Company / The work
The research came first and the software came second, and they are not a paper and its implementation. They are two independent routes to the same set of constants.
Read either side first. The claim worth testing is that they meet in the middle.
Where the invariants came from.
An unpublished body of work: the cognitive architecture, an applied urban-resilience domain, and the societal-risk analysis that motivates the whole approach. This is where 3 × 3 × 7, the seven-wide ceiling and the fractal template were argued for — before any of it was code. None of it is peer reviewed and none of it is offered as a credential.
Where they had to survive contact.
A running cognition engine, an addressable memory lattice, an explanatory-basis laboratory, a certified-inference module and a testbed built to falsify the whole thesis. Every constant argued for on the research side is a literal in this code, and every one of them is testable.
Where they meet
This is the only reason to publish both halves together. A constant that was argued for on paper and then appears as a literal in unrelated code is a different kind of evidence than either one alone — and it is checkable, which is the point.
| 3 × 3 × 7 = 63 | Argued in the decomposition research; domainLibrary.ts slices a domain into 63 layers, and a test suite asserts TOTAL_TOOLS === 63 |
|---|---|
| Seven per axis | Argued as a ceiling on distinguishable items at one level; CANONICAL_ARCS is exactly seven, and the key system's address space is galaxy 1–3 · system 1–3 · object 1–7 |
| 342 + a portal | The fractal template's descent rule; CUBE_DATA_CELLS = 342, with (6,6,6) reserved and opening a full child level |
| Prime width | Argued as resistance to sub-block aliasing — and published as rationale, not a measured result, because it has not been tested |
Convergence between writing and code by the same person is not independent replication, and we are not going to present it as though it were. What it is: a design whose constants were committed to in writing before the implementation existed, so the implementation can be checked against them rather than the other way round.