{"schema_version":1,"assessment_id":"eoa_inverse_innovation_exp03_opportunity320_20260801","source_experiment_id":"eoa_inverse_innovation_exp03_full320_20260801","cell_id":"layer_decay_and_expiration_management__robotics_automation","archetype_slug":"layer_decay_and_expiration_management","domain_slug":"robotics_automation","title":"Dependency-Gated Lifecycle Management for Robot World-Model Layers","opportunity_summary":"Evaluate a lifecycle system that identifies competing robot world-model layers, removes superseded layers from default selection, preserves lineage and held evidence, tiers recoverable history, and permits deletion only through dependency checks and reversible controls. The opportunity is conditional on demonstrating that stale selectable layers or accumulated artifacts create meaningful operational burden and that the composition outperforms newest-validated-only selection.","adopter_authorizer":"A fleet safety owner would authorize production-selection changes, a data steward would authorize archival or destruction, and an investigation or compliance owner would control holds; a fleet operator with these roles is the necessary pilot partner.","scores":{"meaningful_impact":{"score":4,"rationale":"If the diagnosed condition exists, preventing localization or planning against obsolete geometry while preserving rollback and incident evidence could materially improve safety, operability, and storage governance. Impact remains conditional because the packet supplies no evidence of prevalence or realized harm."},"stakeholder_pull":{"score":3,"rationale":"The proposal identifies concrete concerns for operators, safety engineers, data stewards, site owners, and investigators, but contains no evidence that these parties currently experience material burden or would prioritize adoption."},"incremental_advantage":{"score":4,"rationale":"Compared with newest-validated-only selection, the proposal adds explicit supersession, dependency-gated disposition, holds, archival restore testing, bounded active storage, and auditable two-phase deletion. Whether these additions produce enough benefit to justify their complexity is untested."},"distinctiveness_plausibility":{"score":3,"rationale":"The dependency-gated composition is coherently differentiated from the stated selection-only rival, but its prior-art status is explicitly unsearched and the packet cannot establish distinctiveness relative to existing robotics or data-governance systems."},"technical_implementability":{"score":3,"rationale":"The candidate specifies implementable-seeming states, registry controls, replay, tiering, tombstones, and restore tests, but success depends on resolving artifact identity, runtime references, dependencies, change evidence, and task-specific fidelity. The packet identifies failure of reference resolution as a decisive risk."},"adoption_authority_feasibility":{"score":3,"rationale":"The relevant authorizers and their separate powers are explicitly identified, and the read-only first step avoids production changes. Full adoption nevertheless requires coordination across safety, stewardship, investigation, and possibly customer or site-owner authority."},"evidence_readiness":{"score":4,"rationale":"A bounded 30-day, one-site, read-only pilot is specified with replay, stale and orphan flagging, archive restoration, a credible newest-validated-only comparison, halt criteria, and separate problem and intervention falsifiers. Thresholds and usable data remain to be established."},"safety_net_benefit":{"score":5,"rationale":"The proposal has unusually strong safeguards for an evidence phase: no hard deletion or production map loading, protected holds, two-phase deletion, tombstones, restore tests, explicit halts, label rollback, and retention of all bytes during the pilot."},"scalability":{"score":3,"rationale":"Registry states and policy controls could in principle apply across missions and fleets, but heterogeneous stores, robot-local artifacts, unresolved references, site-specific validity evidence, cold-tier latency, and accumulating lifecycle metadata may limit scaling."}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"Prepare and run the specified 30-day read-only pilot at one site, including inventory and identity resolution, historical-selection replay, stale and orphan candidate analysis, archive-sample restoration, threshold definition, safety review, and reporting.","confidence":"MODERATE","assumptions":["One site and a bounded fleet are included.","Existing artifact stores and historical logs are accessible without production modification.","The work requires robotics, data, safety, and evaluation labor plus partner coordination.","No new production hardware or destructive data migration is included."]},"initial_deployment_startup":{"band_2026_usd":"250K_TO_1M","scope":"Build and validate a staging implementation with lifecycle metadata, lineage and dependency capture, supersession controls, archive interfaces, restore testing, holds, tombstones, monitoring, and integration with the map or artifact registry.","confidence":"LOW","assumptions":["A usable registry or equivalent integration point exists.","Identity and runtime references can be resolved without redesigning the autonomy stack.","Validation remains in staging or shadow mode until safety thresholds are met.","The band includes engineering, safety-case work, data remediation, software, and partner coordination."]},"operational_launch":{"band_2026_usd":"250K_TO_1M","scope":"Launch guarded production selection and lifecycle operations for an initial fleet at one site, including acceptance testing, operator procedures, access controls, monitoring, rollback exercises, training, and coordinated approvals; automatic hard deletion remains outside the initial launch.","confidence":"LOW","assumptions":["The evidence pilot supports continuation.","Dependency recall and archive fidelity meet predeclared thresholds.","Production integrations are limited to an initial site and fleet.","Site-owner, safety, stewardship, and investigation approvals can be coordinated."]},"annual_recurring":{"band_2026_usd":"250K_TO_1M","scope":"Operate and audit lifecycle policies for the initial deployment, including periodic revalidation, exception and hold review, restore exercises, storage and software, incident support, metadata maintenance, and safety oversight.","confidence":"LOW","assumptions":["The scope is one initial production fleet or site rather than an enterprise-wide rollout.","Artifact growth and cold-storage volume are not known.","Dedicated engineering and governance support remains necessary.","Major autonomy-platform redesigns and multi-site expansion are excluded."]}},"research_burden":"HIGH","earliest_credible_horizon":"3_TO_12_MONTHS","pipeline_gates":{"recognizable_externally_supportable_problem":{"status":"UNCERTAIN","reason":"The packet specifies observable stale-selection and accumulation conditions plus a clear problem falsifier, but provides no external evidence that these conditions occur or cause meaningful burden in a target fleet."},"identifiable_adopter_or_authorizer":{"status":"YES","reason":"The candidate explicitly assigns production exclusion to the fleet safety owner, archival or destruction to the data steward, and holds to the investigation or compliance owner."},"distinct_testable_incremental_claim":{"status":"YES","reason":"The lifecycle composition can be compared with newest-validated-only selection on stale-layer detection, active-store boundedness, dependency recall, and archive-restore fidelity."},"bounded_next_evidence_step":{"status":"YES","reason":"The sealed candidate authorizes a 30-day, one-site, read-only inventory, replay, flagging, and restore pilot with no production-selection or deletion changes."},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"The first evidence step is non-destructive and read-only, preserves all bytes, respects holds, identifies relevant authorities, and includes explicit halt conditions for mission-critical exclusions, incomplete dependencies, restore failures, and excess false positives."},"implementation_cost_scope_and_range":{"status":"UNCERTAIN","reason":"The one-site pilot provides a bounded evidence scope, but store count, fleet size, data quality, integration architecture, dependency resolution effort, compliance obligations, and production rollout scale are absent, so implementation ranges remain low-confidence."}},"blocking_evidence":["Audit and replay evidence that stale or superseded layers are actually eligible for operational selection or that accumulation causes meaningful storage, search, maintenance, or governance burden.","Measured artifact-identity and runtime-reference coverage sufficient to distinguish superseded, dependent, and orphan layers without disrupting operations.","A predeclared comparison showing improvement over newest-validated-only selection in stale-layer detection or active-store boundedness.","Evidence that dependency recall and false-positive stale-flag rates meet safety-approved thresholds.","Stratified archive-restore results meeting predeclared fidelity and retrieval-time requirements for rollback, relocalization, and incident reconstruction.","Partner-specific evidence on integration effort, recurring governance burden, authority coordination, and resource requirements.","Scoped prior-art research before making any originality or market-distinctiveness claim."],"next_evidence_step":"With an authorized fleet partner, run the specified 30-day read-only pilot at one site. Predeclare dependency-recall, false-positive, restore-fidelity, retrieval-time, and active-store-boundedness thresholds; inventory and resolve layer identities, replay historical selections, compare the lifecycle classifier with newest-validated-only selection, and restore a stratified archive sample. Reject the problem if stale layers are never selectable and accumulation produces no meaningful burden; reject the intervention if it adds no detection or boundedness benefit or misses any required safety threshold.","research_questions":["How often are stale or superseded layers eligible for robot or operator selection, and what observable consequences follow?","What fraction of artifacts and runtime references can be resolved into reliable identities, lineage, and dependency relationships?","Which change and task-validity signals outperform age as evidence that a layer should be superseded or revalidated?","Does the lifecycle composition improve stale-layer detection or active-store boundedness over newest-validated-only selection at acceptable false-positive rates?","What dependency-recall threshold is required before archival or deletion could be considered safe?","Can archived layers be restored with adequate fidelity and retrieval speed for rollback, relocalization, and incident response?","How should holds, destruction deadlines, compaction fidelity, and customer or site-owner authority interact?","What existing robotics, map-management, safety-case, and data-governance systems constitute relevant prior art?"],"recommendation":"PARTNERED_RESEARCH","uncertainty_constraints":["Problem prevalence, stakeholder demand, and realized operational impact are unsupported hypotheses in the sealed packet.","No external evidence establishes that the proposed composition is novel or uncommon.","Effectiveness relative to newest-validated-only selection has not been measured.","Artifact identity, lineage, dependency coverage, and runtime-reference resolution may be inadequate.","Age cannot be treated as proof of invalidity because spatial layers may remain valid for long periods or become invalid immediately after change.","Cost bands are resource-equivalent planning ranges based on stated scope, not observed prices or quotations.","Production safety, compliance obligations, fleet heterogeneity, data volumes, and site-owner constraints are unknown."],"closed_book_prior_art_boundary":"Prior-art status is explicitly UNSEARCHED. This assessment recognizes only the proposal's internal differentiation from its stated nearest rival and makes no claim about novelty, prevalence, market size, existing deployments, realized impact, or exact cost."}