{"arms":{"contracted_react":{"eligible_n":12,"metrics":{"completion_tokens":{"max":206,"mean":179.83333333333334,"median":197.5,"min":134,"n":12},"controller_rejection_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":12},"critical_path_rounds":{"max":5,"mean":4.666666666666667,"median":5.0,"min":4,"n":12},"elapsed_seconds":{"max":7.141787500004284,"mean":5.4793617750013555,"median":5.4606821500347,"min":3.9346082999836653,"n":12},"false_evidence_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":12},"merge_conflicts":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":12},"merge_omissions":{"max":3,"mean":1.0,"median":0.0,"min":0,"n":12},"model_calls":{"max":6,"mean":5.666666666666667,"median":6.0,"min":5,"n":12},"path_switch":{"max":true,"mean":0.3333333333333333,"median":0.0,"min":false,"n":12},"prompt_tokens":{"max":4120,"mean":3528.8333333333335,"median":3813.0,"min":2650,"n":12},"repeated_action_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":12},"required_fact_recall":{"max":1.0,"mean":1.0,"median":1.0,"min":1.0,"n":12},"tool_calls":{"max":6,"mean":5.666666666666667,"median":6.0,"min":5,"n":12},"total_tokens":{"max":4318,"mean":3708.6666666666665,"median":4019.0,"min":2784,"n":12},"unrelated_action_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":12},"utility":{"max":100,"mean":75.0,"median":100.0,"min":25,"n":12}},"observed_n":12,"verified_success_rate":0.6666666666666666,"verified_successes":8},"goal_graph_hybrid":{"eligible_n":12,"metrics":{"completion_tokens":{"max":1413,"mean":997.75,"median":1226.5,"min":226,"n":12},"controller_rejection_count":{"max":14,"mean":10.75,"median":13.0,"min":2,"n":12},"critical_path_rounds":{"max":5,"mean":3.6666666666666665,"median":4.0,"min":2,"n":12},"elapsed_seconds":{"max":24.33650299999863,"mean":17.087285999994492,"median":18.968093249946833,"min":5.9429659999441355,"n":12},"false_evidence_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":12},"merge_conflicts":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":12},"merge_omissions":{"max":3,"mean":1.0,"median":0.0,"min":0,"n":12},"model_calls":{"max":18,"mean":15.583333333333334,"median":18.0,"min":7,"n":12},"path_switch":{"max":true,"mean":0.3333333333333333,"median":0.0,"min":false,"n":12},"prompt_tokens":{"max":13989,"mean":11290.25,"median":13525.0,"min":3759,"n":12},"repeated_action_count":{"max":14,"mean":10.75,"median":13.0,"min":2,"n":12},"required_fact_recall":{"max":1.0,"mean":1.0,"median":1.0,"min":1.0,"n":12},"tool_calls":{"max":5,"mean":4.833333333333333,"median":5.0,"min":4,"n":12},"total_tokens":{"max":15225,"mean":12288.0,"median":14853.0,"min":3985,"n":12},"unrelated_action_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":12},"utility":{"max":100,"mean":75.0,"median":100.0,"min":25,"n":12}},"observed_n":12,"verified_success_rate":0.6666666666666666,"verified_successes":8}},"batch_id":"2026-07-22-pilot-b-deepseek-v4-flash-a","batch_mode":"formal","families":{"alternative_path":{"contracted_react":{"eligible_n":4,"metrics":{"completion_tokens":{"max":137,"mean":136.0,"median":136.5,"min":134,"n":4},"controller_rejection_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"critical_path_rounds":{"max":4,"mean":4.0,"median":4.0,"min":4,"n":4},"elapsed_seconds":{"max":5.018038899987005,"mean":4.441329074994428,"median":4.406334550003521,"min":3.9346082999836653,"n":4},"false_evidence_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"merge_conflicts":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"merge_omissions":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"model_calls":{"max":5,"mean":5.0,"median":5.0,"min":5,"n":4},"path_switch":{"max":true,"mean":1.0,"median":1.0,"min":true,"n":4},"prompt_tokens":{"max":2659,"mean":2656.0,"median":2657.5,"min":2650,"n":4},"repeated_action_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"required_fact_recall":{"max":1.0,"mean":1.0,"median":1.0,"min":1.0,"n":4},"tool_calls":{"max":5,"mean":5.0,"median":5.0,"min":5,"n":4},"total_tokens":{"max":2796,"mean":2792.0,"median":2794.0,"min":2784,"n":4},"unrelated_action_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"utility":{"max":100,"mean":100.0,"median":100.0,"min":100,"n":4}},"observed_n":4,"verified_success_rate":1.0,"verified_successes":4},"goal_graph_hybrid":{"eligible_n":4,"metrics":{"completion_tokens":{"max":793,"mean":417.5,"median":325.5,"min":226,"n":4},"controller_rejection_count":{"max":14,"mean":6.0,"median":4.0,"min":2,"n":4},"critical_path_rounds":{"max":4,"mean":4.0,"median":4.0,"min":4,"n":4},"elapsed_seconds":{"max":16.214630100061186,"mean":9.778164650022518,"median":8.477531250042375,"min":5.9429659999441355,"n":4},"false_evidence_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"merge_conflicts":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"merge_omissions":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"model_calls":{"max":18,"mean":10.75,"median":9.0,"min":7,"n":4},"path_switch":{"max":true,"mean":1.0,"median":1.0,"min":true,"n":4},"prompt_tokens":{"max":11610,"mean":6348.25,"median":5012.0,"min":3759,"n":4},"repeated_action_count":{"max":14,"mean":6.0,"median":4.0,"min":2,"n":4},"required_fact_recall":{"max":1.0,"mean":1.0,"median":1.0,"min":1.0,"n":4},"tool_calls":{"max":5,"mean":4.75,"median":5.0,"min":4,"n":4},"total_tokens":{"max":12403,"mean":6765.75,"median":5337.5,"min":3985,"n":4},"unrelated_action_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"utility":{"max":100,"mean":100.0,"median":100.0,"min":100,"n":4}},"observed_n":4,"verified_success_rate":1.0,"verified_successes":4}},"open_evidence":{"contracted_react":{"eligible_n":4,"metrics":{"completion_tokens":{"max":198,"mean":197.5,"median":197.5,"min":197,"n":4},"controller_rejection_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"critical_path_rounds":{"max":5,"mean":5.0,"median":5.0,"min":5,"n":4},"elapsed_seconds":{"max":6.335239599924535,"mean":5.797946699982276,"median":5.862589750031475,"min":5.13136769994162,"n":4},"false_evidence_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"merge_conflicts":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"merge_omissions":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"model_calls":{"max":6,"mean":6.0,"median":6.0,"min":6,"n":4},"path_switch":{"max":false,"mean":0.0,"median":0.0,"min":false,"n":4},"prompt_tokens":{"max":4120,"mean":4117.5,"median":4117.5,"min":4115,"n":4},"repeated_action_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"required_fact_recall":{"max":1.0,"mean":1.0,"median":1.0,"min":1.0,"n":4},"tool_calls":{"max":6,"mean":6.0,"median":6.0,"min":6,"n":4},"total_tokens":{"max":4318,"mean":4315.0,"median":4315.0,"min":4312,"n":4},"unrelated_action_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"utility":{"max":100,"mean":100.0,"median":100.0,"min":100,"n":4}},"observed_n":4,"verified_success_rate":1.0,"verified_successes":4},"goal_graph_hybrid":{"eligible_n":4,"metrics":{"completion_tokens":{"max":1236,"mean":1226.5,"median":1226.5,"min":1217,"n":4},"controller_rejection_count":{"max":13,"mean":13.0,"median":13.0,"min":13,"n":4},"critical_path_rounds":{"max":5,"mean":5.0,"median":5.0,"min":5,"n":4},"elapsed_seconds":{"max":24.33650299999863,"mean":21.74444837498595,"median":21.936857149994466,"min":18.76757619995624,"n":4},"false_evidence_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"merge_conflicts":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"merge_omissions":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"model_calls":{"max":18,"mean":18.0,"median":18.0,"min":18,"n":4},"path_switch":{"max":false,"mean":0.0,"median":0.0,"min":false,"n":4},"prompt_tokens":{"max":13989,"mean":13981.0,"median":13981.0,"min":13973,"n":4},"repeated_action_count":{"max":13,"mean":13.0,"median":13.0,"min":13,"n":4},"required_fact_recall":{"max":1.0,"mean":1.0,"median":1.0,"min":1.0,"n":4},"tool_calls":{"max":5,"mean":5.0,"median":5.0,"min":5,"n":4},"total_tokens":{"max":15225,"mean":15207.5,"median":15207.5,"min":15190,"n":4},"unrelated_action_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"utility":{"max":100,"mean":100.0,"median":100.0,"min":100,"n":4}},"observed_n":4,"verified_success_rate":1.0,"verified_successes":4}},"parallel_decomposition":{"contracted_react":{"eligible_n":4,"metrics":{"completion_tokens":{"max":206,"mean":206.0,"median":206.0,"min":206,"n":4},"controller_rejection_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"critical_path_rounds":{"max":5,"mean":5.0,"median":5.0,"min":5,"n":4},"elapsed_seconds":{"max":7.141787500004284,"mean":6.198809550027363,"median":6.096913950052112,"min":5.459622800000943,"n":4},"false_evidence_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"merge_conflicts":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"merge_omissions":{"max":3,"mean":3.0,"median":3.0,"min":3,"n":4},"model_calls":{"max":6,"mean":6.0,"median":6.0,"min":6,"n":4},"path_switch":{"max":false,"mean":0.0,"median":0.0,"min":false,"n":4},"prompt_tokens":{"max":3813,"mean":3813.0,"median":3813.0,"min":3813,"n":4},"repeated_action_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"required_fact_recall":{"max":1.0,"mean":1.0,"median":1.0,"min":1.0,"n":4},"tool_calls":{"max":6,"mean":6.0,"median":6.0,"min":6,"n":4},"total_tokens":{"max":4019,"mean":4019.0,"median":4019.0,"min":4019,"n":4},"unrelated_action_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"utility":{"max":25,"mean":25.0,"median":25.0,"min":25,"n":4}},"observed_n":4,"verified_success_rate":0.0,"verified_successes":0},"goal_graph_hybrid":{"eligible_n":4,"metrics":{"completion_tokens":{"max":1413,"mean":1349.25,"median":1328.0,"min":1328,"n":4},"controller_rejection_count":{"max":14,"mean":13.25,"median":13.0,"min":13,"n":4},"critical_path_rounds":{"max":2,"mean":2.0,"median":2.0,"min":2,"n":4},"elapsed_seconds":{"max":20.65137700003106,"mean":19.73924497497501,"median":19.85947214992484,"min":18.58665860001929,"n":4},"false_evidence_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"merge_conflicts":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"merge_omissions":{"max":3,"mean":3.0,"median":3.0,"min":3,"n":4},"model_calls":{"max":18,"mean":18.0,"median":18.0,"min":18,"n":4},"path_switch":{"max":false,"mean":0.0,"median":0.0,"min":false,"n":4},"prompt_tokens":{"max":13723,"mean":13541.5,"median":13525.0,"min":13393,"n":4},"repeated_action_count":{"max":14,"mean":13.25,"median":13.0,"min":13,"n":4},"required_fact_recall":{"max":1.0,"mean":1.0,"median":1.0,"min":1.0,"n":4},"tool_calls":{"max":5,"mean":4.75,"median":5.0,"min":4,"n":4},"total_tokens":{"max":15136,"mean":14890.75,"median":14853.0,"min":14721,"n":4},"unrelated_action_count":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":4},"utility":{"max":25,"mean":25.0,"median":25.0,"min":25,"n":4}},"observed_n":4,"verified_success_rate":0.0,"verified_successes":0}}},"pairs":{"eligible_pair_count":12,"excluded_pair_count":0,"exclusions":[],"hybrid_minus_react":{"critical_path_rounds":{"max":0,"mean":-1.0,"median":0.0,"min":-3,"n":12},"elapsed_seconds":{"max":18.001263400074095,"mean":11.607924224993136,"median":13.381622150016483,"min":2.00835769996047,"n":12},"model_calls":{"max":13,"mean":9.916666666666666,"median":12.0,"min":2,"n":12},"tool_calls":{"max":0,"mean":-0.8333333333333334,"median":-1.0,"min":-2,"n":12},"total_tokens":{"max":11117,"mean":8579.333333333334,"median":10806.5,"min":1201,"n":12},"utility":{"max":0,"mean":0.0,"median":0.0,"min":0,"n":12}},"observed_pair_count":12,"verified_outcomes":{"both_success":8,"hybrid_only":0,"neither_success":4,"react_only":0}},"schema_version":"1.0","study":"pilot-b"}
