05 / DATA
Show the receipts.
A public result should connect to its protocol, configuration, prompts, raw logs, analysis, and corrections.
From claim to raw capture
Every public experiment should carry a stable ID and connect the plain-language result to enough material for another person to inspect or repeat the work.
Question, hypotheses, variables, controls, predictions, stopping rules.
Machine, software, model revision, settings, and measurement tools.
Full prompts, outputs, timings, warnings, interventions, and immutable logs.
Scripts, derived data, interpretation, limitations, and dated corrections.
All 90 experiment pages now show their question, method, controls, measures, confounds, and what would weaken the hypothesis. Those pages derive from the September 26 registry and protocol compendium, version 1.0. Local apparatus evidence has been preserved; no raw machine records are included in this update.