{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp03_full320_20260801","cell_id":"layer_decay_and_expiration_management__library_information_science","trajectory_id":"R","attempt_index":0,"archetype_slug":"layer_decay_and_expiration_management","domain_slug":"library_information_science","decision":"CANDIDATE","problem_id":"stale_superseded_discovery_layers","causal_lever_id":"discovery_lifecycle_state_gating","proposal":{"problem":"Superseded, invalid, or context-stale catalog, holdings, knowledge-base, repository, and research-guide records can remain exposed as current in library discovery systems because deposition is routine but lifecycle state, expiry, dependency, and preservation rules are incomplete. Patrons then encounter false availability or obsolete guidance, while indiscriminate deletion risks erasing provenance, citations, and historically distinct editions.","actors_substrate":["patrons and researchers","catalogers and electronic-resource librarians","archivists and records officers","integrated library system, discovery index, knowledge base, institutional repository, and research guides","bibliographic, holdings, authority, link, license, and version records"],"observable_state":"Measure the share of discovery-exposed records independently adjudicated as superseded, unavailable, or context-stale; failed access attempts attributable to them; unresolved inbound dependencies; time to disposition; and restoration success for archived records.","consequence":"Stale layers compete with current descriptions, cause failed retrieval and misinterpretation, consume maintenance attention, and make cleanup hazardous because historical and dependency value is unclear.","affected_objective":"Improve discovery reliability and maintainability while preserving provenance, lawful retention, reconstruction, and access to historically meaningful records.","structural_mapping":[{"archetype_element":"sequential deposits","domain_realization":"Successive harvests, vendor updates, local edits, holdings changes, guide revisions, and metadata versions accumulate across library systems.","claim_kind":"INFERENCE"},{"archetype_element":"stale layers remain active","domain_realization":"Superseded or invalid records remain searchable without an authoritative lifecycle or supersession state.","claim_kind":"HYPOTHESIS"},{"archetype_element":"interpretive and operational harm","domain_realization":"Visible stale records can imply current availability or guidance and generate failed retrieval or staff remediation.","claim_kind":"HYPOTHESIS"},{"archetype_element":"deletion threatens memory","domain_realization":"Records may support citations, provenance, edition distinctions, audits, local history, or reconstruction.","claim_kind":"INFERENCE"},{"archetype_element":"managed disposition","domain_realization":"Records are refreshed, labeled, suppressed, archived, compacted, preserved by exception, or reversibly removed according to state.","claim_kind":"HYPOTHESIS"}],"component_map":[{"component":"layer_inventory_and_identity_map","status":"direct","domain_realization":"Canonical resource and edition identities linked across catalog, knowledge base, repository, and guide records."},{"component":"deposition_order_and_age_index","status":"direct","domain_realization":"Creation, harvest, validation, revision, and supersession timestamps."},{"component":"retention_policy_matrix","status":"adapted","domain_realization":"Rules by record class, provenance, collection mandate, and evidentiary value."},{"component":"decay_function_or_aging_rule","status":"adapted","domain_realization":"Validation confidence declines with elapsed time and contextual change, not age alone."},{"component":"expiration_trigger_definition","status":"direct","domain_realization":"Broken access, ended license, supersession, owner loss, or overdue validation triggers review."},{"component":"layer_value_and_risk_score","status":"adapted","domain_realization":"Uniqueness, use, provenance, misleadingness, sensitivity, and reconstruction cost."},{"component":"dependency_and_reconstruction_check","status":"direct","domain_realization":"Check citations, authority links, holdings, workflows, exports, and reconstruction paths."},{"component":"pruning_or_archival_pathway","status":"direct","domain_realization":"Refresh, label, suppress, merge, archive, or delete."},{"component":"preservation_exception_register","status":"direct","domain_realization":"Named exceptions for rare, local-history, evidentiary, or preservation records."},{"component":"deletion_audit_and_rollback_window","status":"direct","domain_realization":"Logged soft suppression with a time-boxed restoration window."},{"component":"resource_budget_and_storage_tier","status":"adapted","domain_realization":"Discovery-active, staff-accessible archive, and preservation tiers."},{"component":"review_cadence_and_revalidation_loop","status":"direct","domain_realization":"Periodic validation by class and risk."},{"component":"legal_hold_or_compliance_override","status":"direct","domain_realization":"Privacy, litigation, donor, records-law, and license overrides."},{"component":"hot_warm_cold_access_state","status":"adapted","domain_realization":"Current discovery, limited staff retrieval, and preservation-only access states."},{"component":"orphan_layer_detection","status":"direct","domain_realization":"Find records with no valid holdings, successor, citation, workflow, or collection relation."},{"component":"supersession_marker","status":"direct","domain_realization":"Durable redirect or notice naming the current record and prior identity."},{"component":"layer_compaction_candidate_flag","status":"adapted","domain_realization":"Flag duplicate harvests or redundant snapshots for lossless consolidation."},{"component":"rehydration_or_restore_procedure","status":"direct","domain_realization":"Restore archived metadata, relationships, and access state into a test index."},{"component":"deletion_impact_estimate","status":"direct","domain_realization":"Estimate affected users, references, collections, and restoration difficulty."},{"component":"stale_context_detector","status":"adapted","domain_realization":"Signals include broken links, contradictory holdings, ended licenses, departed owners, and superseding records."}],"mechanism_dispositions":[{"slug":"age_weighted_value_score","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"Age is weak evidence for bibliographic staleness; composite scoring risks false precision. Use class rules and adjudicated signals.","counterfactual_removal":"No causal step changes; prioritization becomes less numerically ranked."},{"slug":"archive_restore_test","disposition":"selected_load_bearing","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Test sampled records across formats and age bands for metadata, relationship, and index restoration.","counterfactual_removal":"Archival cannot safely substitute for active access because recoverability remains unproven."},{"slug":"cache_eviction_rule","disposition":"incompatible","contribution_type":"NONE","adaptation_or_rejection":"Access recency may allocate fast storage but cannot determine descriptive authority or preservation.","counterfactual_removal":"No lifecycle-governance step changes."},{"slug":"dependency_safe_delete_check","disposition":"selected_load_bearing","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Gate destructive action on live inbound-reference and reconstruction checks.","counterfactual_removal":"Deletion could break citations, relationships, exports, or reconstruction, hard-gating viability."},{"slug":"lifecycle_storage_tiering_policy","disposition":"selected_supporting","contribution_type":"OPERATIONAL","adaptation_or_rejection":"Map hot/warm/cold to discovery-active, staff archive, and preservation access.","counterfactual_removal":"Disposition still works, but safe retention costs more and archive latency is unmanaged."},{"slug":"log_rotation_and_cleanup_job","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"Content-blind generational rotation would erase meaningful editions and provenance.","counterfactual_removal":"No loss; library records are not a uniformly disposable append-only stream."},{"slug":"retention_schedule","disposition":"selected_load_bearing","contribution_type":"CORE_CAUSAL","adaptation_or_rejection":"Govern record classes, minimums, maximums, preservation exceptions, and holds.","counterfactual_removal":"Disposition loses legitimate, consistent authority and exceptions become ad hoc."},{"slug":"soft_delete_quarantine_window","disposition":"selected_load_bearing","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Suppress first, log the decision, preserve restoration, and destroy only after review.","counterfactual_removal":"The bounded trial and later cleanup become unacceptably irreversible."},{"slug":"stale_layer_detection_dashboard","disposition":"selected_load_bearing","contribution_type":"CORE_CAUSAL","adaptation_or_rejection":"Unify identities and surface meaning-based stale candidates with owners and evidence.","counterfactual_removal":"The intervention cannot reliably identify which exposed layers require action."},{"slug":"time_to_live_ttl_policy","disposition":"selected_supporting","contribution_type":"OPERATIONAL","adaptation_or_rejection":"Use renewable validation leases that trigger review or labeling, never age-only destruction.","counterfactual_removal":"Lifecycle review becomes sweep-based rather than self-triggering but remains viable."},{"slug":"tombstone_or_deletion_marker","disposition":"selected_load_bearing","contribution_type":"CORE_CAUSAL","adaptation_or_rejection":"Preserve prior identity, deletion state, reason, and successor resolution.","counterfactual_removal":"Suppressed records can become ambiguous gaps, leave dangling references, or be reimported."}],"causal_chain":["Inventory and identity resolution reveal deposited record layers and their relationships.","Context signals and validation leases identify candidates whose current authority is doubtful.","Retention rules, holds, value, and dependency checks determine permissible disposition.","Labels, supersession redirects, or suppression stop doubtful layers masquerading as current while archives preserve memory.","Soft deletion, tombstones, and tested restoration limit irreversible loss.","HYPOTHESIS: this reduces stale-record exposure and attributable access failures without materially suppressing valid resources."],"baseline":"Ordinary practice is complaint-driven link repair, periodic vendor reloads, and manual record cleanup, with many records remaining discoverable until someone notices a failure.","nearest_rival":"A periodic link checker plus manual recataloging, which detects some access failures but lacks unified identity, explicit lifecycle states, dependency gates, and reversible disposition.","authority_safety":{"affected_parties":["patrons and researchers","catalog and electronic-resource staff","archivists and records officers","rights holders and donors","people or communities described in records","partner libraries and downstream metadata users"],"decision_authority":"The library's catalog/discovery governance owner authorizes visibility changes; the records officer, archivist, privacy or legal officer, and collection steward control applicable holds and preservation exceptions.","authorized_first_step":"Run an eight-week pilot on 500 electronic-resource records in one discovery index: compare 250 lifecycle-reviewed records with 250 usual-practice records; permit only labels, redirects, and reversible suppression after staff adjudication, while measuring stale exposure, failed access, staff effort, false suppression, and restore fidelity.","excluded_actions":["hard deletion during the pilot","age-only suppression","altering preservation masters","overriding legal, donor, privacy, or historical holds","suppressing unique records with unresolved dependencies"],"halt_rollback":"Halt new suppressions and restore quarantined records if any hold is breached, restoration fails, unresolved dependencies appear, or adjudicated valid-record suppression exceeds the pre-registered tolerance."}},"negative_tests":{"strongest_counterevidence":"The apparent stale-record burden may primarily arise from upstream vendor-feed errors, transient authentication failures, or ranking behavior; lifecycle treatment of local layers would then address the wrong cause.","analogy_break":"A bibliographic record is not a disposable cache entry: old records may describe distinct editions, past holdings, or historical contexts accurately. Age cannot stand in for invalidity.","failure_condition":"The design fails if identities and dependencies cannot be resolved reliably enough to distinguish obsolete duplicates from distinct or historically necessary records.","problem_falsifier":"A representative audit finds that adjudicated stale or superseded layers are rare, are already clearly marked, and do not materially contribute to failed retrieval, misleading discovery, or maintenance work.","intervention_falsifier":"Compared with usual practice, the bounded pilot produces no meaningful reduction in adjudicated stale exposure or attributable access failures, or its false suppression, labor, unresolved-reference, or restoration harms exceed the pre-registered tolerance.","risks":["valid but infrequently used resources may be demoted","automated signals may encode collection or language bias","suppression may conceal contested cataloging history","archives may be technically present but practically unrecoverable","quarantine may conflict with required destruction timelines","tombstones may themselves accumulate or expose sensitive metadata"]},"null_rationale":null,"classification":{"candidate_kind":"MECHANISM_COMPOSITION","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":0.84,"generator_notes":"Closed-book structural inference from the supplied packet. Empirical effectiveness and prevalence claims remain hypotheses; no novelty or prior-art claim is made."}