{"actors":["Performance analyst who explores GPS, force-plate, wellness, and training-response data","Lead sport scientist who reviews evidence status","Strength and conditioning coaches who propose training-load changes","Team clinician who retains authority over acute medical restrictions","Athletes whose training exposure or selection could be affected"],"affected_objective":"Make athlete-specific training-load recommendations traceable to evidence whose credibility reflects the full search that produced it, without delaying established acute-safety responses.","arm":"COMMON_P1","authority_safety":{"authorized_first_step":"The lead sport scientist may run the gate in shadow mode for one squad over four weekly review cycles, recording analyses and provisional recommendations without changing any athlete's training, treatment, or selection.","decision_authority":"The lead sport scientist owns claim-status decisions; coaches retain ordinary training authority but may not treat a shadow finding as confirmed, and the team clinician retains final authority over medical restrictions and urgent safety action.","excluded_actions":["Changing an athlete's workload, rehabilitation, availability, or selection on the basis of a shadow-stage discovery","Overriding clinician judgment or delaying an established acute-safety protocol","Reusing confirmation data for exploratory model or threshold tuning","Expanding data access beyond existing role-based permissions","Publicly identifying athletes in the discovery record"],"halt_rollback":"Halt the shadow pilot if athlete identifiers escape authorized systems, routine safety decisions are delayed, or attempted looks cannot be captured reliably. Remove the gate from the live review workflow, retain an access-controlled audit record, and revert to the pre-pilot decision process; no athlete-facing decision requires reversal because the pilot authorizes none."},"baseline":"In the inferred baseline, staff inspect combinations of GPS exposures, jump metrics, wellness scores, positions, thresholds, model specifications, and time windows, then place a selected association in a weekly performance memo using an ordinary single-analysis threshold plus coaching plausibility. Null and abandoned looks are not attached to the recommendation, so reviewers cannot tell how many opportunities existed to find the reported pattern.","candidate_id":"multiple_testing_discipline__sport_science__COMMON_P1","causal_chain":["A weekly load review permits many combinations of athlete measures, outcomes, roster segments, thresholds, time windows, and model choices.","Each additional look creates another opportunity for chance variation to produce an attractive association.","When unselected and null looks are absent from the memo, the selected association is interpreted as though it were the sole planned analysis.","That interpretation can give an exploratory athlete or position-specific load signal an action-ready status.","The intervention defines the full weekly decision-linked claim family and records every attempted look before evaluating the selected signal.","A multiplicity-aware screening rule can label a surviving signal as an adjusted discovery, while all other searched patterns remain exploratory or rejected.","A frozen claim tested prospectively on subsequent, untouched sessions must pass its registered confirmation rule before supporting a non-urgent athlete-facing workload restriction.","The preserved registry lets reviewers reconstruct the opportunity set and prevents a failed claim from being revived through a new favorable slice without a new status review."],"cell_id":"multiple_testing_discipline__sport_science","consequence":"A chance-selected signal could prompt unnecessary workload caps or altered training for an athlete or position group, while consuming staff attention and obscuring whether the recommendation rests on discovery evidence or confirmation evidence.","diversity_from_prior_proposals":"Prior proposals were not inspected under runtime isolation. This candidate is differentiated internally by treating the weekly athlete-load recommendation as the decision-linked claim family and coupling cross-sensor search logging with an untouched prospective-session confirmation gate.","experiment_id":"eoa_inverse_innovation_exp13_second_slot_policy60_20260806","intervention":"Install a weekly Athlete Load-Response Claim Gate. Before dashboard review, the lead sport scientist defines the decision-linked family as every GPS, force-plate, wellness, performance, and availability comparison that could support the week's athlete- or position-specific workload recommendation. The analysis environment appends each viewed outcome, subgroup, threshold, window, exclusion, and model to an access-controlled registry. Broad screening uses a predeclared false-discovery-rate rule, but a surviving result receives only adjusted-discovery status. Before any non-urgent restriction based on that result, staff freeze its predictor, outcome, population, direction, model, threshold, and decision criterion, then test it once on subsequent sessions withheld from exploration. The registry records exploratory, adjusted-discovery, confirmed, replicated, or rejected status and preserves null and abandoned analyses.","mechanism_mapping":[{"counterfactual_removal":"Without the registry, analysts and reviewers could not reconstruct how many dashboard views and analytic choices generated the selected load-response claim.","mechanism_slug":"claim_registry","role":"Captures the complete family of attempted sensor, outcome, subgroup, window, threshold, and model looks with owners and statuses."},{"counterfactual_removal":"Without discovery-rate control, an ordinary single-analysis threshold would continue to ignore the number of screened opportunities.","mechanism_slug":"false_discovery_rate_control","role":"Calibrates which broadly screened signals may enter the adjusted-discovery set while preserving their non-confirmed status."},{"counterfactual_removal":"Without advance freezing, the predictor, outcome, model, or decision threshold could be revised after observing confirmation sessions.","mechanism_slug":"preregistration","role":"Locks the selected claim and its confirmation criterion before untouched sessions are examined."},{"counterfactual_removal":"Without untouched sessions, the same evidence used to find the association would also be used to certify it.","mechanism_slug":"holdout_validation","role":"Separates exploratory training sessions from subsequent sessions reserved for the frozen confirmation test."},{"counterfactual_removal":"Without a targeted follow-up gate, an adjusted discovery could flow directly into athlete-facing workload decisions.","mechanism_slug":"confirmatory_follow_up","role":"Requires a prospective, decision-specific test before a non-urgent training restriction can rely on the selected signal."}],"nearest_rivals":["A single-hypothesis testing protocol is insufficient when the reported hypothesis was selected from many outcomes, segments, windows, and models.","A reproducibility protocol could rerun the selected analysis but would not by itself restore the missing family of failed or unselected looks.","A multiverse report could reveal model dependence but would not alone enforce claim status or prospective confirmation before action.","Confounder control addresses distorted associations from third variables, whereas this candidate addresses false-positive opportunity created by repeated search.","Effect-size review addresses practical magnitude after estimation, whereas this candidate addresses whether a searched finding has earned confirmatory status."],"negative_tests":{"intervention_falsifier":"The gate is operationally falsified if, during the four-cycle shadow pilot, independent reviewers cannot reconstruct the attempted-look inventory from the registry, analysts can obtain decision-relevant results through unlogged paths, or supposedly untouched confirmation sessions have already influenced model or threshold selection.","problem_falsifier":"The inferred multiplicity problem is absent if the audit shows that each workload recommendation came from one prospectively specified outcome, population, window, and analysis path, with no additional inspected comparisons capable of supporting the same decision.","risks":["Analysts may route exploratory work through unlogged spreadsheets or informal conversations.","Teams may define claim families narrowly after seeing results, creating correction theater.","Repeated viewing of future sessions may contaminate the confirmation set.","A conservative screening rule may discard useful leads that could have been followed up cheaply.","Registry work may delay routine review or encourage checkbox compliance.","Athlete-level discovery records may expose sensitive performance or health information if access controls fail.","A statistically disciplined result may still be confounded, poorly measured, practically unimportant, or unsafe to act upon."],"strongest_counterevidence":"The strongest counterevidence would be a traceable history showing that workload recommendations were already based on prospectively frozen, decision-specific analyses and untouched prospective confirmation, with all alternative looks recorded and consistently labeled."},"next_evidence_step":"Run a bounded four-week shadow pilot for one squad. Before each weekly dashboard review, freeze the claim-family definition and confirmation-data cutoff; automatically log every eligible view and analysis; process any selected signal through the screening and status rules; and record the recommendation that would have occurred under the baseline and under the gate. At the end, have a reviewer using only the registry attempt to reconstruct every selected claim, its full opportunity set, status, and required follow-up. Collect evidence about capture completeness, family-definition stability, confirmation contamination, and whether any baseline action-ready claim remains merely exploratory; make no athlete-facing changes.","observable_state":"For each weekly review, dashboard and analysis logs show the number and type of GPS, force-plate, wellness, performance, and availability outcomes inspected; athlete and position filters; time windows; thresholds; exclusions; transformations; model specifications; null results; selected results; and timestamps. The claim registry displays a family identifier, frozen specifications, evidence partition, current status, reviewer, and any linked training recommendation.","prior_art_status":"UNSEARCHED","problem":"A sport-science unit searching weekly athlete-monitoring data may inspect many sensor metrics, outcomes, positions, athletes, time windows, thresholds, exclusions, and models, then elevate the most attractive load-response association as if it were a single planned test. When that selected association is used to justify a non-urgent workload restriction, its apparent credibility omits the search opportunities that produced it.","proposal_index":1,"remaining_contrastive_claim":"The candidate's irreducible claim is that a selected athlete-load signal cannot become decision-ready merely because its isolated analysis passes an ordinary threshold or appears coach-plausible; its status must reflect the entire decision-linked search family and advance only through a frozen test on evidence not used during discovery.","revision_record":{"claim_changes":[],"conceptual_changes":[],"evidence_changes":[],"operational_changes":[],"parent_version":null,"progress_targets_addressed":["Specified a concrete many-analysis sport-science decision problem","Preserved claim-family, attempted-look, multiplicity-rule, status, confirmation, and discovery-record causality","Bounded authority and protected acute clinical safety decisions","Defined problem and intervention falsifiers","Limited the first evidence step to a reversible shadow pilot"]},"schema_version":1,"structural_mapping":[{"archetype_element":"Define the claim family","domain_realization":"All metric, athlete or position, outcome, window, threshold, exclusion, and model combinations capable of supporting the same weekly workload decision receive one family identifier."},{"archetype_element":"Inventory attempted looks","domain_realization":"The analysis environment records every dashboard view and model run, including null, abandoned, and non-selected load-response analyses."},{"archetype_element":"Set the error-risk policy","domain_realization":"Broad monitoring tolerates screened leads, but a non-urgent athlete-facing restriction requires prospective confirmation; established acute-safety rules remain under clinical authority."},{"archetype_element":"Apply a multiplicity-aware rule","domain_realization":"A predeclared false-discovery-rate procedure governs the weekly screened set, followed by a single frozen confirmation test on subsequent untouched sessions."},{"archetype_element":"Label claim status","domain_realization":"Each load-response claim is marked exploratory, adjusted-discovery, confirmed, replicated, or rejected in the weekly registry and memo."},{"archetype_element":"Confirm before costly action","domain_realization":"A selected association cannot support a non-urgent workload restriction until its fixed specification passes the registered prospective test."},{"archetype_element":"Preserve the discovery record","domain_realization":"Access-controlled records retain tested alternatives, null findings, analytic choices, evidence cutoffs, status changes, reviewers, and linked recommendations."}],"title":"Athlete Load-Response Claim Gate","version":0}