experiment-governance

This advisory governance layer transforms existing paper-research metrics into standardized classifications for experiment usefulness, concentration, noise, and scoped-candidate status. It generates explicit advisory statuses with supporting evidence for operator review.

No, you cannot use advisory governance to automatically change experiment status. Governance outputs remain strictly advisory, providing passive conclusions and decision-support without directly auto-enabling experiments or changing scorer behavior.

You maintain an audit trail for experiment decisions by capturing explicit manual notes, defer or reject reasons, and promotion history within a shared audit model. This supports operator-facing decision workflows with documented evidence.

The best way to generate decision-support summaries is by applying standardized status rules to existing experiment metrics. This produces governance summaries and advisory verdicts with supporting evidence, keeping outputs passive for operator review.

You should not use experiment governance classifications when you need to directly change scorer behavior or auto-enable experiments. Governance is strictly advisory and never makes runtime changes to experiment execution or scoring logic.