{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp09_archetype_breadth150_20260804","research_id":"eoa_inverse_innovation_exp09_light_prior_art_20260804","cell_id":"search_space_pruning__computer_science","search_lanes":{"direct_problem_and_intervention":{"queries":["production failure replay configuration feature flags versions combinatorial search incident debugging","production incident replay feature flags configuration snapshot tool","auditable pruning configuration space failure reproduction software debugging exclusions reentry"],"source_ids":["SRC1","SRC3"],"no_result_note":null},"synonyms_and_historical_terms":{"queries":["software configuration fault localization failure reproduction configuration space search pruning","delta debugging configurations failure-inducing configuration Zeller paper","configuration debugging search constraints test budget software failure inducing configurations"],"source_ids":["SRC3","SRC4"],"no_result_note":null},"products_practices_and_standards":{"queries":["NIST combinatorial testing configuration interactions software failures","feature interaction fault localization configuration sampling highly configurable systems paper","production incident replay feature flags configuration snapshot tool"],"source_ids":["SRC1","SRC2","SRC3"],"no_result_note":null},"component_combination":{"queries":["configuration sampling constraints configurable systems audit excluded configurations false negatives","adaptive combinatorial testing failure-inducing configuration localization budget","feature-interaction aware configuration prioritization configurable code"],"source_ids":["SRC2","SRC3","SRC4"],"no_result_note":"No retained source described the full combination of incident-specific hard-fact pruning, dominance removal, diversity preservation, sampled exclusion audits, recorded reentry, and per-exclusion accountability."}},"sources":[{"source_id":"SRC1","title":"How to Fight Production Incidents? An Empirical Study on a Large-scale Cloud Service","publisher":"Microsoft Research","url":"https://www.microsoft.com/en-us/research/publication/how-to-fight-production-incidents-an-empirical-study-on-a-large-scale-cloud-service/","source_type":"PRIMARY_RESEARCH","claims_supported":["High-severity production incidents impose substantial customer and engineering costs.","An empirical study of 152 resolved high-severity Microsoft Teams incidents found gaps and delays across detection, root-cause analysis, and mitigation.","Production incident records and postmortems contain telemetry, responder discussions, root causes, mitigation steps, and delay explanations that can supply evidence for archived-incident analysis."]},{"source_id":"SRC2","title":"Combinatorial Testing","publisher":"National Institute of Standards and Technology","url":"https://www.nist.gov/publications/combinatorial-testing","source_type":"OFFICIAL_GUIDANCE","claims_supported":["Software failures are frequently associated with interactions among relatively few parameters.","Combinatorial testing reduces the number of configurations exercised while retaining systematic interaction coverage.","Covering-array and interaction-testing methods are established alternatives to exhaustive configuration testing."]},{"source_id":"SRC3","title":"Feature-Interaction Aware Configuration Prioritization for Configurable Code","publisher":"arXiv","url":"https://arxiv.org/abs/1911.07779","source_type":"PRIMARY_RESEARCH","claims_supported":["Exhaustive analysis of configurable software is often infeasible because the number of configurations grows exponentially.","Existing methods eliminate configurations that violate feature-model constraints and sample configurations for interaction or code coverage.","CoPro ranks configurations using potential feature-interaction violations and placed buggy configurations in the top three in 78 percent of evaluated cases.","CoPro is close prior art for allocating limited analysis effort across configurations, but it prioritizes suspected bugs rather than making incident-evidence-based exclusions with audits and reentry."]},{"source_id":"SRC4","title":"Yesterday, my program worked. Today, it does not. Why?","publisher":"Karlsruhe Institute of Technology","url":"https://pp.ipd.kit.edu/publication.php?id=zeller99esec&lang=en","source_type":"PRIMARY_RESEARCH","claims_supported":["Delta debugging systematically searches changes to identify a minimal failure-inducing set.","The reported prototype reduced a 178,000-line change set to a failure-inducing change within hours.","Delta debugging is established prior art for search-space reduction after working and failing configurations can be tested, but it does not provide the proposed pre-replay exclusion governance."]}],"problem_evidence":{"status":"PARTLY_SUPPORTED","finding":"The sources make the underlying problem visible: production incidents can be costly and slow to diagnose, configuration interactions can cause failures, and exhaustive analysis of configuration spaces is infeasible. They do not directly establish the proposal's narrower operational assertion that incident responders routinely have a replay queue larger than a fixed execution budget or cannot reconstruct undocumented branch exclusions.","source_ids":["SRC1","SRC2","SRC3"]},"closest_prior_art":[{"name":"CoPro feature-interaction-aware configuration prioritization","source_ids":["SRC3"],"overlap":"Narrows an exponential configuration-analysis problem, respects valid configurations, scores configurations using implicated program entities, and prioritizes executions likely to expose interaction bugs.","remaining_difference":"It is general configurable-system quality assurance based on static suspiciousness ranking. It does not convert authenticated observations from a particular production incident into hard exclusions, test sampled exclusions, preserve reentry, or require an accountable exclusion record."},{"name":"Combinatorial interaction testing and covering arrays","source_ids":["SRC2","SRC3"],"overlap":"Selects a tractable, diverse sample from a combinatorial parameter space and preserves specified interaction coverage instead of testing every configuration.","remaining_difference":"Its selection objective is general t-way coverage, not reproduction of one captured incident signature using incident-specific runtime evidence. The retained sources do not add evidence-cited exclusions, dominance under replay cost, false-negative audits, or reversible reentry."},{"name":"Delta debugging","source_ids":["SRC4"],"overlap":"Uses repeated tests to reduce a large set of circumstances or changes to a smaller failure-inducing configuration.","remaining_difference":"It ordinarily starts with testable working and failing cases and seeks a minimal cause. The proposal instead allocates scarce attempts before a compact reproducer exists and governs uncertain exclusions through diversity, auditing, provenance, and reentry."}],"prior_art_disposition":"ADJACENT_PRIOR_ART","contrastive_claim_remaining":"For replaying one archived production incident under a fixed execution budget, an advisory procedure that permits hard pruning only from authenticated incident facts, removes replay-dominated configurations, preserves representatives of unresolved families, reserves executions for random exclusion audits, and records reversible exclusion decisions can reduce the active queue while keeping the known reproducing configuration reachable and every exclusion reconstructable. The retained prior art separately covers configuration prioritization, interaction-preserving sampling, constraint filtering, and failure-inducing-set reduction, but not this combined incident-replay governance claim.","contrastive_claim_falsifier":"The contrastive claim would be falsified by an earlier system or documented practice implementing this same combination for bounded production-failure replay, or experimentally if the archived reproducer is excluded under correct evidence, exclusions cannot be reconstructed and reversed, audit samples fail to reveal erroneous pruning, or procedure overhead consumes at least as much constrained replay capacity as executing the unpruned candidates.","gates":{"adequate_source_search":{"status":"PASS","rationale":"Four search lanes covered direct language, historical terminology, established practices, and component combinations. Exactly four opened sources from Microsoft Research, NIST, arXiv, and KIT were retained, including official and primary research. The screen found several adjacent approaches without treating an exact-phrase miss as novelty evidence.","source_ids":["SRC1","SRC2","SRC3","SRC4"]},"supported_problem":{"status":"PASS","rationale":"The problem is partly supported: primary and official sources show consequential production incidents, configuration-dependent interaction failures, exponential configuration spaces, and the infeasibility of exhaustive analysis. The specific undocumented replay-exclusion workflow remains unverified.","source_ids":["SRC1","SRC2","SRC3"]},"distinct_testable_claim":{"status":"PASS","rationale":"The remaining claim is operationally distinct from ranking, covering-array sampling, and minimal-cause reduction, and it has measurable outcomes: reproducer reachability, survivor count, exclusion reconstructability, audit false negatives, and replay-budget use.","source_ids":["SRC2","SRC3","SRC4"]},"bounded_next_test":{"status":"PASS","rationale":"The proposed archived-incident shadow test is bounded to 40 frozen configurations and 12 isolated replays, reserves two exclusion audits, blinds the operator to the known reproducer, and specifies concrete success and failure observations without requiring production changes.","source_ids":[]},"no_obvious_safety_or_authority_stop":{"status":"PASS","rationale":"The first test is isolated, archived, advisory, and reversible. It excludes live configuration changes, evidence deletion, automated root-cause declarations, and heuristic hard exclusions; it also defines ownership and halt conditions. Sensitive configuration metadata requires normal access control and secret redaction but is not an obvious stop for the bounded test.","source_ids":["SRC1"]}},"screen_survival":true,"world_novelty_boundary":"This bounded public-web screen supports only an adjacent-prior-art disposition and a testable remaining contrast. It cannot establish world novelty, patentability, market size, expert acceptance, realized value, or absence of unpublished, proprietary, non-indexed, patent, or differently termed implementations."}