{"schema_version":1,"assessment_id":"eoa_inverse_innovation_exp03_opportunity320_20260801","source_experiment_id":"eoa_inverse_innovation_exp03_full320_20260801","cell_id":"computability_boundary_mapping__art_aesthetics","archetype_slug":"computability_boundary_mapping","domain_slug":"art_aesthetics","title":"Scope-Aware Certification for Formal Properties of Generative Art","opportunity_summary":"Test whether generative-art platforms and commissioning institutions can reduce false universal compliance claims by restricting exact analysis to decidable or finitely bounded artwork classes and labeling all other results as bounded, incomplete, unknown, or out of scope. The proposal concerns machine-checkable visual constraints rather than beauty or artistic merit.","adopter_authorizer":"A generative-art platform or commissioning institution that controls its accepted artwork language, analyzer, and certificate labels; participating artists authorize submission, while curators retain exhibition judgment.","scores":{"meaningful_impact":{"score":3,"rationale":"The proposal could prevent later noncompliant outputs, false rejection of compliant works, and wasted attempts to build an impossible unrestricted decider. Impact is limited by the sealed packet's lack of evidence about how often institutions request or rely on universal formal-property certification."},"stakeholder_pull":{"score":2,"rationale":"The candidate names plausible users and frames an institution as seeking an analyzer, but supplies no observed requests, adoption commitments, incident history, or evidence that sampled results are currently interpreted as universal certificates."},"incremental_advantage":{"score":4,"rationale":"Relative to finite randomized testing and human review, decidable fragments, bounded exhaustive analysis, explicit UNKNOWN results, and guarantee provenance directly address overclaiming. The advantage would be smaller where only empirical coverage or single-image checking is required."},"distinctiveness_plausibility":{"score":2,"rationale":"Boundary routing and explicit guarantee labels form a coherent package, but prior art is explicitly unsearched and the packet provides no basis for distinguishing it from existing program-verification, creative-coding, or certification practices."},"technical_implementability":{"score":4,"rationale":"The proposed 20-work sandbox pilot, finite seed sets, 100-frame traces, and one decidable frame predicate are bounded and technically plausible. Valid treatment of unrestricted behavior, sound abstractions, and language-specific decidability boundaries would require substantially greater formal work."},"adoption_authority_feasibility":{"score":4,"rationale":"The operating institution can define accepted languages and labels, artists may choose whether to enter the restricted regime, and curators retain final judgment. Feasibility is reduced by the risk that institutional certificates acquire authority beyond their stated scope."},"evidence_readiness":{"score":4,"rationale":"The candidate specifies a consenting 20-artwork corpus, sandboxing, a bounded exhaustive comparator, four result labels, falsifiers, and rollback conditions. It does not yet specify the corpus, seed-space sizes, label-comprehension metric, or acceptance thresholds."},"safety_net_benefit":{"score":5,"rationale":"UNKNOWN and OUT_OF_SCOPE states, prohibition on treating timeouts as violations, sandboxed execution, retained human judgment, logged assumptions, certificate withdrawal, and rollback to UNKNOWN provide unusually explicit safeguards against overclaiming and unsafe execution."},"scalability":{"score":3,"rationale":"The labeling and boundary-management pattern could be reused across artworks, but each artwork language, renderer, predicate, computation model, and external capability may require renewed modeling and proof review. Restricting languages may also exclude important expressive works."}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"10K_TO_50K","scope":"A bounded offline study with 20 consenting artworks: formalize one predicate, inventory program bounds, sandbox execution, compare sampled testing with exhaustive finite-seed and 100-frame analysis, seed counterexamples, and test interpretation of result labels.","confidence":"MODERATE","assumptions":["Existing artwork runtimes can be sandboxed without purchasing specialized hardware.","The finite seed sets and 100-frame traces are small enough for ordinary compute resources.","One formal-methods-capable engineer and limited curator or artist participation are available.","The study issues no live or universal certificates."]},"initial_deployment_startup":{"band_2026_usd":"50K_TO_250K","scope":"Build a restricted certification prototype with artwork-language validation, bounded execution, result-label generation, assumption records, logs, certificate withdrawal, and platform integration in a non-production environment.","confidence":"LOW","assumptions":["Only one artwork language and a small predicate family are initially supported.","Existing identity, submission, rendering, and storage infrastructure can be reused.","External formal proof review and security testing are limited in scope.","No unrestricted total analyzer is promised."]},"operational_launch":{"band_2026_usd":"250K_TO_1M","scope":"Production hardening for one institution, including secure untrusted-code execution, policy and user-interface integration, proof and scope review, accessibility and comprehension evaluation, incident handling, staff training, and certificate lifecycle management.","confidence":"LOW","assumptions":["Launch is limited to one platform or commissioning institution.","The accepted fragment and supported predicates remain deliberately narrow.","Security and compliance requirements are comparable to a normal hosted creative-code runtime.","Artists can appeal or opt out without constructing a separate adjudication system."]},"annual_recurring":{"band_2026_usd":"50K_TO_250K","scope":"Maintain the analyzer and sandbox, revalidate boundaries after language or renderer changes, review new predicates, monitor certificate interpretation, handle incidents and appeals, and preserve audit records.","confidence":"LOW","assumptions":["Deployment remains limited to one institution and a modest supported fragment.","Major renderer or language changes are infrequent.","Compute volume is moderate and bounded analyses do not require specialized infrastructure.","Independent proof or security review is periodic rather than continuous."]}},"research_burden":"HIGH","earliest_credible_horizon":"3_TO_12_MONTHS","pipeline_gates":{"recognizable_externally_supportable_problem":{"status":"UNCERTAIN","reason":"The packet gives a coherent failure mechanism and observable state, but no external evidence that institutions currently seek universal certificates, misreport timeouts, or suffer consequential certification failures."},"identifiable_adopter_or_authorizer":{"status":"YES","reason":"The institution operating the generative-art platform or commissioning process can define the accepted language and labels; artist submission authority and curator exhibition authority are separately identified."},"distinct_testable_incremental_claim":{"status":"YES","reason":"The proposal can test whether bounded exhaustive analysis plus BOUNDED_PASS, COUNTEREXAMPLE, UNKNOWN, and OUT_OF_SCOPE labels produces fewer false universal conclusions than sampled pass/fail testing."},"bounded_next_evidence_step":{"status":"YES","reason":"The authorized sandbox study is limited to 20 consenting artworks, one predicate, finite seeds, and 100-frame traces, with sampled testing as the comparator and finite decidability plus label interpretation as falsifiers."},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"The first step is sandboxed and consent-based, excludes merit judgments and silent rejection, preserves artist and curator authority, and specifies certificate withdrawal and rollback if scope or soundness fails."},"implementation_cost_scope_and_range":{"status":"UNCERTAIN","reason":"The pilot scope supports a broad first-evidence band, but production cost depends on unspecified artwork-language complexity, seed-space size, runtime security, integration requirements, proof-review effort, and submission volume."}},"blocking_evidence":["Whether any target institution actually requests or relies on universal guarantees over future generative-art outputs rather than checks of completed images or finite runs.","Whether the accepted artwork language is already finite-state, syntactically terminating, or otherwise fully enumerable, which would falsify the proposed unrestricted-boundary problem.","Whether stakeholders can state stable machine-checkable output predicates without treating them as judgments of beauty or artistic merit.","Whether an independently reviewed impossibility reduction is valid for the exact artwork language, renderer, predicate, quantifiers, and computation model.","Whether the proposed labels measurably reduce universal misinterpretation relative to sampled pass/fail output.","Whether closely related certification architectures or guarantee-labeling systems already exist; prior art is unsearched."],"next_evidence_step":"Run a preregistered, non-live sandbox study on 20 consenting artworks. For one declared frame predicate, first inventory whether every accepted program has a finite effectively enumerable state space; then compare the baseline sampled pass/fail method with exhaustive analysis over declared finite seeds and 100-frame traces, including seeded late counterexamples. Blind reviewers should classify each result as bounded, universal, unknown, or violation. Stop or redirect if all accepted programs are already fully decidable, exhaustive analysis omits a declared case, or the new labels do not reduce false universal interpretations.","research_questions":["Do target platforms or commissioning institutions make reusable universal output guarantees, and what decisions depend on them?","Are their accepted artwork languages and parameter spaces actually unrestricted or unbounded under the requested quantifiers?","Which formal predicates are stable, machine-checkable, and operationally important without encoding artistic merit?","For the exact declared computation model, is the proposed impossibility reduction valid under independent review?","How often do sampled testing and timeout handling produce false universal conclusions or false violations on the test corpus?","Do BOUNDED_PASS, COUNTEREXAMPLE, UNKNOWN, and OUT_OF_SCOPE labels improve user comprehension and downstream use relative to pass/fail labels?","What existing systems provide comparable restricted analysis, sound-incomplete analysis, provenance, and governed unknown states?","How much expressiveness and artist participation would be lost under each decidable fragment?","What production sandbox, integration, proof-review, and maintenance resources would the adopting institution require?"],"recommendation":"VALIDATE_PROBLEM_FIRST","uncertainty_constraints":["The assessment is closed-book and cannot establish prior art, novelty, prevalence, market size, realized impact, or exact cost.","The stated institutional demand is part of a hypothetical candidate, not evidence of stakeholder pull or adoption intent.","Applicability depends on quantification over future or unbounded program behavior; completed finite images may be directly decidable.","The problem disappears or changes materially if every accepted artwork has an effectively enumerable finite state space.","Technical conclusions depend on the precise language, renderer, predicate, external capabilities, and computation model.","The communication benefit of the result labels is hypothesized but not measured.","Cost bands are resource-equivalent planning ranges based on the stated scope, not vendor quotes or observed deployments."],"closed_book_prior_art_boundary":"No external search or prior-art evidence was used. The packet explicitly marks prior art as UNSEARCHED and novelty evidence as weak, so this assessment makes no claim that the architecture, computability framing, restricted fragments, fallback analyses, or guarantee labels are new or uncommon."}