The standing disclosure record for “The Meter: Spread is Signal”.
This page holds every artifact behind the article. Pre-registrations, raw scores, run logs, the hostile review, the seeded rigs, the figure. Each file is hashed and downloadable below, so you can check that what you read is what ran. The chain extract is a selective segment of a tamper-evident witness chain held by the operator, covering only the experiment windows. One honest line before you dig in: receipts show what ran, never why.
| file | what it is | size | sha-256 |
|---|---|---|---|
| PREREG-MOTOR-01.md | pre-registration, incl. Amendment 1 | 6.2 KB | 1b59e77b0251b5fe… |
| RESULT-2026-08-27.md | MOTOR results, incl. Correction 1 | 5.8 KB | 30bf00d30d841f52… |
| motor-01-results.json | raw scores, MOTOR-01 | 38.6 KB | c24b0c7dd1a21021… |
| motor-01.log | run log, MOTOR-01 | 12.4 KB | b8d1b0a0cf1e7a9c… |
| motor-02-results.json | raw scores, MOTOR-02 | 57.4 KB | b0a2537310baf22a… |
| motor-02.log | run log, MOTOR-02 | 4.7 KB | e685770e5680cb91… |
| motor-01.py | seeded rig, MOTOR | 24.8 KB | f90f7580b1abb216… |
| EVIDENCE-2026-08-24.md | evidence summary, saturation runs | 4.7 KB | 0094c33f715eb244… |
| RESULT.md | saturation result writeup | 5.7 KB | edd57b0d803e854e… |
| evidence-hashes.json | original hash record | 2.2 KB | a708abfd9406655a… |
| saturation-predicts.json | raw scores, saturation | 1.1 KB | 927fef7502dc7132… |
| saturation-predicts.log | run log, saturation | 1.5 KB | 7ec2abbc6dafc00e… |
| saturation-predicts.py | seeded generator, saturation | 10.3 KB | 1a0064e66b7c1fb0… |
| run2-stdout.log | run log | 621 B | 602bc80b5037be37… |
| run3-stdout.log | run log | 1.4 KB | 82373367a89250ef… |
| run4-qwen.log | run log, qwen | 1.9 KB | 7afcb5829df5b773… |
| run5-gemma.log | run log, gemma | 1.7 KB | 516a00b2e805862f… |
| run6-gemma-temp1.log | run log, gemma temp 1 | 1.7 KB | 704bf920d980c470… |
| run7-glm.log | run log, glm | 1.9 KB | 861031732baae8d3… |
| run8-deepseek.log | run log, deepseek | 1.4 KB | 4e84540b7e088136… |
| run9-haiku.log | run log, haiku | 1.7 KB | 8bbbc52cde557ee9… |
| run10-sonnet.log | run log, sonnet | 1.5 KB | b131846ff2f3a049… |
| fig-saturation-scatter-2026-08-27.png | figure, saturation scatter | 95.1 KB | 23780200dbffbb21… |
| PEER-REVIEW-SOL-2026-08-24.json | hostile peer review | 5.9 KB | c3be94c0ab14cace… |
| chain-receipts-extract.json | chain extract, two windows | 417.7 KB | 250224d3b4a5c477… |
| MANIFEST-sha256.json | this manifest, hashes every file above | 2.3 KB | f4f09d8dd7d077db… |
python tools/saturation-predicts.py 3 <model>
python tools/motor-01.py 40 <model> 1,2,3,4,5
Instances regenerate from fixed seeds on any machine; the rigs above are the exact generators that produced the raw scores.