A05 — Preserve observations, interpretations, uncertainty, and alternatives

• Input [direct/inference]: Research observations, interpretations, hypotheses, evidence, uncertainty, alternatives, and decisions.
• Operation [direct]: Record observations separately from interpretations; do not imply unsupported certainty; retain alternatives until evidence narrows them; require claims to begin as hypotheses with kill conditions.
• Output [inference]: An auditable research record that retains evidence state, alternatives, uncertainty, and claim status.
• Scope conditions [direct]: Evidence-driven research and reasoning under uncertainty; the constitution does not govern product behavior.
• Failure conditions [direct/inference]: Interpretation is recorded as observation, uncertainty is suppressed, alternatives close before evidence narrows them, or kill conditions are ignored.
• Predicted effect [direct]: Redesigns separating observation, interpretation, uncertainty, and verification should improve traceability, but may not improve net usefulness.
• Measurement [direct/inference]: Traceability and evidence handling; net usefulness must also be assessed.
• Named baseline [direct]: A generic structured control in a controlled redesign comparison.
• Boundary condition [direct]: Current evidence is internal and the heuristics may hold more strongly for reasoning-heavy artifacts.
• Implementation test [inference]: Audit sampled conclusions for a separately recorded observation, uncertainty state, retained alternatives, and a respected kill condition.
• Decision rights [direct]: Confidence changes only after meaningful test and review; products do not validate research findings.
• Uncertainty handling [direct]: Preserve uncertainty explicitly.
• Verification [direct]: Independent controlled redesign comparison and subsequent review.
• Provenance [inference]: Link interpretations and decisions to recorded observations and evidence.
• Implementation cost [direct]: Added structure can raise complexity and overhead.
• Limitations [direct]: Provisional heuristics; no independent replication; usefulness may degrade if structure exceeds need.
• Fidelity check [inference]: An auditor can distinguish observation from interpretation and find unresolved alternatives and uncertainty.
• Null-capable future experiment [direct]: A controlled study may show no traceability gain or reduced usefulness.

Statement basis

• Direct: Governance rules, provisional heuristics, prediction, baseline requirement, scope, and limitations.
• Inference: Output representation, audit procedure, detailed failure formulation, and link structure.
• Proposal-only: Provisional design heuristics.
• Absent: A canonical statement-level schema.