{"schema_version":1,"research_id":"eoa_inverse_innovation_exp03_external48_20260801","source_assessment_id":"eoa_inverse_innovation_exp03_opportunity320_20260801","cell_id":"layer_decay_and_expiration_management__innovation_entrepreneurship","selection_stratum":"DEPLOYABLE_PRIORITY","search_queries":["venture experiment repository superseded evidence lifecycle status decision logs stale knowledge research","site:learn.microsoft.com Purview retention labels disposition review records hold official","site:atlassian.com Confluence content lifecycle archive stale pages official","architecture decision record superseded status official documentation","startup experiment documentation learning evidence repository empirical study decisions pivot records","product teams outdated documentation wrong decisions empirical study knowledge repositories stale information","organizational memory outdated knowledge decision making empirical research stale knowledge","innovation teams experiment documentation repositories empirical study","site:archives.gov records management disposition temporary records legal hold official lifecycle","site:iso.org ISO 15489 records management controls disposition official","site:learn.microsoft.com purview retention labels legal hold event based retention official","site:docs.growthbook.io experiment archive status official","site:docs.statsig.com experiments archive completed official","site:support.optimizely.com experiment archive results official","Knowledge Retention Challenges in Information Systems Development Teams DOI"],"sources":[{"source_id":"S1","title":"Knowledge Retention Challenges in Information Systems Development Teams: A Revelatory Story From Developers in New Zealand","publisher":"International Journal of Knowledge Management / IGI Global","url":"https://doi.org/10.4018/IJKM.291096","source_class":"PRIMARY_RESEARCH","publication_date":"2022-01-01","accessed_at":"2026-08-02","claims_supported":["A case study using interviews, organizational documents, and observations found out-of-date, poorly categorized documentation and difficulty obtaining relevant, correct information in an information-systems development organization.","The study associated ineffective knowledge retention with search burden, role stress, slower delivery, lower project quality, and reduced information-systems development agility.","The evidence is adjacent rather than venture-specific and comes from one organization."]},{"source_id":"S2","title":"How Do Software Startups Pivot? Empirical Results from a Multiple Case Study","publisher":"arXiv","url":"https://arxiv.org/abs/1711.00760","source_class":"PRIMARY_RESEARCH","publication_date":"2017-10-29","accessed_at":"2026-08-02","claims_supported":["A multiple-case study of four software startups found that startups made business and technology pivots triggered by factors including negative customer feedback.","The study supports the premise that startup decision context changes, but it does not test whether stale repository artifacts cause bad decisions."]},{"source_id":"S3","title":"ISO 15489-1:2016 — Information and documentation — Records management — Part 1: Concepts and principles","publisher":"International Organization for Standardization","url":"https://www.iso.org/standard/62542.html","source_class":"STANDARD","publication_date":"2016-04","accessed_at":"2026-08-02","claims_supported":["The standard covers records, metadata, records systems, assigned responsibilities, monitoring, training, records controls, and processes for creating, capturing, and managing records over time.","Structured lifecycle governance for business records is established standards-based practice rather than a new general mechanism."]},{"source_id":"S4","title":"Architectural decision record process","publisher":"Amazon Web Services","url":"https://docs.aws.amazon.com/prescriptive-guidance/latest/architectural-decision-records/adr-process.html","source_class":"OFFICIAL_GUIDANCE","publication_date":"undated","accessed_at":"2026-08-02","claims_supported":["AWS guidance defines decision-record states, ownership, review, immutable accepted records, and a Superseded state when an accepted new record replaces an old one.","The resulting decision log preserves context, consequences, stakeholders, versions, timestamps, rejected decisions, and decision history.","This is close prior art for preventing obsolete decisions from appearing authoritative while retaining provenance."]},{"source_id":"S5","title":"Manage experiments and campaigns","publisher":"Optimizely","url":"https://support.optimizely.com/hc/en-us/articles/4410289799309-Manage-experiments-and-campaigns","source_class":"OFFICIAL_PRODUCT_DOCUMENTATION","publication_date":"2025-08-21","accessed_at":"2026-08-02","claims_supported":["Optimizely experiments already have Draft, Running, Paused, Concluded, and Archived states and can be filtered by status.","Concluding an experiment records a Positive, Negative, or Inconclusive outcome and a findings description.","Archiving creates a results snapshot that remains referenceable; the product archives rather than deletes experiments."]},{"source_id":"S6","title":"Disposition of content","publisher":"Microsoft","url":"https://learn.microsoft.com/en-us/purview/disposition","source_class":"OFFICIAL_PRODUCT_DOCUMENTATION","publication_date":"undated","accessed_at":"2026-08-02","claims_supported":["Microsoft Purview implements retention-label-triggered disposition review, designated reviewer roles, multiple review stages, relabeling, extensions, archive options, audit history, and permission separation.","Deletion can be suspended for litigation or audit, and content pending review is not permanently deleted until an authorized final reviewer acts unless optional auto-approval is deliberately configured.","The documentation demonstrates both technical implementability and the need for records-management authority and audit controls."]},{"source_id":"S7","title":"Archive content items","publisher":"Atlassian","url":"https://support.atlassian.com/confluence-cloud/docs/archive-pages/","source_class":"OFFICIAL_PRODUCT_DOCUMENTATION","publication_date":"undated","accessed_at":"2026-08-02","claims_supported":["Confluence archiving removes outdated content from the content tree and quick search while preserving it for advanced search and direct links.","Archived links receive a visible Archived marker, archived items preserve their state, and items can be restored.","Archiving permissions are role-controlled, and Atlassian warns that inherited permissions should be checked before archiving."]},{"source_id":"S8","title":"Preparing Disposition Instructions","publisher":"U.S. National Archives and Records Administration","url":"https://www.archives.gov/records-mgmt/scheduling/instructions","source_class":"GOVERNMENT_OR_REGULATOR","publication_date":"2019-06-06","accessed_at":"2026-08-02","claims_supported":["NARA guidance requires inventory, evaluation, explicit retention periods, clear disposition instructions, and authorization before destruction.","Event-based retention after completion, supersession, revision, acceptance, or rejection is established records-management practice.","NARA cautions that subjective destruction rules such as 'when no longer needed' can cause loss of control and inadvertent destruction of still-needed records."]}],"problem_evidence":{"support":"MODERATE","rationale":"The problem is externally recognizable but not established at the claimed venture-wide prevalence or consequence. A primary case study found that out-of-date and poorly structured documentation impeded access to relevant, correct information and was associated with search burden and reduced development agility. Startup case research confirms that venture context and decisions change through pivots. Optimizely's experiment-specific status, outcome, filtering, and archiving features also show that experiment accumulation and lifecycle handling are real product concerns. None of these sources measures how often superseded venture artifacts are mistaken for current authority or alter roadmap, funding, or go-to-market decisions.","source_ids":["S1","S2","S5"]},"stakeholder_evidence":{"support":"WEAK","rationale":"AWS, Microsoft, Atlassian, and Optimizely identify operational owners, reviewers, administrators, analysts, and program teams for closely analogous workflows, making an adopter and authorizer structure credible. However, the search found no venture-founder interviews, procurement evidence, budgets, adoption requests, or willingness-to-pay data for the proposed combined workflow.","source_ids":["S4","S5","S6","S7"]},"prior_art":{"proximity":"SUBSTANTIAL_COLLISION","closest_analogues":[{"name":"Optimizely experiment lifecycle and archive","similarity":"Experiment-specific prior art already supplies lifecycle states, explicit conclusion outcomes, status filtering, preserved result snapshots, and archive-instead-of-delete behavior.","remaining_difference":"It does not establish a cross-artifact supersession graph, class-based legal retention and holds, dependency-gated disposition, or comparative evidence that authority-facing labels reduce incorrect-current selections.","source_ids":["S5"]},{"name":"AWS architectural decision record lifecycle","similarity":"ADR practice already assigns owners and states, makes accepted records immutable, creates a replacement record when evidence changes, marks the old record Superseded, and preserves the historical decision log and rationale.","remaining_difference":"It is scoped to architectural decisions rather than heterogeneous venture experiments, market analyses, assumption versions, and pitch claims; it does not provide retention schedules, holds, quarantine, or restoration testing.","source_ids":["S4"]},{"name":"Microsoft Purview records management combined with Confluence archiving","similarity":"Established products already combine labels, role-gated and audited disposition, litigation or audit suspension, relabeling and extensions, archive options, search suppression, visible archive markers, preserved links, and restoration.","remaining_difference":"These controls manage record status and discoverability but do not by themselves encode which empirical claim or venture assumption is currently decision-authoritative, nor test effects on venture decisions.","source_ids":["S6","S7"]},{"name":"ISO 15489 and NARA lifecycle governance","similarity":"Standards and government guidance already establish metadata, responsibilities, monitoring, inventory, explicit retention, authorized disposition, and event-based triggers including supersession.","remaining_difference":"They do not specify a venture-evidence authority layer or the proposed retrieval and incorrect-selection outcome measures.","source_ids":["S3","S8"]}],"distinctive_claim_remaining":"The remaining testable claim is narrow: applying a linked epistemic-authority layer—current, superseded, review-due, and explicit replacement relationships—together with records-compliant disposition gates across heterogeneous venture evidence will reduce time to find authoritative evidence and incorrect-current selections versus an otherwise identical untreated corpus, without increasing missed dependencies, contested classifications, or failed retrievals. The search did not establish that outcome.","confidence":"HIGH"},"implementation_evidence":{"support":"STRONG","rationale":"The constituent mechanisms are implemented in current products and formal guidance: experiment states and preserved snapshots, immutable and superseded decision records, search-suppressed but restorable archives, retention labels, authorized multi-stage disposition, holds, audit history, and event-based retention. A read-only mirrored evaluation therefore requires configuration and evaluation work rather than new foundational technology. Cross-tool identity resolution, tacit dependencies, and venture-specific authority classification remain unverified implementation challenges.","source_ids":["S3","S4","S5","S6","S7","S8"]},"scores":{"meaningful_impact":{"score":3,"rationale":"Adjacent primary evidence links outdated or disorganized documentation with search difficulty and reduced development agility, and startup decisions do change through pivots. Direct evidence that stale venture artifacts materially alter roadmap, investment, or go-to-market decisions is absent.","source_ids":["S1","S2"]},"stakeholder_pull":{"score":2,"rationale":"Multiple vendors support analogous workflows and define operational roles, but no venture-specific demand, budget, procurement, or adoption evidence was found.","source_ids":["S4","S5","S6","S7"]},"incremental_advantage":{"score":3,"rationale":"Relative to ad hoc cleanup, the combined epistemic-authority and records-governance treatment is more coherent and testable. Its individual mechanisms are already established, and no comparative outcome evidence shows that the combination outperforms simpler experiment status plus archive practices.","source_ids":["S4","S5","S6","S7","S8"]},"distinctiveness_plausibility":{"score":2,"rationale":"The bounded search found substantial collision across experiment platforms, ADR practice, content archives, and records management. Only the venture-specific cross-artifact configuration and comparative outcome claim remain plausibly distinctive.","source_ids":["S3","S4","S5","S6","S7","S8"]},"technical_implementability":{"score":4,"rationale":"Existing platforms demonstrate nearly all components, so a bounded mirror and metadata treatment are feasible. Reliable identity resolution and dependency detection across heterogeneous repositories remain material complications.","source_ids":["S4","S5","S6","S7"]},"adoption_authority_feasibility":{"score":4,"rationale":"Close analogues assign record owners, project-team reviewers, space administrators, and disposition-specific roles, supporting a credible split between content authority and irreversible-disposition authority. Organization-specific legal, privacy, and records mappings would still be required.","source_ids":["S3","S4","S6","S7","S8"]},"evidence_readiness":{"score":5,"rationale":"The candidate permits a small, read-only-first corpus with an untreated comparison and measurable retrieval, selection, dependency, contest, and restoration outcomes. Existing experiment and lifecycle states make a non-production mirror practical.","source_ids":["S4","S5","S7"]},"safety_net_benefit":{"score":4,"rationale":"Visible archive markers, retained links, restoration, holds, role-gated disposition, and audit history are credible safeguards. The score is below maximum because permissions can change during archiving, optional auto-approval can cause deletion, and tacit dependencies may remain undetected.","source_ids":["S6","S7","S8"]},"scalability":{"score":3,"rationale":"Standards and enterprise products show that lifecycle controls can operate across large repositories. Scaling the venture-specific layer still depends on recurring classification, cross-tool integration, reviewer capacity, and maintenance of supersession and dependency links.","source_ids":["S3","S5","S6","S7"]}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"10K_TO_50K","scope":"Six-week non-production evaluation of no more than 100 closed experiments from one squad, including corpus inventory, duplicate and identity review, a read-only mirror, provisional lifecycle and supersession metadata, matched untreated comparison, reviewer tasks, analysis, records/privacy screening, and a written decision report. No quarantine or deletion occurs.","confidence":"MODERATE","assumptions":["Existing repository export or read access is available.","One product or evaluation lead, one analyst, and part-time artifact-owner and records/privacy review are sufficient.","No custom production integration, software procurement, customer-data disposition, or legal opinion is required.","The band is a resource-equivalent estimate; no direct source reported the cost of this exact study."],"source_ids":["S1","S4","S5","S6","S7"]},"initial_deployment_startup":{"band_2026_usd":"50K_TO_250K","scope":"One venture unit and its principal repository, including lifecycle taxonomy, authority and contest roles, retention mapping, hold checks, metadata configuration, limited connectors, audit design, training, accessibility and security review, restore testing, and evaluation instrumentation.","confidence":"LOW","assumptions":["Existing collaboration and records platforms can be configured rather than replaced.","Historical backfill is selective and limited to decision-relevant artifacts.","Legal, privacy, and records specialists participate part-time.","Tacit dependencies require manual owner review rather than a new enterprise knowledge graph.","The band includes labor, coordination, software configuration, compliance, training, and evaluation but is not vendor-quoted."],"source_ids":["S3","S4","S5","S6","S7","S8"]},"operational_launch":{"band_2026_usd":"250K_TO_1M","scope":"Launch across multiple venture teams and several repositories, including integrations, identity resolution, prioritized historical backfill, role and access mapping, governance, exception handling, legal/privacy/security review, training, support, audit and restore exercises, and comparative outcome evaluation.","confidence":"LOW","assumptions":["Launch is limited to one organization rather than a commercial multi-tenant product.","Several heterogeneous repositories require connector and workflow work.","Sensitive, contractual, financial, intellectual-property, and held records retain separate controls.","Existing enterprise software covers most storage, search, audit, and retention functions.","The band includes labor, data work, software, compliance, coordination, testing, and evaluation."],"source_ids":["S3","S5","S6","S7","S8"]},"annual_recurring":{"band_2026_usd":"50K_TO_250K","scope":"Annual operation within one organization, including stewardship, review queues, classification contests, retention and hold updates, dependency checks, restore drills, access and audit reviews, software administration, training, user support, and monitoring of retrieval and selection outcomes.","confidence":"LOW","assumptions":["Repository volume and policy complexity remain moderate.","Automation supports reminders and routine labels, while consequential classifications receive human review.","One part-time stewardship function plus distributed artifact owners and technical support is sufficient.","Major litigation, repository migration, enterprise-license expansion, and commercial product support are excluded.","The band includes recurring labor, software administration, compliance, coordination, and evaluation."],"source_ids":["S3","S4","S6","S7","S8"]}},"verified_pipeline_gates":{"externally_supported_problem":{"status":"YES","reason":"Adjacent primary research documents difficulty obtaining correct information from out-of-date, poorly categorized repositories, while official experimentation documentation implements lifecycle states and archiving for growing experiment collections. Venture-specific prevalence and decision impact remain blocking uncertainties, but the underlying problem is externally supported.","source_ids":["S1","S5"]},"externally_credible_adopter_or_authorizer":{"status":"YES","reason":"Official guidance assigns decision-record ownership and team review, while Microsoft and Atlassian implement distinct administrator, archive, and disposition-review permissions. A product lead plus records/privacy authority is therefore a credible role structure, though no venture-specific buyer demand was found.","source_ids":["S4","S6","S7"]},"distinct_testable_incremental_claim":{"status":"YES","reason":"Because status, supersession, archive, retention, and disposition are established separately, the incremental claim can be isolated as the effect of a linked venture-evidence authority layer on retrieval time and incorrect-current selection, with dependency and retrieval failures as safety outcomes.","source_ids":["S4","S5","S6","S7"]},"bounded_next_evidence_step":{"status":"YES","reason":"A six-week, at-most-100-experiment read-only mirror can compare provisional lifecycle and supersession metadata with unchanged artifacts without production deployment, quarantine, or deletion.","source_ids":["S4","S5","S7"]},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"For the next evidence step only, all changes occur in a non-production mirror and all dispositions are simulated. Official guidance supports preserving links and restorability, separating disposition permissions, honoring litigation or audit suspension, and withholding destruction without authority.","source_ids":["S6","S7","S8"]},"credible_cost_scope_and_range":{"status":"YES","reason":"The four bands explicitly cover labor, repository and metadata work, integrations, governance, compliance, coordination, software administration, training, restoration testing, and evaluation. Later bands remain low-confidence because no direct cost study or vendor quote covers this configuration.","source_ids":["S3","S4","S5","S6","S7","S8"]}},"next_evidence_step":"Run a six-week, non-production study using at most 100 closed experiments from one squad. After records/privacy screening, create a read-only mirror and match artifact clusters by age and topic; randomly assign half to receive provisional lifecycle states, explicit superseded-by links, and an authority-first result view, while half retain their existing metadata and search presentation. Have reviewers who were not the original authors complete identical current-evidence retrieval and interpretation tasks. Compare median retrieval time and incorrect-current selection rates, while recording missing dependencies, failed retrievals, and classification contests. Simulate every disposition and make no production changes, quarantine, or deletion. Falsify progression if reviewers are already reliably correct in the untreated corpus, the treatment fails to exceed a predeclared minimum meaningful improvement, or any safety outcome exceeds its predeclared tolerance.","blocking_evidence":["Venture-specific prevalence: the frequency of obsolete, duplicate, or ambiguously authoritative artifacts in representative experiment repositories is unknown.","Decision consequence: no study found here measures whether such artifacts change venture roadmap, funding, or go-to-market judgments.","Comparative effectiveness: lifecycle and supersession treatment has not been compared with unchanged search presentation or simpler status-and-archive practice.","Classification validity: the accuracy, contest rate, and maintenance burden of authority and supersession labels are unknown.","Dependency safety: tacit, political, contractual, or off-repository dependencies may not be discoverable from metadata.","Stakeholder pull: no venture-specific adoption request, budget, procurement evidence, or willingness-to-sustain governance was found.","Behavioral response: shadow copying, selective storytelling, and unjustified deference to weak but current evidence remain unmeasured."],"research_disposition":"PROBLEM_PREVALENCE_STUDY","world_novelty_boundary":"This was a bounded web search through 2026-08-02 covering primary research, official product documentation, a current records standard, government guidance, experiment management, content archiving, and decision-record practice. It found substantial close prior art for nearly every mechanism. It was not an exhaustive vendor, patent, standards, procurement, non-English, or private-deployment search and cannot establish world novelty, patentability, or freedom to operate. The only remaining potentially distinctive proposition is the venture-specific combined configuration and its falsifiable comparative outcome."}