Wiki · The Colony & the Method
Morphology Ablation Report
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and planning — specs marked as proposed rather than applied, organs designed but not built, plans that were later superseded — and each page states which it is.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.
Purpose
Verify Validation Invariants #6, #9 and Acceptance Gate #5 (morphology matters): actions are gated by appendages, development cannot create impossible bodies, and a body that never develops cannot perform the consequential late game.
Method
- Gating (deterministic): assert a seed body cannot perform appendage-gated actions; growing an organ without prerequisites is rejected.
- Development validity (property): hundreds of random genomes developed for many ticks always yield valid body graphs.
- Utilisation (statistical): run the Infrastructure baseline with development
disabled (
dev_intervalbeyond horizon) vs enabled, and measure structures built and expansions.
Artifacts used
SP.BodyTest,SP.Body.DevelopmentTest(property),SP.EvalTest,SP.InvariantsTest(#6, #9).scripts/evidence.exsmorphology section.
Result summary
Gating (exact): a seed body returns false for excavate, build_*,
open_seam; Body.grow(seed, :excavator, …) ⇒ {:error, {:prereqs_unmet, :excavator, [:manipulator]}}. Locomotion (move/orient/probe) needs no appendage.
Development validity: SP.Body.DevelopmentTest develops 150 random genomes
for 80 ticks each; all resulting bodies pass Body.valid?/1 (Invariant #9).
Utilisation (from scripts/evidence.exs):
never-develop body: stage=0 structures=0 expansions=0 (cannot build/excavate/expand)
developing Infrastructure (seeds 11..40): builders=9/30 total_structures=36 max_stage=4
A body that never develops is permanently confined to locomotion and dies without constructing anything. A developing body reaches stage 4, builds infrastructure in 12/30 runs, and can open seams (see open-endedness report).
Pass/Fail
PASS. Morphology is necessary: it gates the entire action ladder (exact), can never be malformed (property), and is the prerequisite for all niche construction and expansion (statistical).
Residual risks
- Whether a developing body reaches a given stage within an episode depends on surviving long enough (energy economy); harsh seeds cap development. This is the intended bootstrapping difficulty, not a defect.
sha256 b3b3e3051f1e78f6 — of the original file, so what was ingested stays checkable.
Plain — written for this website, not the source document
This report checks that the body an agent grows actually matters. Three claims are tested: actions are gated by having the right appendage, development cannot produce an impossible body, and a body that never develops cannot do the consequential later work.
The method mixes three kinds of evidence. An exact check that a starting body is refused the gated actions and that growing an organ without its prerequisite is rejected by name. A property test that develops many random genomes over many ticks and requires every resulting body to be valid. And a statistical comparison between runs with development switched off and runs with it on.
The result is that a body that never develops is confined to moving about and dies without building anything, while developing bodies reach later stages and build.
The verdict is a pass, and the report notes that harsh seeds cap development, which it calls intended difficulty rather than a defect.
Plain · written 2026-08-01 by claude-opus-5 · not yet checked by a person · about the document whose sha256 is b3b3e3051f1e78f6
Clear — written for this website, not the source document
This is one of a set of short validation reports, and it tests whether the body an agent grows is load-bearing rather than decorative. It names the invariants and the acceptance gate it is verifying.
Three claims are separated, each tested with a different kind of evidence. The first is gating, tested exactly: a starting body is asked to perform the appendage-gated actions and refuses each one, and an attempt to grow an organ whose prerequisite is missing returns a named error rather than succeeding quietly. Moving, orienting and probing need no appendage, which is stated so the boundary is clear.
The second is development validity, tested as a property: many random genomes are developed over many ticks, and every resulting body must pass the validity check. The report says all of them did.
The third is utilisation, tested statistically: the same strategy runs with development effectively disabled, then with it enabled. A body that never develops stays at the starting stage, builds nothing, and cannot excavate or expand. A developing body reaches a later stage, builds in a minority of runs, and becomes able to open seams, with a separate report cited for that.
The verdict is a pass. The residual risk is stated honestly: reaching a given stage inside one episode depends on surviving long enough, and harsh seeds cap development. The report calls that intended difficulty rather than a defect, which is a claim about design intent rather than a measurement.
Clear · written 2026-08-01 by claude-opus-5 · not yet checked by a person · about the document whose sha256 is b3b3e3051f1e78f6