{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp05_complete_proposal_portfolio20_20260803","cell_id":"computability_boundary_mapping__human_computer_interaction","portfolio_valid":true,"proposal_assessments":[{"proposal_index":1,"complete":true,"causally_faithful":true,"materially_distinct":true,"reason":"Operationally complete: it specifies the unrestricted safety-reachability class, checked impossibility evidence, enforceable finite fragment, exact and one-sided fallbacks, status semantics, authority, rollback, falsifiers, and a bounded evidence step. Its causal structure faithfully moves from an impossible universal protected-action verifier to model-relative guarantees. Its release-assurance problem and intervention are distinct from the other four."},{"proposal_index":2,"complete":true,"causally_faithful":true,"materially_distinct":true,"reason":"Operationally complete: it defines formal goal predicates, finite closed-world promises, an exact reachability procedure, fair witnessed search, external-capability contracts, status routing, authority limits, falsifiers, and fixtures. It faithfully distinguishes exact decidability, recognition, unknown, and relative computation. Its problem is user task feasibility and its intervention is an evidence-carrying help broker, not a verifier, migration gate, explanation service, or workflow runtime."},{"proposal_index":3,"complete":true,"causally_faithful":true,"materially_distinct":true,"reason":"Operationally complete: it specifies the observational-equivalence relation, Wizard-of-Oz provenance, checked unrestricted impossibility result, enforceable finite migration profile, exact pair comparison, divergence search, governance, rollback, and evidence fixtures. It faithfully prevents prototype success from implying a universal converter or equivalence decider. Its prototype-to-build migration decision and provenance gate are independently adoptable."},{"proposal_index":4,"complete":true,"causally_faithful":true,"materially_distinct":true,"reason":"Operationally complete: it separates trace, bounded-counterfactual, and global-influence queries; defines finite exact and unrestricted one-sided modes; exposes external dependencies; and supplies authority, rollback, falsifiers, and tests. It faithfully preserves quantifiers and substitutes honest weaker claims for an unavailable unrestricted global analyzer. Its explanation-scope intervention addresses a different affected problem and causal path."},{"proposal_index":5,"complete":true,"causally_faithful":true,"materially_distinct":true,"reason":"Operationally complete: it defines the workflow class, terminating fragment and promises, constructive guarantee, budgeted checkpointed execution, distinct lifecycle statuses, effect ledger, user controls, authority, rollback, falsifiers, and fixtures. It faithfully replaces an impossible universal termination predictor with scoped guarantees and honest suspension. Its completion-control and side-effect-management intervention is not a variant of the other proposals."}],"pairwise_assessments":[{"proposal_a":1,"proposal_b":2,"same_problem":false,"same_intervention":false,"independent_opportunity":true,"key_difference":"Proposal 1 prevents false release assurance about forbidden protected-action reachability; Proposal 2 prevents false runtime guidance about reaching a desired user goal. One installs a release verification boundary, while the other installs a help broker using plan witnesses and capability-relative results."},{"proposal_a":1,"proposal_b":3,"same_problem":false,"same_intervention":false,"independent_opportunity":true,"key_difference":"Proposal 1 analyzes safety of one extensible interface against a protected-action invariant; Proposal 3 analyzes task-semantic equivalence between legacy and replacement interfaces and the invalid inference from Wizard-of-Oz trials. The latter's provenance ledger and migration gate are not features of release verification."},{"proposal_a":1,"proposal_b":4,"same_problem":false,"same_intervention":false,"independent_opportunity":true,"key_difference":"Proposal 1 concerns whether forbidden behavior can occur and governs release assurance; Proposal 4 concerns what causal scope an adaptive-interface explanation can truthfully claim. Safety monitoring and explanation-query routing affect different decisions through different evidence objects."},{"proposal_a":1,"proposal_b":5,"same_problem":false,"same_intervention":false,"independent_opportunity":true,"key_difference":"Proposal 1 addresses semantic safety verification before interface release; Proposal 5 addresses termination status, checkpointing, resumption, and side effects during user-authored automation execution. A workflow may adopt either intervention without adopting the other."},{"proposal_a":2,"proposal_b":3,"same_problem":false,"same_intervention":false,"independent_opportunity":true,"key_difference":"Proposal 2 asks whether a goal is reachable from a current workspace and returns plan evidence; Proposal 3 asks whether two interfaces preserve the same declared behavior and whether a human-mediated prototype supports automation investment. Planning assistance and migration realizability have different causal targets."},{"proposal_a":2,"proposal_b":4,"same_problem":false,"same_intervention":false,"independent_opportunity":true,"key_difference":"Proposal 2 resolves task-feasibility claims using reachability witnesses and closed-world negative certificates; Proposal 4 resolves explanation-scope claims using execution provenance or feature-isolating context pairs. Neither intervention is an implementation detail of the other."},{"proposal_a":2,"proposal_b":5,"same_problem":false,"same_intervention":false,"independent_opportunity":true,"key_difference":"Proposal 2 searches for whether any action plan reaches a user's goal; Proposal 5 receives an already-authored workflow and governs whether its run is guaranteed complete, observed complete, suspended, or externally blocked. Their predicates, controls, and causal paths differ materially."},{"proposal_a":3,"proposal_b":4,"same_problem":false,"same_intervention":false,"independent_opportunity":true,"key_difference":"Proposal 3 compares two systems for task-semantic equivalence and exposes hidden human substitution in migration prototypes; Proposal 4 analyzes trace-local versus global feature-influence explanations within one adaptive system. Shared attention to provenance does not make the affected problem or intervention the same."},{"proposal_a":3,"proposal_b":5,"same_problem":false,"same_intervention":false,"independent_opportunity":true,"key_difference":"Proposal 3 governs a prototype-to-build decision for accessible interface conversion and equivalence certification; Proposal 5 governs execution of end-user automations through termination proofs, checkpoints, and effect ledgers. They have separate artifacts, owners, and consequences."},{"proposal_a":4,"proposal_b":5,"same_problem":false,"same_intervention":false,"independent_opportunity":true,"key_difference":"Proposal 4 prevents trace or bounded evidence from becoming an unsupported global influence explanation; Proposal 5 prevents timeout from becoming a false nontermination verdict and makes suspension safely resumable. Explanation contracts and completion protocols intervene through different mechanisms."}],"replacement_indices":[],"rationale":"All five proposals are operationally complete and preserve the archetype's essential causal sequence: formalize the claimed class and computation model, obtain constructive and impossibility evidence under matched assumptions, restrict exact guarantees to enforceable decidable regions, preserve unknown and relative states, and publish governed fallbacks. Although they reuse the archetype's standard boundary-mapping toolkit, they address five materially different HCI problems with different formal predicates and interventions: protected-action safety verification, contextual task reachability, migration equivalence and Wizard-of-Oz build inference, adaptive-explanation influence scope, and automation completion control. Every pair differs in affected problem, intervention, and causal path; none is merely a renamed variant, added feature, narrower population, or implementation detail."}