{"actors":["Research participants whose consented motion data are collected","Biomechanics laboratory operators who calibrate equipment, place sensors, and run protocols","Instrument and synchronization technicians who manage capture systems","Biomechanics analysts who clean signals, identify phases, and derive variables","Study investigators who approve protocol interpretations and trial eligibility rules","Data stewards and ethics personnel who govern identity, access, retention, and correction"],"affected_objective":"Reproduce which captured trials, protocol deviations, calibration states, synchronization choices, exclusions, corrections, and processing versions produced a named biomechanics analysis dataset, without treating the reconstructed dataset as a clinical, selection, or coaching decision.","arm":"ORDINARY_DIVERSE_P2","authority_safety":{"authorized_first_step":"Under the study's existing ethics and data-governance approvals, construct a read-only prototype from a copied, de-identified bundle for one completed biomechanics protocol. It may reconstruct trial manifests and compare derived metadata but cannot alter source recordings, approved datasets, publications, or participant records.","decision_authority":"The principal investigator retains authority over study interpretations and analysis-plan changes; designated analysts retain responsibility for documented processing decisions; ethics and data-governance personnel control consent, access, retention, and redaction; the prototype has no authority over participant care, athlete selection, or training.","excluded_actions":["Changing trial eligibility, endpoint definitions, or protocol interpretations without the study's governed approval process","Writing corrections directly into raw recordings, approved analysis datasets, publications, or participant records","Using temporal sequence alone to infer biomechanical causation, participant intent, fatigue, injury, or misconduct","Re-identifying participants or exposing video, body measurements, health information, or restricted metadata outside approved roles","Treating computational tie-breaks, automated phase detections, or confidence scores as adjudicated scientific facts","Using reconstructed outputs for clinical, selection, disciplinary, or live coaching decisions"],"halt_rollback":"Stop if the prototype links an event to the wrong participant or trial, exposes restricted media or identifiers, lacks provenance for a source transformation, or fails deterministic replay under the frozen contract. Revoke prototype access, delete copied data and derived projections through the approved procedure, preserve only the authorized incident record, and leave original research assets unchanged."},"baseline":"A biomechanics study stores raw recordings, trial spreadsheets, calibration files, exclusion notes, processing scripts, and exported variable tables as separate mutable artifacts. The latest trial-status column or processed dataset may show what is currently included, but it does not necessarily preserve the ordered protocol, calibration, synchronization, adjudication, and transformation history needed to reproduce how that analysis set was assembled.","candidate_id":"event_log_centered_modeling__sport_science__ORDINARY_DIVERSE_P2","causal_chain":["A laboratory session generates distinct protocol acts and assertions: equipment calibration, sensor placement, capture initiation, trial attempt, protocol deviation, synchronization, quality review, exclusion decision, correction, and processing release.","When trial spreadsheets and processed files are updated in place, the current inclusion flag or derived variable table collapses the original occurrence, later observation, analyst decision, source provenance, and superseded interpretation.","Recalibration, revised event-phase detection, corrected trial identity, or a changed exclusion adjudication can then alter the analysis cohort and outputs without a reconstructable lineage to the affected recordings.","The intervention admits bounded laboratory and analysis assertions to a canonical log with stable identities, participant and instrument roles, event and record times, referenced source artifacts, transformation deltas, validity states, and linked corrections.","A pinned projection contract selects eligible trials and processing assertions, resolves their declared partial order, and produces a versioned trial manifest and analysis-dataset lineage while leaving continuous recordings in governed source storage.","Frontiers, unresolved conflicts, protocol coverage, and rebuild diffs expose whether the manifest omits an adjudication, uses an obsolete calibration, or differs from the canonical history.","Investigators can inspect and govern the evidentiary path that produced a dataset without silently rewriting earlier study states or granting the projection independent authority."],"cell_id":"event_log_centered_modeling__sport_science","consequence":"A reported biomechanics dataset may be difficult to trace to the exact recordings, calibration state, protocol execution, trial-eligibility decisions, corrections, and processing versions that produced it. Reanalysis can therefore select a different trial set or apply incompatible transformations while still appearing to use the same study data.","diversity_from_prior_proposals":"Unlike the sealed P1's athlete-level training, recovery, and daily planning problem, this opportunity concerns laboratory experiment reproducibility. Its canonical events describe protocol execution and analytical transformations; its derived objects are trial manifests and dataset lineages; and its causal path runs from mutable research artifacts through untraceable cohort construction to replayable analysis provenance, not from asynchronous monitoring inputs to athlete-day decision views.","experiment_id":"eoa_inverse_innovation_exp13_second_slot_policy60_20260806","intervention":"Create a governed biomechanics provenance ledger whose canonical records represent bounded laboratory and analytical assertions such as calibration performed, sensor configuration established, trial attempted, recording committed, protocol deviation reported, synchronization accepted, phase boundary inferred, trial-quality judgment issued, eligibility adjudicated, processing transformation executed, dataset release declared, and prior assertion corrected or retracted. Each record carries a stable event or assertion identifier, pseudonymous participant and instrument references with explicit roles, event time, record time, laboratory context, referenced source-artifact identifier and integrity digest, transformation delta, provenance, uncertainty, validity status, and access class. Continuous force, motion, electromyography, and video streams remain governed source artifacts rather than being forced into semantic micro-events. Duplicate imports are idempotent; uncertain trial links and competing phase annotations remain unresolved assertions; corrections append supersession or adjudication links. Versioned projectors derive a trial-eligibility manifest, calibration-and-processing lineage, and reproducible dataset release view with declared ordering, schema, reference files, frontier, conflict handling, and completeness limits. Rebuild-and-diff checks compare these projections with the preserved release manifest without directly patching either source data or canonical history.","mechanism_mapping":[{"counterfactual_removal":"Without immutable acceptance and linked corrections, revised trial decisions or processing assertions can replace earlier study states without preserving what changed.","mechanism_slug":"append_only_event_store","role":"Makes bounded protocol, adjudication, and processing assertions the governed historical substrate for dataset construction."},{"counterfactual_removal":"Without separate event and record times, a retrospective exclusion or synchronization correction can appear to have been part of the original laboratory session or an earlier analysis release.","mechanism_slug":"bitemporal_event_register","role":"Separates when a capture or protocol act occurred from when it was observed, entered, reviewed, corrected, or adjudicated."},{"counterfactual_removal":"Without registered event schemas and upcasting rules, older calibration, trial, and processing assertions may acquire new meanings during replay.","mechanism_slug":"versioned_event_schema_registry","role":"Pins field semantics and compatibility rules so historical assertions remain interpretable across analysis-pipeline changes."},{"counterfactual_removal":"Without a declared projector, the eligible-trial list and dataset lineage remain mutable files rather than reproducible functions of accepted assertions.","mechanism_slug":"event_sourced_projection","role":"Derives versioned trial manifests and release lineages from eligible canonical events under explicit rules."},{"counterfactual_removal":"Without rebuild comparison, undocumented spreadsheet edits, stale calibration links, or non-deterministic transformations may remain concealed in an analysis release.","mechanism_slug":"projection_rebuild_and_diff","role":"Reconstructs manifests and lineage metadata from the ledger and compares them with preserved release artifacts and declared invariants."},{"counterfactual_removal":"Without evidence-sensitive reconciliation, conflicting trial identifiers, phase annotations, or quality judgments may be prematurely collapsed into one apparently certain account.","mechanism_slug":"provenance_weighted_event_reconciliation","role":"Preserves competing assertions and their sources until an authorized adjudication policy determines projection treatment."}],"nearest_rivals":["A reproducible analysis repository containing scripts, environments, and frozen data: it can rerun computation, but may not represent protocol execution, trial identity, later adjudications, and corrections as canonical role-bearing events.","An electronic laboratory notebook: it preserves chronological notes, but free-form entries may lack stable event identities, explicit transformations, idempotent ingestion, ordering semantics, and rebuildable projection contracts.","A data-provenance graph linking files and processing steps: it can trace artifact dependencies, but may omit participant roles, protocol deviations, dual time, disputed assertions, and decision-time trial eligibility.","A version-controlled trial spreadsheet: it retains file revisions, but whole-file versions rather than bounded laboratory and analytical events remain primary, and derived manifests may still require manual interpretation."],"negative_tests":{"intervention_falsifier":"The intervention is falsified for this use if a frozen projector cannot reproduce the reference trial manifest and declared lineage invariants solely from admitted events plus integrity-checked source references, or if the required scientific meaning resides in continuous signals that cannot be represented through bounded protocol and transformation events without material distortion.","problem_falsifier":"The problem is falsified if the existing research package already provides immutable identities for trials and transformations, dual-time provenance, append-only correction lineage, explicit adjudication history, versioned projection rules, visible completeness limits, and deterministic reconstruction of sampled release manifests without manual interpretation.","risks":["Event definitions may falsely discretize continuous movement or hide uncertainty in phase boundaries.","Detailed video, body measurements, or protocol metadata may enable re-identification or exceed participant consent.","Incorrect trial or participant identity links may contaminate derived manifests.","A formal ledger may give automated detections or investigator judgments unwarranted epistemic authority.","Incomplete logging of informal laboratory decisions may make the recorded process appear exhaustive.","Schema changes may reinterpret older protocol deviations or quality labels during replay.","Long-lived correction histories may conflict with withdrawal, deletion, or redaction obligations.","Replayable computation may still reproduce a scientifically inappropriate analysis contract.","Operational burden may encourage staff to bypass event capture and maintain shadow spreadsheets."],"strongest_counterevidence":"A frozen raw-data archive, release manifest, analysis script, environment specification, and concise laboratory notebook may already provide sufficient reconstruction. If protocol acts and analytical decisions are few, unambiguous, and never revised, an event-first substrate could add governance and privacy costs without improving reproducibility."},"next_evidence_step":"Using one completed, de-identified biomechanics session bundle under its existing approvals, freeze event definitions for calibration, configuration, trial attempt, recording commitment, deviation, eligibility judgment, processing execution, and correction. Freeze identity keys, ordering rules, one trial-manifest projector, allowed source references, and exact lineage invariants before replay. Reconstruct the preserved release manifest, then record unexplained trial-membership differences, missing provenance links, unresolved identity conflicts, obsolete calibration references, and replay nondeterminism. Inspect no unrestricted video or direct identifiers, make no new biomechanical or participant-level inference, and stop at the predefined bundle boundary.","observable_state":"For a released biomechanics dataset, an investigator can inspect the current trial list and processing outputs but cannot consistently enumerate, in machine-replayable form, which protocol attempts, calibration states, source recordings, deviations, eligibility adjudications, superseded annotations, and processing executions produced each included result.","prior_art_status":"UNSEARCHED","problem":"Biomechanics studies transform laboratory captures into analysis datasets through calibration, sensor configuration, trial execution, synchronization, phase identification, quality review, eligibility adjudication, signal processing, and correction. When these steps are represented mainly by mutable trial spreadsheets, filenames, notebook entries, and exported tables, the current dataset can lose the lineage of why a trial was included, which calibration and transformation applied, what was known at release time, and which later correction changed the interpretation. This impedes reproducible reconstruction of the analysis cohort and its processing path.","proposal_index":2,"remaining_contrastive_claim":"For reconstruction of a biomechanics analysis release, the defining claim is that bounded protocol, adjudication, and processing assertions become canonical while trial manifests and dataset lineages remain discardable, versioned projections with visible frontiers and conflicts. Versioned files or rerunnable scripts alone do not satisfy the claim unless the governed event history can reproduce which evidence and decisions generated the release.","revision_record":{"claim_changes":[],"conceptual_changes":[],"evidence_changes":[],"operational_changes":[],"parent_version":null,"progress_targets_addressed":["Independent sport-science problem centered on biomechanics research reproducibility","Event boundaries distinguish protocol acts, observations, inferences, adjudications, transformations, and corrections","Continuous movement signals remain source artifacts rather than artificial semantic micro-events","Canonical history, projection contracts, replay, frontiers, and conflict handling preserve the event-first structure","Authority, consent, access, falsification, and rollback boundaries constrain the initial test"]},"schema_version":1,"structural_mapping":[{"archetype_element":"Event boundary and type contract","domain_realization":"Defines calibration, configuration, trial attempt, recording commitment, protocol deviation, synchronization decision, inferred phase annotation, eligibility adjudication, processing execution, release, and correction as distinct assertions; continuous samples are not individually promoted to semantic events."},{"archetype_element":"Canonical event record","domain_realization":"Each accepted laboratory or analytical assertion records identity, type, participant and instrument roles, dual time, laboratory context, source reference, transformation or evidentiary delta, provenance, uncertainty, status, and access class."},{"archetype_element":"Stable event and participant identity","domain_realization":"Source-scoped identifiers distinguish sessions, trial attempts, recordings, annotations, and processing runs; pseudonymous participant references persist while uncertain trial-recording matches remain unresolved."},{"archetype_element":"Event time and record time","domain_realization":"The protocol or capture time is separate from later entry, review, annotation, correction, adjudication, and dataset-release time."},{"archetype_element":"Transformation delta and provenance","domain_realization":"A processing event identifies its input artifacts, integrity digests, pinned code and parameter references, declared output, operator or system, and validity status without embedding mutable outputs as historical truth."},{"archetype_element":"Ordering and concurrency semantics","domain_realization":"Protocol order, device-local sequences, analytical dependencies, uncertain simultaneity, and computational tie-breaks are represented separately; dependency or temporal precedence is not labeled causal evidence."},{"archetype_element":"Correction and conflict lineage","domain_realization":"Revised trial identities, exclusions, calibrations, and phase annotations append retraction, supersession, or adjudication links while preserving earlier release-time knowledge states and competing assertions."},{"archetype_element":"Versioned projection definition","domain_realization":"A named projector declares eligible event types, identity and ordering rules, evidence-status filters, calibration selection, conflict handling, schema versions, source references, and output contract for a trial manifest or dataset lineage."},{"archetype_element":"Projection frontier and completeness statement","domain_realization":"Each projected release identifies the latest processed assertion, pending adjudications, missing laboratory records, inaccessible artifacts, unmodeled informal decisions, and protocol stages outside coverage."},{"archetype_element":"Replay and projection reconciliation","domain_realization":"A pinned projector rebuilds the release manifest and lineage from accepted assertions and compares identifiers, artifact digests, transformation dependencies, and declared invariants with the preserved release package."},{"archetype_element":"Retention, redaction, and access policy","domain_realization":"Consent scope, pseudonymization, role-specific access, restricted media references, retention horizons, withdrawal handling, and tested redaction propagation govern the ledger, projections, indexes, snapshots, and exports."}],"title":"Replayable Biomechanics Trial Provenance Ledger","version":0}