{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp06_four_proposal_generalization60_20260803","source_assessment_id":"predictive_residual_processing__environmental_climate:P4:v0","cell_id":"predictive_residual_processing__environmental_climate","proposal_index":4,"qualification":"EMPIRICAL_PARTNER_CANDIDATE","criteria":{"specific_differentiated_claim":{"status":"YES","reason":"The proposal retains a specific comparator-based claim: relative to the unchanged raw-threshold dashboard, an actuator-verified, command-conditioned residual channel must reduce routine review volume by at least 30% while meeting explicit detection, protected-routing, reconstruction, and fallback thresholds."},"credible_problem_signal":{"status":"YES","reason":"External evidence supports substantial release-driven downstream variation, continuous operational water-quality monitoring, incident-response duties, and hydropower nuisance-alarm burden, although the prevalence of release-period misattribution remains unmeasured."},"identifiable_partner_or_adopter":{"status":"YES","reason":"A reservoir water-quality organization such as a USACE district team is an identifiable adopter and can coordinate with reservoir operations, model owners, field technicians, and compliance-data custodians to authorize the evaluation and provide records."},"partner_access_is_necessary":{"status":"YES","reason":"The decisive uncertainties cannot be resolved through public research because the test requires site-specific synchronized command, measured-actuator, upstream, tributary, meteorological, raw-sensor, alarm, acknowledgement, and operator-workflow data."},"safe_authorized_first_step":{"status":"YES","reason":"The first step is a bounded retrospective replay authorized by the environmental-monitoring lead, with any subsequent scheduled-release observation limited to approved read-only shadow mode; the raw dashboard, mandatory records, operating authority, emergency procedures, and protected signals remain authoritative."},"bounded_decisive_empirical_design":{"status":"YES","reason":"The design freezes the model and scope, reserves an untouched holdout, compares against the unchanged raw-dashboard baseline, injects blinded coincident events and defined faults, measures operational burden and safety behavior, and specifies quantitative success and falsification thresholds."},"no_material_negative_gate":{"status":"YES","reason":"All substantive verified pipeline gates are YES. Cost scope is UNCERTAIN rather than materially NO, and cost uncertainty is not the sole reason for seeking a partner."},"not_merely_more_research":{"status":"YES","reason":"The next step is a preregistered partner-specific replay and conditional shadow observation with named inputs, comparator, injections, metrics, thresholds, and stopping conditions, not open-ended investigation."}},"uncertainty_types":["PROBLEM_PREVALENCE","ADOPTER_PULL","INCREMENTAL_EFFECT","WORKFLOW_FIT","DATA_ACCESS","COST_SCOPE"],"partner_profile":"A managed-reservoir water-quality team with authority to approve retrospective and read-only shadow evaluation, access to reservoir operations and environmental-monitoring records, an existing raw-threshold dashboard, and staff able to review residual alerts without changing gate control or emergency procedures.","required_access":"One controlled release's synchronized issued-command and measured-actuator records; upstream, tributary, meteorological, and downstream multivariable raw observations; model and station configurations; alarm and acknowledgement logs; operator review participation; and, only after replay success and local approval, read-only observation of one scheduled release.","bounded_empirical_test":"Preregister and replay one historical controlled release across a fixed station set and five variables using a frozen command-conditioned model and untouched holdout. Superpose blinded conductivity, turbidity, and dissolved-oxygen departures; inject actuator mismatch, sensor dropout, checksum conflict, travel-time shift, and persistent bias; compare residual routing with the unchanged fixed-threshold raw dashboard; audit random and transition windows; and condition one read-only shadow release on passing every replay threshold.","success_condition":"Routine notification or review burden falls by at least 30%; all protected signals route in full; at least 95% of scripted coincident departures are detected and acknowledged; every declared fault triggers full-signal fallback; held-out observations reconstruct within preregistered variable-specific tolerances; and total processing and review burden is lower than the raw-dashboard baseline.","falsification_condition":"Falsify the claim if burden reduction is below 30%, any protected signal is suppressed, scripted-event detection is below 95%, any injected fault fails to trigger fallback, command-actuator discrepancy or missingness is hidden, held-out reconstruction exceeds tolerance, false attribution or audit disagreement is unacceptable, or total burden is not lower than baseline.","rationale":"Adjacent prior art establishes reservoir-operation forecasting, water-quality modeling, continuous-sensor QC, and alarm management, but not the complete actuator-verified residual-attention architecture with reconstructibility, independent raw audit, protected bypasses, and deterministic fallback. The remaining incremental-effect and safety questions require proprietary synchronized site records and operator participation, making this a well-bounded empirical-partner candidate rather than a public-research question."}