Wiki · The Colony & the Method
SP.Lab — User Acceptance Test (UAT) checklist
How to read this page
Three ways to read this page. Precise is the document itself, exactly as it is written in the repository. Plain and Clear were written for this website to help you meet that document — they are about it. They are not it, and they are not evidence.
Eighty-four pages about the colony. Each agent is an Elixir process holding a generative model and doing inference, attached to a body that logs into a Minecraft world as an ordinary player. Around that sit the broadcast suite that films them and the runbooks that keep the whole thing running. There are typed specifications for each organ of the model, plus the world and genome specs. There are also the adversarial review personas used to attack a proposed change before it ships.
It is for the reader curious how a running system is put together and how it is held to account. The accountability half is the more distinctive. There is a lab protocol governing evidence and attribution, and a claim fence that restricts the vocabulary a claim is allowed to use. There is a public gate log. And there is a standing invitation to reproduce any verdict from the commit and the seed named in its receipt.
Start with the public read, then the lab protocol, then the falsification invitation. If you want the mathematics rather than the operations, go straight to the typed organ specs.
What it is not: a description of a mind, and not all one kind of document. A large part of this corpus is design and planning — specs marked as proposed rather than applied, organs designed but not built, plans that were later superseded — and each page states which it is. A specification is not a running system, and these pages are careful about the difference; the reader should be too. Eight documents were withheld from publication because they describe private infrastructure.
Your browser cannot switch reading levels, so the document itself is shown.
Precise — the source document
This is the document. Rendered from the repository at the commit above, with nothing rewritten for the web. A gate re-renders it on every deploy and fails the build if a single byte differs.
Maps each of the 20 "hard tests" from the original lab build brief to its covering evidence.
Status is one of PASS (automated assertion or documented + cross-checked), HOST (out of
the lab's scope; covered by the host StratifiedPalimpsest system), or NOT-BUILT (honestly
out of scope for this lab; flagged in lab/docs/SCIENTIFIC_LIMITS.md).
Reproduce all PASS rows: mix test test/sp/lab/ and mix sp.lab.validate.
| # | Hard test | Status | Evidence |
|---|---|---|---|
| 1 | Unit / dimensional consistency | PASS | lab/proofs/dimensional_analysis.md; doctests; mix sp.lab.validate |
| 2 | Earth gravity from g = GM/R² |
PASS | physics_test "reproduces every body within 2%"; doctest surface_gravity; validate "Newtonian envelope ≤0.36%" |
| 3 | Moon gravity persists at ~vacuum pressure | PASS | physics_test "Moon: substantial gravity at ~vacuum" (g=1.62 at 3e-15 bar) |
| 4 | Mercury gravity persists at ~vacuum pressure | PASS | physics_test "Mercury: gravity ≈ Mars at ~vacuum" (g=3.70 at 5e-15 bar) |
| 5 | Titan pressure–gravity contradiction | PASS | physics_test "Titan overshoots"; validate "Pressure overshoots Titan ~10x" (10.49×) |
| 6 | Venus pressure–gravity contradiction | PASS | physics_test "Venus overshoots"; validate "Pressure overshoots Venus ~100x" (100.4×) |
| 7 | Ozone optical-depth reduces UV (correct direction) | PASS | radiation_test "more ozone column → lower surface UV (monotone)" |
| 8 | Zero ozone increases UV hazard | PASS | radiation_test "zero absorber → full transmission"; validate "Zero ozone → transmittance 1.0" |
| 9 | Proton gradient increases ATP proxy | PASS | bioenergetics_test "steeper proton gradient yields more ATP (monotone)" |
| 10 | Loss of membrane integrity collapses viability | PASS | bioenergetics_test "membrane breach collapses viability" |
| 11 | Strictly-aerobic cell fails without O₂ | PASS | bioenergetics_test "fails without O2"; validate "Aerobic cell w/o O2 is nonviable" |
| 12 | Anaerobic cell survives without O₂ (valid acceptor) | PASS | bioenergetics_test "anaerobic survives with alt acceptor"; validate "Anaerobic+sulfate viable" |
| 13 | Active inference updates priors under evidence | HOST | Out of lab scope. Host system: docs/EVIDENCE.md §2.5, mix sp.brain.verify. Lab documents the bounds only (lab/proofs/active_inference_bounds.md). |
| 14 | Mis-set precision delays/blocks belief update | HOST | Out of lab scope. Host system: docs/EVIDENCE.md §2.6 (dynamic precision). |
| 15 | Model comparison penalizes circular formula | PASS | model_compare_test "rubric prefers Newtonian; pressure penalised"; validate "Model verdict == :newtonian_dominates" |
| 16 | D-Value flags hidden scaling / unit ambiguity | PASS | lab/proofs/dgst_d_value_audit.md (D-value reduces to angular diameter; gravity is an implicit input; undeclared ×10⁵). Cross-checked: pressure model fails 5/5. |
| 17 | Solar vacuum reduces convective loss, not radiative | PASS | solar_energy_test "vacuum removes convective loss but radiative persists"; validate "Radiative floor ~893 W" |
| 18 | Long-horizon simulations mark uncertainty explosion | NOT-BUILT | The lab has no time-evolution simulator (the planet/solar-system time-scale module was not built). Flagged in lab/docs/SCIENTIFIC_LIMITS.md. |
| 19 | UI displays evidence class for every claim | PASS | lab/ui/index.html — every row carries an A/B/C/D/U/X pill badge. Screenshot in lab/evidence/captures/. |
| 20 | No module reports unsupported claims as "proven" | PASS | model_compare_test "vocabulary closed: 'proven' not a result/class"; validate "Vocabulary closed" |
Summary
- 16 / 20 PASS (automated or documented + cross-checked).
- 2 HOST (#13, #14 — active-inference belief updating lives in the host system, not the lab; the lab fences itself to the AIF bounds).
- 1 NOT-BUILT (#18 — no time-evolution simulator; flagged, not hidden).
- 1 visual PASS (#19 — UI evidence-class badges, captured as a screenshot).
No row is failed. The honest gaps (#13, #14 host; #18 not-built) are recorded, not papered over.
sha256 89d45b6923bdd3fc — of the original file, so what was ingested stays checkable.
Plain — written for this website, not the source document
A mapping table, and little else: it takes twenty hard tests named in an original brief and, for each, says where the evidence for it lives. One passing row exists to show that a model comparison penalises a circular formula.
What makes the page worth reading is the status vocabulary. A row can pass, or it can be marked as belonging to a different system entirely, or it can be marked as not built. That third mark is the honest one, and one row carries it: there is no time-evolution simulator, so the test about long-horizon uncertainty could not be run, and the page says so rather than quietly dropping the row.
Two further rows are marked as living in the host system rather than this smaller lab, which fences the lab to a narrower claim.
The summary counts each category and ends on the point of the exercise: no row is failed, and the honest gaps are recorded rather than papered over.
Plain · written 2026-08-01 by claude-opus-5 · not yet checked by a person · about the document whose sha256 is 89d45b6923bdd3fc
Clear — written for this website, not the source document
A traceability checklist, row by row. It takes twenty hard tests from an original build brief and maps each one to the evidence that covers it, with a command given at the top for reproducing every passing row.
The status vocabulary is declared before the table and is the reason the page is useful. A row can pass, meaning an automated assertion or a documented and cross-checked result. It can be marked as belonging to the host system, meaning it is out of this smaller lab's scope. Or it can be marked as not built, meaning it is honestly out of scope and flagged in a separate limits document.
The table itself covers a wide range. There is consistency of units, and reproducing surface gravity for several bodies from a standard formula. There is a check that gravity persists at near-vacuum pressure on two airless bodies, and two cases where a rival pressure-based account overshoots badly. There is the direction of an effect linking an absorber to surface radiation, and the case of no absorber at all. And there are two bioenergetic monotonicity checks, the collapse of viability when a membrane fails, and the difference between a cell that requires oxygen and one that does not. Further rows cover a model comparison penalising a circular formula, an audit that flags a hidden scaling factor, and a thermal case separating one loss mechanism from another. Then a user interface requirement that every claim carry an evidence class, and a vocabulary check that no module reports an unsupported claim using a forbidden word.
Two rows are marked as belonging to the host system, both concerning belief updating under evidence and the effect of mis-set precision. The page states that the lab documents only the bounds of that topic, which is a deliberate narrowing of what this lab claims.
One row is marked as not built: there is no time-evolution simulator, so the test about long-horizon uncertainty could not be run at all. It is flagged rather than removed.
The summary counts each category and closes on the point of the whole exercise. No row is failed, and the gaps that exist are recorded rather than papered over. Read that way, the checklist is less a score than an index into where the evidence is and where it is missing.
Clear · written 2026-08-01 by claude-opus-5 · not yet checked by a person · about the document whose sha256 is 89d45b6923bdd3fc