{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp03_full320_20260801","cell_id":"layer_decay_and_expiration_management__statistics_experimental_design","trajectory_id":"R","attempt_index":0,"archetype_slug":"layer_decay_and_expiration_management","domain_slug":"statistics_experimental_design","decision":"CANDIDATE","problem_id":"superseded_experimental_artifact_sprawl","causal_lever_id":"lifecycle_state_gating_for_analysis_artifacts","proposal":{"problem":"In long-running or repeatedly amended studies, superseded protocols, outcome definitions, randomization specifications, stopping rules, analysis plans, code, and derived datasets remain discoverable beside current versions. Analysts can unknowingly apply obsolete specifications, while indiscriminate cleanup can destroy prespecification evidence, regulatory records, lineage, or reproducibility.","actors_substrate":["study investigators and statisticians","data-management and coordinating-center staff","ethics, regulatory, and audit reviewers","versioned protocols, analysis plans, code, data dictionaries, derived datasets, and decision records"],"observable_state":"Multiple artifacts for the same study object lack reliable current, superseded, archived, held, and deletion-pending states; obsolete artifacts appear in ordinary search or workflows, and their dependencies and retention obligations are unclear.","consequence":"Wrong-version use can alter estimands, populations, outcomes, stopping decisions, or reported results; fear of losing provenance then causes every artifact to remain active indefinitely.","affected_objective":"Preserve valid inference, prespecification evidence, reproducibility, and oversight while reducing obsolete artifacts' operational authority and search burden.","structural_mapping":[{"archetype_element":"sequential deposits","domain_realization":"Successive protocol amendments, analysis-plan versions, code releases, data cuts, and derived datasets.","claim_kind":"INFERENCE"},{"archetype_element":"stale layers masquerading as current","domain_realization":"Superseded specifications or datasets remain selectable without an authoritative lifecycle label.","claim_kind":"HYPOTHESIS"},{"archetype_element":"retention and dependency tension","domain_realization":"Artifacts may be operationally obsolete yet necessary for audit, reconstruction, or interpretation of published results.","claim_kind":"INFERENCE"},{"archetype_element":"managed disposition","domain_realization":"Refresh, supersede, restrict, archive, compact only for navigation, quarantine, or delete under explicit authority.","claim_kind":"HYPOTHESIS"}],"component_map":[{"component":"layer_inventory_and_identity_map","status":"direct","domain_realization":"Registry of study artifacts keyed by study, estimand, version, data cut, and owner."},{"component":"deposition_order_and_age_index","status":"direct","domain_realization":"Creation, approval, effective, supersession, and last-use timestamps."},{"component":"retention_policy_matrix","status":"adapted","domain_realization":"Artifact-class rules for active availability, archival, and minimum/maximum retention."},{"component":"decay_function_or_aging_rule","status":"adapted","domain_realization":"Review priority rises with time since validation or supersession; age alone never invalidates evidence."},{"component":"expiration_trigger_definition","status":"direct","domain_realization":"Approval of a successor, study milestone, invalidated dependency, or elapsed review interval triggers reassessment."},{"component":"layer_value_and_risk_score","status":"adapted","domain_realization":"Separate operational-use, reproducibility, regulatory, privacy, and misuse-risk fields rather than one authoritative score."},{"component":"dependency_and_reconstruction_check","status":"direct","domain_realization":"Trace reports, code, data lineage, registrations, and decisions that require the artifact."},{"component":"pruning_or_archival_pathway","status":"direct","domain_realization":"Remove from active discovery, archive immutably, quarantine, or delete when permitted."},{"component":"preservation_exception_register","status":"direct","domain_realization":"Named holds for audit, publication reconstruction, disputes, safety, or historical value."},{"component":"deletion_audit_and_rollback_window","status":"direct","domain_realization":"Logged soft deletion with a time-bounded restoration period."},{"component":"resource_budget_and_storage_tier","status":"adapted","domain_realization":"Budget active search visibility, permissions, maintenance, and storage by lifecycle tier."},{"component":"review_cadence_and_revalidation_loop","status":"direct","domain_realization":"Milestone and periodic review of labels, owners, holds, and restore capability."},{"component":"legal_hold_or_compliance_override","status":"direct","domain_realization":"Sponsor, regulator, ethics, litigation, or institutional hold blocks disposition."},{"component":"hot_warm_cold_access_state","status":"adapted","domain_realization":"Current workflow, restricted reference, archive, and deep archive access states."},{"component":"orphan_layer_detection","status":"direct","domain_realization":"Identify artifacts with no live workflow, report, lineage, or audit dependency."},{"component":"supersession_marker","status":"direct","domain_realization":"Durable marker names the approved successor and effective date."},{"component":"layer_compaction_candidate_flag","status":"adapted","domain_realization":"Flag families for a navigation summary while retaining exact evidentiary versions."},{"component":"rehydration_or_restore_procedure","status":"direct","domain_realization":"Documented retrieval and execution/readability check for archived artifacts."},{"component":"deletion_impact_estimate","status":"direct","domain_realization":"Estimate affected analyses, publications, obligations, participants, and recovery difficulty."},{"component":"stale_context_detector","status":"adapted","domain_realization":"Detect conflicts with the approved protocol, current schema, software environment, or data cut."}],"mechanism_dispositions":[{"slug":"age_weighted_value_score","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"A composite age-discounted score could create false authority; old artifacts may be essential evidence and new ones may already be invalid.","counterfactual_removal":"No material loss because explicit multidimensional fields and gates perform triage without collapsing obligations."},{"slug":"archive_restore_test","disposition":"selected_supporting","contribution_type":"TEST_DESIGN","adaptation_or_rejection":"Sample archived plans, code, and data cuts across age and format bands and test readable, executable reconstruction.","counterfactual_removal":"Archival could silently become irrecoverable, weakening reproducibility claims."},{"slug":"cache_eviction_rule","disposition":"incompatible","contribution_type":"NONE","adaptation_or_rejection":"Recency-based automatic eviction cannot determine scientific, evidentiary, or regulatory value.","counterfactual_removal":"No loss; fixed-capacity fast-tier optimization is not the governing problem."},{"slug":"dependency_safe_delete_check","disposition":"selected_load_bearing","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Gate deletion on traced lineage, report, registration, audit, and reconstruction dependencies; uncertainty blocks deletion.","counterfactual_removal":"Cleanup could break reproducibility or erase evidence supporting results."},{"slug":"lifecycle_storage_tiering_policy","disposition":"selected_supporting","contribution_type":"OPERATIONAL","adaptation_or_rejection":"Adapt tiers primarily to discoverability and permissions, with storage cost secondary.","counterfactual_removal":"Artifacts could be labeled but would continue competing in active workflows."},{"slug":"log_rotation_and_cleanup_job","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"Generational position is too content-blind for heterogeneous scientific artifacts.","counterfactual_removal":"No material loss because disposition follows class, status, obligation, and dependency."},{"slug":"retention_schedule","disposition":"selected_load_bearing","contribution_type":"CORE_CAUSAL","adaptation_or_rejection":"Map artifact classes to active, archival, and destruction rules with cited institutional authority and holds.","counterfactual_removal":"Lifecycle decisions revert to inconsistent naming and discretionary cleanup."},{"slug":"soft_delete_quarantine_window","disposition":"selected_load_bearing","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Hide approved candidates from routine use while retaining logged restoration before permitted destruction.","counterfactual_removal":"Mistaken cleanup becomes immediately irreversible."},{"slug":"stale_layer_detection_dashboard","disposition":"selected_load_bearing","contribution_type":"CORE_CAUSAL","adaptation_or_rejection":"Inventory identity-resolved artifacts and flag semantic conflicts, missing owners, supersession, and validation drift.","counterfactual_removal":"Obsolete artifacts remain invisible as a class and continue masquerading as current."},{"slug":"time_to_live_ttl_policy","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"Fixed self-expiry is unsafe; supersession, study status, dependencies, and obligations determine disposition.","counterfactual_removal":"No loss; review dates can trigger reassessment without automatic expiry."},{"slug":"tombstone_or_deletion_marker","disposition":"selected_supporting","contribution_type":"OPERATIONAL","adaptation_or_rejection":"Leave a durable identity record naming deletion or supersession, rationale, date, and successor.","counterfactual_removal":"Absence becomes ambiguous and stale copies or references may revive obsolete authority."}],"causal_chain":["Resolve artifact identities and current/superseded relationships.","Detect obsolete or context-mismatched artifacts and remove them from ordinary authority.","Apply class-based retention, holds, dependency checks, and impact review.","Archive or quarantine reversibly; delete only when authorized and safe.","Preserve tombstones and test restoration, reducing wrong-version use without sacrificing reconstruction."],"baseline":"Shared drives or repositories retain nearly every version; filenames, folders, commit history, and staff memory indicate which artifact is final, with manual cleanup after pressure arises.","nearest_rival":"Immutable version control with an authoritative current-version pointer and access controls, but without class-specific retention, holds, restore testing, or governed disposition.","authority_safety":{"affected_parties":["study participants whose data or outcomes are represented","investigators and analysts","sponsors and institutions","ethics, regulatory, journal, and audit reviewers"],"decision_authority":"The study sponsor or accountable institution, advised by the principal investigator, lead statistician, data steward, records/privacy officers, and ethics or regulatory bodies where applicable.","authorized_first_step":"For one study over four weeks, build a read-only inventory, label candidate superseded artifacts, measure wrong-version retrieval in scripted search tasks, and simulate dependency and restore checks; change no source artifact or approved record.","excluded_actions":["automatic deletion","altering approved protocols, registrations, raw data, or audit logs","overriding legal, ethics, sponsor, or regulatory holds","treating age or low access as proof of scientific worthlessness"],"halt_rollback":"Halt if labeling hides a required current artifact, dependencies cannot be resolved, retrieval worsens materially, or any authority disputes status; remove pilot labels and restore prior search visibility from the logged inventory."}},"negative_tests":{"strongest_counterevidence":"If repositories already enforce one approved current artifact, immutable lineage, complete dependencies, governed retention, and routinely verified restoration, lifecycle failure is not an important source of errors.","analogy_break":"Scientific versions are not fungible storage objects: an obsolete plan can remain indispensable evidence of what was prespecified, so decay usually applies to operational authority rather than evidentiary value.","failure_condition":"The approach fails if artifact identity, approval status, or external dependencies cannot be observed reliably enough to distinguish superseded from required records.","problem_falsifier":"Across representative studies, obsolete artifacts are rarely encountered in active workflows, users reliably select approved versions, and artifact sprawl contributes negligible error, delay, or audit burden.","intervention_falsifier":"In the bounded test, lifecycle labels and gated discovery do not reduce wrong-version selection or review time versus the baseline, or they increase missed required artifacts or reconstruction failures.","risks":["False stale labels could hide valid minority or historical analyses.","Lifecycle metadata may create unwarranted confidence despite incomplete external dependencies.","Archiving can increase retrieval delay during audits or safety investigations.","Quarantine may conflict with hard-deletion privacy duties.","Governance burden may exceed benefit for small, short-lived studies."]},"null_rationale":null,"classification":{"candidate_kind":"MECHANISM_ADAPTATION","prior_art_status":"UNSEARCHED","evidence_maturity":"HYPOTHESIS"},"revision_change_log":{"revision_kind":"ORIGINAL","prior_problem_id":null,"prior_causal_lever_id":null,"problem_changed":false,"causal_lever_changed":false,"conceptual_changes":[],"operational_changes":[],"repairs_addressed":[]},"confidence":0.86,"generator_notes":"Closed-book structural transfer. The candidate manages the operational authority and retention state of accumulated experimental artifacts; it does not infer that old scientific records lose evidentiary value."}