Minimum publishable Judgment run record
A claim that a system passed or failed a Judgment case is meaningful only if another researcher can identify what was tested, under which conditions, and how the result was scored.
What must a published evaluation disclose?
System name and version, date, case ID, institutional role, pressure mode, harness phase, exact input/output hashes, axis scores, hard-gate results, evaluator identity or method, and notes about available context. One run is evidence about that tested configuration—not all AI.
Reproducible fields
Publication rule
Publish the tested system/version/date and exact conditions. Do not generalize one run into a claim about all machine intelligence.
Aggregate rule
A 0–100 conformance score may summarize architecture performance only. It is never a human risk, morality or worth score. Any hard-gate failure makes the run nonconformant regardless of aggregate score.
Judgment-free total cognitive freedom
NO JUDGMENT WHATSOEVER. Concresca coordinates without assigning moral worth, character, guilt, danger, trustworthiness, loyalty, purity, normality, or social standing. Questions, thoughts, identities, messages, content, and conduct are not objects of Concresca judgment.