{"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__tech_ethics_ai_governance","archetype_slug":"layer_decay_and_expiration_management","domain_slug":"tech_ethics_ai_governance","title":"Lifecycle gating for AI governance artifacts","opportunity_summary":"Evaluate a lifecycle registry and review process that binds AI-governance artifacts to owners, system versions, authority states, review dates, dependencies, holds, and successors. The proposed mechanism aims to prevent obsolete artifacts from remaining operationally authoritative without deleting evidence needed for accountability, appeals, audits, or incident reconstruction. Both defect prevalence and comparative effectiveness remain untested hypotheses.","adopter_authorizer":"The likely adopter is an organization operating a governed AI service. Its accountable AI-governance body authorizes lifecycle classes and operational use, subject to scoped vetoes from legal, privacy, records, security, and affected-system owners.","scores":{"meaningful_impact":{"score":4,"rationale":"If stale exceptions or approval conditions currently influence decisions, distinguishing current authority from preserved historical evidence could materially reduce unmanaged risk and improve audit, appeal, and reconstruction paths. The magnitude is conditional because defect prevalence and realized harm are unsupported."},"stakeholder_pull":{"score":3,"rationale":"Risk, compliance, legal, audit, system owners, investigators, and affected people have identifiable interests in current guidance and preserved evidence, but the packet contains no adoption inquiries, demand observations, or evidence that organizations prioritize this problem."},"incremental_advantage":{"score":4,"rationale":"Against the stated baseline and nearest rival, the composition adds artifact-class retention rules, explicit lifecycle states, dependency and hold gates, successor markers, reversible disposition, and restoration testing. Whether these additions outperform mature existing practice is unobserved."},"distinctiveness_plausibility":{"score":3,"rationale":"The integrated combination is specifically articulated and differs from the stated nearest rival, but prior art is explicitly unsearched and overlap with records-management, exception-expiry, policy-recertification, and archival-control systems is unresolved."},"technical_implementability":{"score":4,"rationale":"A read-only inventory, metadata registry, rule-based flagging, expert adjudication, quarantine, and restoration testing are technically plausible with conventional document and workflow systems. Implementability falls if ownership, version bindings, dependencies, or legal duties cannot be observed reliably."},"adoption_authority_feasibility":{"score":3,"rationale":"The packet identifies an accountable governance body and necessary specialist veto holders, making authority structurally clear. Coordinating several functions and resolving contested artifact legitimacy could nevertheless slow or prevent adoption."},"evidence_readiness":{"score":4,"rationale":"The candidate specifies a bounded eight-week read-only pilot, baseline comparison, problem and intervention falsifiers, safety outcomes, exclusions, and rollback conditions. It lacks a predefined sample, thresholds, adjudication protocol, and any observed results."},"safety_net_benefit":{"score":5,"rationale":"The proposal explicitly separates expiration of operational authority from destruction, preserves holds and affected-party rights, requires successor or deletion markers, excludes automated age-based destruction, and provides quarantine, restoration, freeze, and rollback mechanisms."},"scalability":{"score":3,"rationale":"Lifecycle states and metadata could be reused across services, but scaling depends on heterogeneous repositories, artifact classes, legal regimes, authority structures, dependency mapping, and sustained owner participation. No multi-service evidence is supplied."}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"Design and conduct the specified eight-week read-only pilot for one bounded AI service, including artifact sampling, inventory and metadata extraction, blinded expert applicability review, baseline comparison, safety adjudication, and reporting.","confidence":"MODERATE","assumptions":["The artifact population is accessible without major procurement or repository reconstruction.","Existing staff from governance, legal, privacy, records, security, and the system team can participate part-time.","The pilot uses configuration or lightweight analysis rather than production-grade integration.","No automatic deletion, authority change, or live control-path modification occurs."]},"initial_deployment_startup":{"band_2026_usd":"250K_TO_1M","scope":"Build a production-capable lifecycle registry and workflows for an initial organizational deployment, including repository connectors, access controls, lifecycle taxonomy, version binding, hold and dependency checks, audit logging, quarantine, restoration, and validation.","confidence":"LOW","assumptions":["Several existing document or ticket repositories require integration.","A bounded set of artifact classes and AI services is included initially.","Specialist review and compliance validation are required.","Existing identity, records, and workflow infrastructure can be reused."]},"operational_launch":{"band_2026_usd":"1M_TO_5M","scope":"Roll out the mechanism across multiple AI services or business units, migrate and adjudicate legacy artifacts, train owners and reviewers, establish governance operations, validate recovery, and monitor safety gates.","confidence":"LOW","assumptions":["Legacy metadata quality is uneven and requires substantial human review.","Multiple governance and records stakeholders must coordinate launch decisions.","The organization has a moderate portfolio of governed AI services.","The scope excludes enterprise-wide replacement of document, ticketing, and records platforms."]},"annual_recurring":{"band_2026_usd":"250K_TO_1M","scope":"Operate lifecycle reviews, exception revalidation, hold and dependency administration, registry maintenance, monitoring, audits, restoration exercises, training, and repository integration support.","confidence":"LOW","assumptions":["Review remains partly human because artifact validity is normative and authority-dependent.","The initial multi-service scope remains stable.","Major litigation, investigation, or regulatory-response surges are excluded.","Existing platform licenses and core infrastructure remain available."]}},"research_burden":"MODERATE","earliest_credible_horizon":"0_TO_3_MONTHS","pipeline_gates":{"recognizable_externally_supportable_problem":{"status":"YES","reason":"The candidate identifies a coherent failure mode: obsolete governance artifacts can retain apparent authority while indiscriminate cleanup can destroy accountability evidence. External evidence is still needed to establish material prevalence."},"identifiable_adopter_or_authorizer":{"status":"YES","reason":"An AI-operating organization is the adopter; its accountable AI-governance body authorizes lifecycle classes, with explicit scoped vetoes for legal, privacy, records, security, and affected-system owners."},"distinct_testable_incremental_claim":{"status":"YES","reason":"The testable claim is that lifecycle metadata and gated disposition improve blinded identification of currently applicable artifacts relative to existing review, without increasing missed controls, protected-record exposure, or restoration failures."},"bounded_next_evidence_step":{"status":"YES","reason":"The packet authorizes an eight-week, one-service, read-only pilot with expert adjudication, a baseline comparison, explicit falsifiers, and no automatic deletion or authority changes."},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"The first step is read-only and preserves specialist vetoes, holds, review rights, visibility, and rollback. The packet also defines zero-tolerance halt conditions for protected records, active controls, and affected-party rights."},"implementation_cost_scope_and_range":{"status":"UNCERTAIN","reason":"The pilot scope is bounded, but repository count, artifact volume, metadata quality, integration complexity, staffing availability, legal obligations, and eventual deployment breadth are unspecified, so only assumption-dependent broad bands are supportable."}},"blocking_evidence":["No representative audit establishes how often lifecycle fields, ownership, version scope, dependencies, or hold status are missing.","No evidence shows that users rely on superseded artifacts more often than on current artifacts.","No comparative pilot demonstrates improved identification of applicable artifacts or acceptable safety outcomes.","The observability of ownership, inbound dependencies, legal duties, and version bindings is untested.","Prior art is unsearched, so distinctiveness from existing governance and records-management systems is unknown.","Production integration scope, legacy adjudication burden, and recurring review workload are unspecified."],"next_evidence_step":"Pre-register and run the authorized eight-week read-only pilot on one AI service. Sample its exception and approval artifacts, compare blinded reviewers using the lifecycle-gating view against reviewers using the existing process, and measure correct identification of currently applicable artifacts. Reject the intervention if it provides no improvement or causes any missed active control, protected-record exposure, impaired review right, or restoration failure under the preset zero-tolerance safety gates.","research_questions":["What proportion of a predefined representative artifact sample lacks enforceable lifecycle state, owner, system-version binding, review date, supersession link, dependency status, or hold status?","How often do blinded users classify superseded artifacts as currently authoritative under the existing process?","Does the lifecycle-gating view improve blinded applicability judgments relative to the baseline without increasing missed controls or protected-record risks?","Can ownership, authority, dependencies, holds, and legal duties be observed with sufficient reliability to support disposition decisions?","How much of the proposed composition is already present in governance registries, exception-expiry workflows, records schedules, policy-recertification systems, or archival controls?","What human adjudication, integration, and recurring-review workload would a multi-service deployment require?","Do quarantine and archival processes preserve timely appeals, investigations, audits, and tested restoration?","What mechanism prevents holds and exceptions from becoming indefinitely permanent and recreating the accumulation problem?"] ,"recommendation":"PILOT_NOW","uncertainty_constraints":["Problem prevalence and effect size are unsupported hypotheses.","Stakeholder demand and willingness to adopt have not been observed.","World novelty, competitive prevalence, and market size are unmeasured.","The nearest-rival comparison is supplied by the sealed candidate and is not externally validated.","Cost bands are resource-equivalent planning ranges based only on the described scope and stated assumptions.","Normative validity cannot safely be inferred from age, access frequency, or technical dependency alone.","Deployment feasibility depends on metadata observability and cross-functional authority coordination."],"closed_book_prior_art_boundary":"No novelty or prior-art conclusion is supported. The packet labels prior art as unsearched; therefore the assessment only evaluates the internal plausibility of incremental gain relative to the candidate's stated baseline and nearest rival, without asserting distance from real-world records-management, model-governance, exception-expiry, policy-recertification, or archival-control practices."}