{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp06_four_proposal_generalization60_20260803","source_assessment_id":"computability_boundary_mapping__film_media_production:P2:v0","cell_id":"computability_boundary_mapping__film_media_production","proposal_index":2,"qualification":"EMPIRICAL_PARTNER_CANDIDATE","criteria":{"specific_differentiated_claim":{"status":"YES","reason":"The falsifiable remaining claim is that a film-production gate restricted to certificate-backed equivalence can prevent sampled agreement, timeout, and failed search from becoming universal EQUIVALENT labels while still approving useful transformations inside a formally enforced fragment. This is specifically differentiated from ordinary golden-frame or OpenImageIO comparison despite close compiler-verification prior art."},"credible_problem_signal":{"status":"YES","reason":"External evidence establishes consequential audiovisual QC, complex stateful and environment-dependent plug-in semantics, and widespread instance-level comparison tooling. It does not establish prevalence of the exact overclaim behavior, but it provides a credible signal strong enough for a bounded prevalence audit rather than vague problem fishing."},"identifiable_partner_or_adopter":{"status":"YES","reason":"A post-production facility operating render graphs is a concrete partner class. Its post-production supervisor or release-master custodian can authorize the study, while pipeline engineers and QC reviewers can supply de-identified decisions, run the comparator, and assess whether certified substitutions are operationally useful."},"partner_access_is_necessary":{"status":"YES","reason":"Public research has already resolved the adjacent-prior-art and theoretical-baseline questions. The remaining uncertainties require proprietary historical substitution records, a facility-specific execution model and representative graphs, workflow measurements, and confirmation of decision authority and adopter pull."},"safe_authorized_first_step":{"status":"YES","reason":"The proposed six- to eight-week study is non-production and uses de-identified historical records, synthetic seeded graph pairs, frozen semantics, and ordinary comparison tooling. It prohibits production graph replacement, preserves supervisor authority, and defines halt, quarantine, rollback, and hard false-approval safeguards."},"bounded_decisive_empirical_design":{"status":"YES","reason":"The study fixes 10 historical decisions, 20 seeded graph pairs, an equal-budget golden-frame/OpenImageIO comparator, measurable labels and costs, independent reduction review, and predetermined falsifiers for problem prevalence, theoretical validity, checker safety, stale-certificate handling, and fragment adoptability."},"no_material_negative_gate":{"status":"YES","reason":"No verified pipeline gate is materially NO: the incremental claim, bounded evidence step, safety and authority posture, and cost scoping pass. Problem support and adopter credibility are UNCERTAIN rather than contradicted, and those uncertainties are exactly what the partner study is designed to resolve."},"not_merely_more_research":{"status":"YES","reason":"The next step is an operationally specified partner audit and controlled comparative trial with fixed cases, metrics, authority boundaries, and stop conditions—not an instruction to conduct additional general research or web searching."}},"uncertainty_types":["PROBLEM_PREVALENCE","ADOPTER_PULL","INCREMENTAL_EFFECT","WORKFLOW_FIT","DATA_ACCESS","COST_SCOPE"],"partner_profile":"One post-production or VFX facility using executable render graphs and golden-frame or OpenImageIO-style comparison, with an identified post-production supervisor or release-master custodian, pipeline engineering support, QC participation, and authority to permit a non-production study.","required_access":"Written non-production pilot authorization; 10 de-identified historical graph-substitution decisions; a frozen OpenFX/MaterialX-like representation and execution model; representative but non-production substitution examples; ordinary comparison infrastructure under a matched render budget; staff time and runtime measurements; and controlled access for an independent reduction reviewer.","bounded_empirical_test":"Pre-register a six- to eight-week study. Audit 10 de-identified historical decisions for reusable exact-equivalence claims, timeout-as-verdict behavior, or sampled-frame overclaims. Independently review the reduction against the frozen facility model. Implement a minimal certificate checker and run the specified 20 seeded pairs, including valid certificates, seeded differences, malformed or out-of-fragment inputs, late differences, and semantic-version mismatches. Compare its results with the facility's ordinary golden-frame/OpenImageIO workflow using the same render budget, recording false-EQUIVALENT count, label accuracy, certificate invalidation, counterexample yield, useful approvals, runtime, and staff minutes.","success_condition":"The audit finds at least one relevant unrestricted-equivalence or incompleteness-overclaim behavior; independent review validates the model-specific reduction; the gate produces zero false EQUIVALENT labels, never promotes late bounded agreement to universal equivalence, invalidates every semantic-version mismatch, and correctly separates certified, bounded, different, and unknown outcomes; at least one representative useful substitution fits the certified fragment; and the facility accepts the stronger guarantee at measured workflow cost.","falsification_condition":"Do not advance if none of the 10 historical decisions exhibits the claimed overreach; the reduction fails under the frozen model; any seeded inequivalent pair is labeled EQUIVALENT; any late-difference pair receives universal approval; any semantic mismatch preserves an old certificate; no operationally useful representative substitution fits the fragment; or staff and runtime burden exceed the comparator without an accepted stronger guarantee.","rationale":"This belongs in the narrow empirical-partner lane because the differentiated claim and safe comparative design survive adjacent prior art, while the decisive unknowns are partner-dependent: whether the specific overclaim occurs, whether a useful share of real substitutions fits the restricted fragment, whether the gate improves decision labels against an existing comparator, and whether supervisors accept its workflow burden. These questions cannot be settled by additional public web research."}