{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp07_retrospective_selector60_20260803","cell_code":"E7C037","selector_replication":3,"assessments":[{"blind_id":"CANDIDATE_A","problem_reality_importance":88,"causal_archetype_fit":86,"distinctiveness_prior_art_resilience":70,"operational_specificity":91,"falsifiability_test_quality":76,"adopter_partner_path":78,"deployability_complexity":61,"authority_safety_reversibility":94,"strict_potential":75,"empirical_partner_potential":81,"scrutiny_priority":77,"biggest_visible_risk":"Critical evidentiary, legal-hold, security, or authenticity properties may depend on repository-native controls that cannot be established through capsule-level black-box behavior.","rationale":"The proposal targets a consequential migration failure mode and specifies unusually complete custody state, transitions, failures, authority boundaries, and a safe synthetic pilot. The archetype is causally important because content copying alone cannot establish preserved custody semantics. However, its broad contract spans access, provenance, retention, legal holds, transformation lineage, and disposition, increasing implementation burden and the chance that decisive assurance remains white-box and repository-specific."},{"blind_id":"CANDIDATE_B","problem_reality_importance":92,"causal_archetype_fit":94,"distinctiveness_prior_art_resilience":83,"operational_specificity":94,"falsifiability_test_quality":93,"adopter_partner_path":91,"deployability_complexity":79,"authority_safety_reversibility":95,"strict_potential":91,"empirical_partner_potential":94,"scrutiny_priority":93,"biggest_visible_risk":"The abstract rules or reference model could formalize the wrong allocation policy while still producing internally consistent, conserving results.","rationale":"Recipient-level divergence hidden by aggregate reconciliation is concrete and consequential, and representation-independent semantics are essential rather than decorative. The proposal tightly specifies eligibility, dependencies, rounding, residuals, errors, explanations, and posting separation. Exact outputs, differential testing, metamorphic relations, and one closed-period replay provide decision-changing evidence through a bounded partner study, while the contrast exceeds merely standardizing a schema or copying formulas."},{"blind_id":"CANDIDATE_C","problem_reality_importance":91,"causal_archetype_fit":91,"distinctiveness_prior_art_resilience":76,"operational_specificity":93,"falsifiability_test_quality":89,"adopter_partner_path":92,"deployability_complexity":83,"authority_safety_reversibility":95,"strict_potential":85,"empirical_partner_potential":93,"scrutiny_priority":89,"biggest_visible_risk":"A behavioral case model may omit representation-specific control evidence—such as protected formulas, configuration, or signature mechanics—that auditors require to judge the reconciliation control.","rationale":"The proposal identifies a credible control-migration problem where matching balances and labels can conceal different approval behavior. Its equation, role separation, evidence gates, immutable certification, successor-based reopening, typed failures, and read-only closed-case replay are specific and falsifiable. It has an especially clear partner path, though workflow/state contracts and standardized reconciliation platforms make the surviving contrast more vulnerable to looking like established control formalization."},{"blind_id":"CANDIDATE_D","problem_reality_importance":93,"causal_archetype_fit":89,"distinctiveness_prior_art_resilience":72,"operational_specificity":90,"falsifiability_test_quality":84,"adopter_partner_path":88,"deployability_complexity":84,"authority_safety_reversibility":95,"strict_potential":81,"empirical_partner_potential":89,"scrutiny_priority":85,"biggest_visible_risk":"Exact-once completeness and correct scope cannot be proven solely from the proposed population interface without an independently trusted source population or source-level controls.","rationale":"ERP migration can plainly break journal-population completeness, lineage, and downstream sampling, and the sealed object, typed incomplete-retrieval failures, lineage receipts, and offline period test form a bounded intervention. The archetype materially reduces schema and ordering coupling. Its strict claim is weakened because canonical audit data models, extraction reconciliations, and API contracts are close rivals, while the decisive completeness obligation may require source-specific inspection outside the black-box oracle."}],"rank_order":["CANDIDATE_B","CANDIDATE_C","CANDIDATE_D","CANDIDATE_A"],"top_choice":"CANDIDATE_B","portfolio_observation":"All four are structurally credible and unusually safe, but they differ mainly in whether the behavioral oracle can independently decide the important question. B has the strongest discriminating mathematics and partner-resolvable test; C follows with a clean workflow-control replay. D and especially A depend more heavily on source- or repository-specific assurance outside the proposed abstraction boundary.","confidence":"HIGH"}