{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp04_retrieval_first_paired20_20260802","cell_id":"computability_boundary_mapping__engineering_design","considered_ids":["H1","H5"],"selected_id":"H5","dispositions":[{"hypothesis_id":"H1","lane":"RESEARCH","reason":"The computability boundary is well established, while the residual pre-authorization governance gate is supported only as an untested workflow distinction; the critic found no direct evidence that it improves engineering funding or scoping decisions."},{"hypothesis_id":"H5","lane":"FINALIST","reason":"Despite substantial component collision, the critic identifies a material, falsifiable residual: a mechanically enforced final-gate invariant spanning simulator, laboratory, vendor, and human evaluators that preserves typed unresolved states and blocks unsupported definitive verdicts. Direct sources establish both the recurring failure problem and the narrower gaps in existing solver-, interface-, and documentation-level practices."}],"selection_reason":"H5 offers the strongest recognizable engineering problem and a concrete residual claim testable against deployed workflows. Its novelty boundary is narrow enough to evaluate as an end-to-end gate invariant, whereas H1 primarily repackages established computability and assurance mechanisms into a governance requirement without direct outcome support."}