{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp09_archetype_breadth150_20260804","research_id":"eoa_inverse_innovation_exp09_light_prior_art_20260804","cell_id":"priority_based_admission__computer_science","search_lanes":{"direct_problem_and_intervention":{"queries":["fuzzer-generated regression tests runtime budget admission priority permanent suite archive excluded witnesses","fuzzing bug-triggering inputs select permanent regression suite severity novelty execution cost","\"fuzz\" \"regression suite\" \"runtime\" test case selection"],"source_ids":["SRC1","SRC2"],"no_result_note":null},"synonyms_and_historical_terms":{"queries":["fuzzing corpus culling distillation minimization execution latency coverage","safe regression test selection subset reduce cost","test suite reduction selection prioritization limited time"],"source_ids":["SRC2","SRC3"],"no_result_note":null},"products_practices_and_standards":{"queries":["ClusterFuzz reproducible testcase minimized crash fixed testing documentation","AFL corpus culling favored inputs latency file size afl-cmin documentation"],"source_ids":["SRC1","SRC2"],"no_result_note":null},"component_combination":{"queries":["regression test prioritization fault severity execution time redundancy","value based regression test prioritization severity cost nonduplicate faults","test suite selection budget knapsack fault severity execution cost"],"source_ids":["SRC3","SRC4"],"no_result_note":null}},"sources":[{"source_id":"SRC1","title":"Fixing a bug","publisher":"Google ClusterFuzz","url":"https://google.github.io/clusterfuzz/using-clusterfuzz/workflows/fixing-a-bug/","source_type":"OFFICIAL_GUIDANCE","claims_supported":["ClusterFuzz supplies downloadable crash test cases and reproduction information.","ClusterFuzz reruns test cases against current builds until fixes are detected.","Reliability of reproduction affects how ClusterFuzz treats crash test cases."]},{"source_id":"SRC2","title":"Technical Details","publisher":"AFL++ Project","url":"https://aflplus.plus/docs/technical_details/","source_type":"FIRST_PARTY_PRODUCT","claims_supported":["AFL selects a smaller favored corpus while preserving observed coverage tuples.","Its selection score incorporates execution latency and input size.","Nonfavored entries may be retained but skipped, while afl-cmin can permanently discard redundant entries.","AFL distinguishes crashes using trace novelty criteria."]},{"source_id":"SRC3","title":"Empirical Studies of a Safe Regression Test Selection Technique","publisher":"IEEE Transactions on Software Engineering / University of Nebraska–Lincoln","url":"https://digitalcommons.unl.edu/csearticles/11/","source_type":"PRIMARY_RESEARCH","claims_supported":["Regression-test selection is an established technique for reducing testing cost by selecting a subset of an existing suite.","Safe selection seeks to exclude no test that would reveal a relevant fault under defined conditions.","Empirical results show that selection benefits and costs depend on suite and coverage characteristics."]},{"source_id":"SRC4","title":"Value-Based Test Case Prioritization for Regression Testing Using Genetic Algorithms","publisher":"Tech Science Press","url":"https://www.techscience.com/cmc/v74n1/49871/html","source_type":"PRIMARY_RESEARCH","claims_supported":["Regression testing can be curtailed by limited time and cost.","The proposed severity-detection score combines fault severity, execution cost, and whether another test already detects the fault.","The method prioritizes execution order rather than removing tests or governing permanent-suite membership."]}],"problem_evidence":{"status":"PARTLY_SUPPORTED","finding":"The component problem is visible: official fuzzing systems preserve, minimize, deduplicate, and repeatedly execute crash inputs; AFL explicitly trades corpus coverage against input size and latency; and regression-testing research treats suite cost, subset selection, fault severity, redundancy, and limited execution resources as established concerns. The retained evidence does not directly demonstrate a repository where distinct reproducible fuzzer witnesses exceed a declared permanent-regression runtime budget after ordinary minimization, so that exact scarcity remains repository-specific.","source_ids":["SRC1","SRC2","SRC3","SRC4"]},"closest_prior_art":[{"name":"AFL favored-corpus culling and afl-cmin","source_ids":["SRC2"],"overlap":"Selects inputs using coverage novelty, execution latency, and size; suppresses or removes redundant inputs to control corpus cost.","remaining_difference":"It optimizes a fuzzing corpus rather than governed admission of legitimate defect witnesses to a permanent regression suite, and it lacks consequence-based reserved capacity, recorded exclusion reasons, appeal, and starvation-aware reconsideration."},{"name":"Safe regression-test selection","source_ids":["SRC3"],"overlap":"Selects a cost-reducing subset from an existing regression suite while attempting to preserve fault-revealing tests.","remaining_difference":"It is change-sensitive regression selection, not a durable admission boundary for accumulated fuzzer witnesses with consequence ranking, archival handling, reserved capacity, and review."},{"name":"Value-Cognizant Fault Detection-Based Test Case Prioritization","source_ids":["SRC4"],"overlap":"Ranks tests using fault severity per execution time and discounts faults already detected by another test, closely matching consequence, cost, and redundancy attributes.","remaining_difference":"It orders all tests for execution and expressly distinguishes prioritization from selection or removal; it does not establish permanent membership, reserved capacity, appeal, or a reasoned archive for nonadmitted witnesses."},{"name":"ClusterFuzz crash-testcase lifecycle","source_ids":["SRC1"],"overlap":"Maintains reproducible crash test cases, reruns them, tracks fixes, and treats unreliable witnesses differently.","remaining_difference":"The guidance does not document comparative admission to a fixed-runtime permanent suite or the proposal's multi-attribute governance and review mechanisms."}],"prior_art_disposition":"ADJACENT_PRIOR_ART","contrastive_claim_remaining":"Under a genuinely binding recurring runtime ceiling, governed permanent-suite admission that jointly considers defect consequence, sole-witness status, behavioral novelty, reproducibility, redundancy, and measured cost—and adds protected capacity for uncovered high-consequence classes plus reasoned archival, appeal, and reconsideration—will retain more independently judged high-consequence distinct failure classes than discovery-order or ad hoc membership. The retained sources cover several ingredients separately but do not disclose this complete governed-boundary combination.","contrastive_claim_falsifier":"The contrast fails if closer prior art already implements the same permanent-membership policy and governance combination, or if a preregistered shadow study finds that ordinary minimization removes the scarcity, the policy covers no more high-consequence distinct failure classes than baseline, frequently excludes their sole witnesses, yields materially inconsistent independent classifications, or consumes assessment effort comparable to the runtime it allocates.","gates":{"adequate_source_search":{"status":"PASS","rationale":"The bounded search covered the proposal directly, historical and synonymous regression-selection terminology, first-party fuzzing practices, and combinations of severity, redundancy, novelty, and execution cost. Exactly four opened sources from four publisher contexts were retained, including official guidance, first-party documentation, and primary research.","source_ids":["SRC1","SRC2","SRC3","SRC4"]},"supported_problem":{"status":"PASS","rationale":"The general resource, redundancy, reliability, and prioritization pressures are supported, although the exact post-minimization scarcity at a permanent-suite boundary is only partly supported and must be measured in the target repository.","source_ids":["SRC1","SRC2","SRC3","SRC4"]},"distinct_testable_claim":{"status":"PASS","rationale":"A measurable contrast remains between governed permanent membership and the adjacent practices of corpus culling, regression selection, and execution-order prioritization. Outcomes include distinct high-consequence class coverage, sole-witness exclusions, runtime, classification agreement, and administrative cost.","source_ids":["SRC2","SRC3","SRC4"]},"bounded_next_test":{"status":"PASS","rationale":"A read-only shadow classification of at most 60 historical witnesses under the existing runtime ceiling is bounded and can test actual scarcity, inter-rater agreement, admitted failure-class coverage, sole-witness exclusions, appeal effects, runtime, and assessment effort without modifying CI.","source_ids":["SRC1","SRC2","SRC3","SRC4"]},"no_obvious_safety_or_authority_stop":{"status":"PASS","rationale":"The proposed first step changes no suite membership, production code, or release gate and leaves final authority with designated maintainers. Fuzzer witnesses should be executed only in an isolated environment with resource limits and controlled access to potentially sensitive crash artifacts, but these are manageable controls rather than an obvious stop.","source_ids":["SRC1","SRC2"]}},"screen_survival":true,"world_novelty_boundary":"This four-source public-web screen supports only an adjacent-prior-art disposition and a researchable residual claim. It cannot establish world novelty, patentability, freedom to operate, market size, expert acceptance, or realized value; patents, proprietary CI systems, undocumented repository policies, standards, and additional literature may contain closer matches."}