A05 — Preserve observation, interpretation, uncertainty, and alternatives

• Input [inference]: Research observations, interpretations, hypotheses, evidence, and candidate alternatives generated during evidence-driven work.
• Operation [direct]: "Observations and interpretations must be recorded separately" (research-constitution.md, Rule 3); "Separate Observation From Interpretation," "Preserve Uncertainty," and "Preserve Alternatives Until Evidence Narrows Them" are stated as design heuristics (emergent-design-laws-v0.1.md).
• Output [inference]: A record in which observation, interpretation, uncertainty, and unresolved alternatives remain individually identifiable rather than collapsed into a single conclusion.
• Scope conditions [direct]: Evidence-driven research reasoning; the research constitution's Interpretation Rule states "The constitution governs research behavior, not product behavior," so this scope is explicitly bounded away from product use.
• Failure conditions [inference]: Interpretation recorded as if it were observation, uncertainty not preserved, or alternatives closed before evidence narrows them — inferred as the direct complements of the stated rules and heuristics rather than stated as failure conditions in their own right.
• Predicted effect [direct]: "Framework redesigns that separate observation, interpretation, uncertainty, and verification will usually improve traceability, but not always net usefulness" (theory doc, Predictions).
• Measurement [inference]: Traceability is named as the affected quantity in the prediction above; net usefulness is named as a quantity that must also be checked but is not defined further.
• Implementation test [absent]: No audit or scoring procedure is defined in the permitted canonical set.
• Decision rights [direct]: "Confidence may not be updated directly from an evidence run. Confidence may increase only when evidence survives a meaningful test and subsequent review" (research-constitution.md, Rules 10–11); "Products consume research findings; products do not validate research findings" (Rule 4).
• Uncertainty handling [direct]: Preserving uncertainty is itself the mechanism, per the "Preserve Uncertainty" heuristic and the constitution's rule that every claim begins as a hypothesis with a kill condition (Rules 1–2).
• Verification [inference]: No dedicated verification procedure for this candidate is stated; the general rule that hypotheses require kill conditions and review before confidence changes (Rules 1–2, 10–11) is applied here by inference as the closest available verification path.
• Provenance [inference]: Not directly specified; inferred from the general requirement to record observations and interpretations "separately" (Rule 3), which implies each must be traceable to its own record.
• Implementation cost [direct]: Added structure of this kind "tends to add complexity" (Theory v0.1 claim 2, discussing structured redesign generally).
• Limitations [direct]: The emergent design laws are explicitly labeled "evidence-supported heuristics rather than accepted engineering laws," with "Promotion to law requires independent replication," and an Open Caution that they "may not be universal" and "may hold more strongly for reasoning-heavy frameworks than for execution-heavy or highly representational artifacts."

Statement basis

• Direct: Core operation, scope boundary, predicted effect, decision rights, uncertainty-handling rule, cost, and limitations.
• Inference: Input/output framing, failure-condition complements, measurement scope, verification path, and provenance implication.
• Proposal-only: The emergent design heuristics themselves are labeled provisional research synthesis rather than doctrine.
• Absent: A canonical statement-level implementation test or schema.