment. Modular adoption, multi-agency crisis coordination, dynamic resource allocation, and ethical oversight written in rather than bolted on.
It was also answered against the Heilmeier catechism, the question set DARPA uses to interrogate a research proposal: what are you trying to do, how is it done today, what are the limits, who cares, what are the risks. That document exists because the questions are the right ones to be asked, not because anybody asked them.
There is now also a simulator — terrain, road network, EMS allocation, utilities, and a doctrine clause system — which runs and typechecks clean. It is days old and has no test suite, and there is no deployed municipal system. See SeCURED.
| Manual | 228 pages · ~69,500 words · unpublished |
|---|---|
| Method | The three-pillar, seven-category frame applied to one city-scale domain |
| Interrogated by | The Heilmeier catechism, answered in full |
| Simulator | 16,022 lines · builds clean · zero tests |
| Deployment | None. No municipality runs this |
Where it became code
Each row is a constant argued for in the 2024 thesis material and a place it now appears as a literal, asserted by a test or countable on disk. This is the only claim this page is really making.
| 3 × 3 × 7 = 63 | Three tiers (node / cluster / super) × three pillars × seven categories →
domainLibrary.ts slices a domain into 63 layers
("3 states × 3 axes × 7 layers"), and a suite asserts TOTAL_TOOLS === 63 |
|---|---|
| Seven as a ceiling | Seven crisis domains per pillar → CANONICAL_ARCS is exactly seven, and
the key system addresses galaxy 1–3 · system 1–3 · object 1–7 — 63 of 63 object
keys generated and present on disk |
| Same shape at every tier | The fractal template's scale-up/scale-down rule → a cell that contains a complete
instance of the structure; CUBE_DATA_CELLS = 342 with (6,6,6) reserved as the
portal that opens a full child level |
| Weighted projection with sign | SIS = [(DIS + IIS)·W1 + CMS·W2] × RAP, where a topic can compound or
mitigate → arc scoring with magnitude, signed alignment and influence weight |
It is not independent replication. The same person wrote the thesis and the code, so convergence between them is not evidence in the way a second team reproducing a result would be. Anyone treating this table as validation has misread it.
What it is: the constants were committed to in writing first, and the implementation can therefore be checked against them rather than rationalised backwards into them. None of this material is peer reviewed. It is offered because it explains why the system has the shape it has, not as a credential.