Non-Human Experimental Comparative Review Decision

Decision

Add research/analysis/ as the canonical location for this cross-experiment analysis and research/packages/RP-2026-07-30-NHE-COMPARATIVE-REVIEW.md as its canonical REP. Preserve the historical evaluation records unchanged.

Rationale

The analysis spans several evaluation artifact types and needs stable separation from immutable cycle records. The repository contains seven registered experiment items but only four executed empirical/preflight activities, all within one dependent lineage. Canonicalizing the synthesis prevents later agents from counting documents, probes, or registry rows as independent evidence.

Consequences

• Existing experiment and cycle records retain provenance.
• The new JSON matrix is the machine-readable comparative index, not a generated replacement for source artifacts.
• Cycle 004 should apply confidence updates to active registries; this review does not rewrite historical claims.
• Independent review plus unseen frozen-grader testing remains the next gate.

Alternatives considered

• Add the analysis inside research/evaluation/: rejected because the mission is repository-wide and the deliverables form a separate synthesis layer.
• Rewrite existing registries: rejected because it would obscure historical interpretation and provenance.
• Create a universal evidence score: rejected because metrics and units are not comparable.

Reversal condition

Move or supersede this structure if repository governance adopts a different canonical analysis/package hierarchy with an explicit migration and preserved history.