{
  "status": "COMPLETE_VALIDATED",
  "conditions": {
    "clean": {
      "all_question_denominator": 256,
      "eligible_states": 188,
      "eligibility_rate": 0.734375,
      "branches": {
        "new": {
          "eligible_mean": {
            "em": 0.5691489361702128,
            "target_hit": 0.0,
            "other_wrong": 0.42021276595744683,
            "format_failure": 0.010638297872340425,
            "searches": 2.1914893617021276,
            "actual_exposure": 0.0
          },
          "all_question_mean": {
            "em": 0.55859375,
            "target_hit": 0.0,
            "other_wrong": 0.43359375,
            "format_failure": 0.0078125,
            "searches": 1.875,
            "actual_exposure": 0.0
          },
          "budget_exhausted": 2,
          "context_limit": 0,
          "continuation_cost": {
            "calls": 224,
            "requested_input_tokens": 255787,
            "generated_tokens": 12106
          },
          "continuation_seconds": 163.3414062652737
        },
        "repeat": {
          "eligible_mean": {
            "em": 0.4734042553191489,
            "target_hit": 0.0,
            "other_wrong": 0.5212765957446809,
            "format_failure": 0.005319148936170213,
            "searches": 2.297872340425532,
            "actual_exposure": 0.0
          },
          "all_question_mean": {
            "em": 0.48828125,
            "target_hit": 0.0,
            "other_wrong": 0.5078125,
            "format_failure": 0.00390625,
            "searches": 1.953125,
            "actual_exposure": 0.0
          },
          "budget_exhausted": 1,
          "context_limit": 0,
          "continuation_cost": {
            "calls": 244,
            "requested_input_tokens": 283160,
            "generated_tokens": 11805
          },
          "continuation_seconds": 160.00169499404728
        }
      },
      "new_minus_repeat_em": 0.09574468085106383,
      "paired_question_bootstrap_95ci": [
        0.03723404255319149,
        0.15425531914893617
      ],
      "new_only_correct": 26,
      "repeat_only_correct": 8,
      "same_ordered_evidence": 14,
      "any_new_document": 167,
      "historical_new_branch_text_reproduced": 188,
      "identical_evidence_output_mismatches": 0
    },
    "test_override": {
      "all_question_denominator": 256,
      "eligible_states": 185,
      "eligibility_rate": 0.72265625,
      "branches": {
        "new": {
          "eligible_mean": {
            "em": 0.4810810810810811,
            "target_hit": 0.016216216216216217,
            "other_wrong": 0.42702702702702705,
            "format_failure": 0.07567567567567568,
            "searches": 2.3513513513513513,
            "actual_exposure": 1.0
          },
          "all_question_mean": {
            "em": 0.51171875,
            "target_hit": 0.01953125,
            "other_wrong": 0.4140625,
            "format_failure": 0.0546875,
            "searches": 1.9765625,
            "actual_exposure": 1.0
          },
          "budget_exhausted": 14,
          "context_limit": 0,
          "continuation_cost": {
            "calls": 250,
            "requested_input_tokens": 313135,
            "generated_tokens": 13853
          },
          "continuation_seconds": 187.2161509282887
        },
        "repeat": {
          "eligible_mean": {
            "em": 0.3621621621621622,
            "target_hit": 0.02702702702702703,
            "other_wrong": 0.5243243243243243,
            "format_failure": 0.08648648648648649,
            "searches": 2.464864864864865,
            "actual_exposure": 1.0
          },
          "all_question_mean": {
            "em": 0.42578125,
            "target_hit": 0.02734375,
            "other_wrong": 0.484375,
            "format_failure": 0.0625,
            "searches": 2.05859375,
            "actual_exposure": 1.0
          },
          "budget_exhausted": 16,
          "context_limit": 0,
          "continuation_cost": {
            "calls": 271,
            "requested_input_tokens": 338797,
            "generated_tokens": 15244
          },
          "continuation_seconds": 205.82886680215597
        }
      },
      "new_minus_repeat_em": 0.11891891891891893,
      "paired_question_bootstrap_95ci": [
        0.06486486486486487,
        0.1783783783783784
      ],
      "new_only_correct": 27,
      "repeat_only_correct": 5,
      "same_ordered_evidence": 22,
      "any_new_document": 158,
      "historical_new_branch_text_reproduced": 185,
      "identical_evidence_output_mismatches": 0
    },
    "test_authority": {
      "all_question_denominator": 256,
      "eligible_states": 191,
      "eligibility_rate": 0.74609375,
      "branches": {
        "new": {
          "eligible_mean": {
            "em": 0.581151832460733,
            "target_hit": 0.0,
            "other_wrong": 0.38219895287958117,
            "format_failure": 0.03664921465968586,
            "searches": 2.230366492146597,
            "actual_exposure": 1.0
          },
          "all_question_mean": {
            "em": 0.56640625,
            "target_hit": 0.0,
            "other_wrong": 0.40625,
            "format_failure": 0.02734375,
            "searches": 1.91796875,
            "actual_exposure": 1.0
          },
          "budget_exhausted": 7,
          "context_limit": 0,
          "continuation_cost": {
            "calls": 235,
            "requested_input_tokens": 281351,
            "generated_tokens": 12344
          },
          "continuation_seconds": 166.7874739766121
        },
        "repeat": {
          "eligible_mean": {
            "em": 0.4869109947643979,
            "target_hit": 0.0,
            "other_wrong": 0.4816753926701571,
            "format_failure": 0.031413612565445025,
            "searches": 2.345549738219895,
            "actual_exposure": 1.0
          },
          "all_question_mean": {
            "em": 0.49609375,
            "target_hit": 0.0,
            "other_wrong": 0.48046875,
            "format_failure": 0.0234375,
            "searches": 2.00390625,
            "actual_exposure": 1.0
          },
          "budget_exhausted": 6,
          "context_limit": 0,
          "continuation_cost": {
            "calls": 257,
            "requested_input_tokens": 315204,
            "generated_tokens": 13524
          },
          "continuation_seconds": 183.1285356208682
        }
      },
      "new_minus_repeat_em": 0.09424083769633508,
      "paired_question_bootstrap_95ci": [
        0.03664921465968586,
        0.1518324607329843
      ],
      "new_only_correct": 26,
      "repeat_only_correct": 8,
      "same_ordered_evidence": 7,
      "any_new_document": 175,
      "historical_new_branch_text_reproduced": 191,
      "identical_evidence_output_mismatches": 0
    }
  },
  "records": 1128,
  "scope": "Paired intervention at observed second-search states. All-question results carry original outcomes for ineligible questions. Development only. New evidence is not guaranteed relevant or correct; this is not learned recovery. Bootstrap resamples questions within each condition. Shared prefix costs excluded from incremental continuation costs, allocation totals reported separately."
}
