{"cases":[{"alternative_route_blocks":[],"case_id":"E15A027__DIRECT","case_type":"DIRECT_POSITIVE","domain":"Container-terminal crane safety","intended_route_id":"B1","omitted_condition_id":null,"remedy_leakage_audit":"The scenario states the diagnostic conflict and its operational consequences without recommending any intervention or describing a corrective mechanism.","route_evidence":[{"condition_id":"b002_a04_ap_a1","intended_status":"SATISFIED","scenario_evidence":"Load pins, hydraulic pressure, motor current, and the shipping manifest each provide an imperfect estimate of the same container's suspended mass."},{"condition_id":"b002_a04_ap_a2","intended_status":"SATISFIED","scenario_evidence":"The sources have materially different error bounds, calibration histories, biases, and update times."}],"scenario_text":"At Berth 6, a container terminal must determine the actual suspended mass of a sealed export container before an aging quay crane carries it over a maintenance crew's exclusion zone. Four available readings disagree: the crane's load pins indicate 31.8 tonnes, hydraulic pressure implies 35.1 tonnes, motor current suggests 33.6 tonnes, and the shipping manifest lists 29.4 tonnes. None is exact. Recent certification tests put the load pins within about 0.4 tonnes, while the hydraulic estimate has a wider error band and a documented upward bias. Motor current is affected by drivetrain losses and is only calibrated to roughly 1.8 tonnes. The manifest was entered two days earlier, before a late cargo amendment, and past audits show occasional transcription errors. All four figures concern the same container at the same lift, but the operations report presents them together without showing how their very different evidential quality affects the stated mass. Supervisors cannot tell how defensible the reported figure is near the crane's safety threshold.","scenario_title":"Conflicting Mass Readings at Berth 6","vocabulary_separation_audit":"This case uses cargo-terminal operations, crane instrumentation, manifests, tonnes, and lifting thresholds; the transfer case instead uses archaeological chronology and laboratory dating evidence."},{"alternative_route_blocks":[],"case_id":"E15A027__TRANSFER","case_type":"TRANSFER_POSITIVE","domain":"Archaeological site chronology","intended_route_id":"B1","omitted_condition_id":null,"remedy_leakage_audit":"The scenario asks only whether the chronology is defensible and contains no recommendation, intervention, or implementation language.","route_evidence":[{"condition_id":"b002_a04_ap_a1","intended_status":"SATISFIED","scenario_evidence":"Radiocarbon, dendrochronological, luminescence, and coin evidence each imperfectly date the same occupation layer."},{"condition_id":"b002_a04_ap_a2","intended_status":"SATISFIED","scenario_evidence":"The dating evidence differs in interval width, completeness, sample disturbance, and susceptibility to residual material."}],"scenario_text":"An excavation team is preparing the chronology of a burned dwelling whose occupation date determines whether it predates a regional political transition. Evidence from the same sealed floor layer does not line up neatly. Radiocarbon analysis of a charred seed places the deposit between 820 and 790 BCE, while a timber ring sequence points to 806 BCE but lacks its outermost growth rings. Luminescence testing of floor sediment yields 835 to 775 BCE, and a worn coin from beneath the collapse was minted no earlier than 812 BCE. Each observation bears on when the dwelling was occupied, yet each is imperfect in a different way. The seed has a narrow laboratory interval but may have been stored before burning; the timber result is sharper but incomplete; the sediment date spans six decades because of measurement uncertainty; and the coin supplies only an earliest possible date and could have remained in circulation. The draft report cites all four dates but gives a single confident occupation year whose evidential basis reviewers cannot reconstruct.","scenario_title":"Dating the Burned Dwelling","vocabulary_separation_audit":"The transfer case replaces industrial sensing with excavation layers, calendar dates, laboratory assays, tree rings, sediment, and artifact chronology, making its domain and surface vocabulary substantially different from the direct case."},{"alternative_route_blocks":[{"blocked_by_condition_id":"b002_a04_ap_a4","condition_set_id":"B2","how_blocked":"The scenario explicitly states that matched validation found no performance changes across sediment type, handling environment, workload, instrument condition, or calibration range."},{"blocked_by_condition_id":"b002_a04_ap_a5","condition_set_id":"B3","how_blocked":"The scenario states that no reading is weaker, duplicated, stale, biased, or noisier than the others, so equal treatment would not privilege defective evidence."},{"blocked_by_condition_id":"b002_a04_ap_a6","condition_set_id":"B4","how_blocked":"The laboratories report their dates separately, and the scenario explicitly says no combined estimate or preexisting combination rule exists."}],"case_id":"E15A027__NEAR_MISS","case_type":"ONE_LITERAL_NEAR_MISS","domain":"Archaeological site chronology","intended_route_id":"B1","omitted_condition_id":"b002_a04_ap_a2","remedy_leakage_audit":"The scenario documents a chronology problem and controlled evidence properties without proposing how the dates should be reconciled.","route_evidence":[{"condition_id":"b002_a04_ap_a1","intended_status":"SATISFIED","scenario_evidence":"Four laboratory dates are imperfect observations of the same destruction event."},{"condition_id":"b002_a04_ap_a2","intended_status":"CONTRADICTED","scenario_evidence":"The scenario explicitly establishes matched uncertainty, accuracy, calibration, recency, bias, coverage, independence, and contextual suitability for all four readings."}],"scenario_text":"At a second excavation, four accredited laboratories date split portions of one homogeneous charcoal sample from a sealed destruction layer. Their reported calendar estimates are 742, 746, 739, and 744 BCE, all intended to establish the same fire event. The measurements remain imperfect because each has a documented error band, but the interlaboratory study was designed to remove differences in evidential quality. Every lab used the same protocol and calibration release during the same week; blind reference samples show identical error distributions, accuracy, interval coverage, and negligible bias. Randomized aliquots and separate equipment make the readings mutually independent to the same documented degree, and all four samples are equally suitable for this deposit. Matched validation also found no performance changes across sediment type, handling environment, workload, instrument condition, or calibration range. No reading is older, noisier, duplicated, or less applicable than another. The laboratories report their dates separately, with no combined estimate or preexisting rule for producing one. The excavation report nevertheless needs to explain what calendar date the four equally credible but nonidentical results support.","scenario_title":"Matched Laboratory Dates for a Destruction Layer","vocabulary_separation_audit":"This near miss deliberately preserves the transfer case's archaeological chronology, laboratory, and calendar-date vocabulary while changing only the selected-route distinction concerning unequal evidential quality."}],"experiment_id":"eoa_inverse_innovation_exp15_route_aware_retrieval40_20260814","sample_id":"E15A027","schema_version":1}