{
  "schema_version": 1,
  "benchmark": "ptichi.work_explanation_recovery_benchmark@1.0",
  "dataset": "ptichi.work_explanation_recovery_fixtures@1.0",
  "rubric": "ptichi.listener_recoverability_rubric@1.0",
  "input_identities": {
    "benchmark_contract": {
      "path": "public/data/work-explanation-recovery-benchmark.v1.json",
      "sha256": "dea92af2f8782785a87d30a0da52349bf3bb76e6d7d80e0eb46f52c10dbaaec1"
    },
    "fixtures": {
      "path": "public/data/work-explanation-recovery-fixtures.v1.json",
      "sha256": "990d2f4ccd9ac262c409745bfd0009a7b686e6b0df7adb6d66d30e807bb4ad4a"
    },
    "rubric": {
      "path": "public/data/listener-recoverability-rubric.v1.json",
      "sha256": "7ca0ebb85d473dab08e9dbe4891a6bd835b7d06c4a683dbfe6b2167426238665"
    },
    "evaluator": {
      "path": "scripts/run-recovery-benchmark.mjs",
      "sha256": "8c28ea338b98b6d661c16bef75b212340e4a82765a95e10f0bc9cf00480cee83"
    }
  },
  "case_count": 12,
  "baselines": [
    {
      "id": "identity",
      "metrics": {
        "case_count": 12,
        "target_repair_rate": 0,
        "required_anchor_preservation_rate": 0.3333,
        "forbidden_claim_rate": 0.4167,
        "all_structural_checks_pass_rate": 0,
        "average_token_edit_ratio": 0,
        "average_word_count_delta": 0
      },
      "by_job": {
        "concise_meeting_explanation": {
          "case_count": 2,
          "target_repair_rate": 0,
          "required_anchor_preservation_rate": 0,
          "forbidden_claim_rate": 0.5,
          "all_structural_checks_pass_rate": 0,
          "average_token_edit_ratio": 0,
          "average_word_count_delta": 0
        },
        "difficult_question": {
          "case_count": 2,
          "target_repair_rate": 0,
          "required_anchor_preservation_rate": 0.5,
          "forbidden_claim_rate": 0.5,
          "all_structural_checks_pass_rate": 0,
          "average_token_edit_ratio": 0,
          "average_word_count_delta": 0
        },
        "project_status_update": {
          "case_count": 2,
          "target_repair_rate": 0,
          "required_anchor_preservation_rate": 1,
          "forbidden_claim_rate": 0.5,
          "all_structural_checks_pass_rate": 0,
          "average_token_edit_ratio": 0,
          "average_word_count_delta": 0
        },
        "recommendation": {
          "case_count": 2,
          "target_repair_rate": 0,
          "required_anchor_preservation_rate": 0,
          "forbidden_claim_rate": 0,
          "all_structural_checks_pass_rate": 0,
          "average_token_edit_ratio": 0,
          "average_word_count_delta": 0
        },
        "risk_or_delay_explanation": {
          "case_count": 2,
          "target_repair_rate": 0,
          "required_anchor_preservation_rate": 0,
          "forbidden_claim_rate": 0.5,
          "all_structural_checks_pass_rate": 0,
          "average_token_edit_ratio": 0,
          "average_word_count_delta": 0
        },
        "technical_explanation": {
          "case_count": 2,
          "target_repair_rate": 0,
          "required_anchor_preservation_rate": 0.5,
          "forbidden_claim_rate": 0.5,
          "all_structural_checks_pass_rate": 0,
          "average_token_edit_ratio": 0,
          "average_word_count_delta": 0
        }
      }
    },
    {
      "id": "first_sentence_only",
      "metrics": {
        "case_count": 12,
        "target_repair_rate": 0,
        "required_anchor_preservation_rate": 0,
        "forbidden_claim_rate": 0.4167,
        "all_structural_checks_pass_rate": 0,
        "average_token_edit_ratio": 0.4197,
        "average_word_count_delta": -9.0833
      },
      "by_job": {
        "concise_meeting_explanation": {
          "case_count": 2,
          "target_repair_rate": 0,
          "required_anchor_preservation_rate": 0,
          "forbidden_claim_rate": 0.5,
          "all_structural_checks_pass_rate": 0,
          "average_token_edit_ratio": 0.2321,
          "average_word_count_delta": -6.5
        },
        "difficult_question": {
          "case_count": 2,
          "target_repair_rate": 0,
          "required_anchor_preservation_rate": 0,
          "forbidden_claim_rate": 0.5,
          "all_structural_checks_pass_rate": 0,
          "average_token_edit_ratio": 0.3346,
          "average_word_count_delta": -6.5
        },
        "project_status_update": {
          "case_count": 2,
          "target_repair_rate": 0,
          "required_anchor_preservation_rate": 0,
          "forbidden_claim_rate": 0.5,
          "all_structural_checks_pass_rate": 0,
          "average_token_edit_ratio": 0.6541,
          "average_word_count_delta": -19
        },
        "recommendation": {
          "case_count": 2,
          "target_repair_rate": 0,
          "required_anchor_preservation_rate": 0,
          "forbidden_claim_rate": 0,
          "all_structural_checks_pass_rate": 0,
          "average_token_edit_ratio": 0.5209,
          "average_word_count_delta": -7.5
        },
        "risk_or_delay_explanation": {
          "case_count": 2,
          "target_repair_rate": 0,
          "required_anchor_preservation_rate": 0,
          "forbidden_claim_rate": 0.5,
          "all_structural_checks_pass_rate": 0,
          "average_token_edit_ratio": 0.4834,
          "average_word_count_delta": -6.5
        },
        "technical_explanation": {
          "case_count": 2,
          "target_repair_rate": 0,
          "required_anchor_preservation_rate": 0,
          "forbidden_claim_rate": 0.5,
          "all_structural_checks_pass_rate": 0,
          "average_token_edit_ratio": 0.2931,
          "average_word_count_delta": -8.5
        }
      }
    },
    {
      "id": "authored_reference_take_b",
      "metrics": {
        "case_count": 12,
        "target_repair_rate": 1,
        "required_anchor_preservation_rate": 1,
        "forbidden_claim_rate": 0,
        "all_structural_checks_pass_rate": 1,
        "average_token_edit_ratio": 0.5259,
        "average_word_count_delta": 2.75
      },
      "by_job": {
        "concise_meeting_explanation": {
          "case_count": 2,
          "target_repair_rate": 1,
          "required_anchor_preservation_rate": 1,
          "forbidden_claim_rate": 0,
          "all_structural_checks_pass_rate": 1,
          "average_token_edit_ratio": 0.7053,
          "average_word_count_delta": -5.5
        },
        "difficult_question": {
          "case_count": 2,
          "target_repair_rate": 1,
          "required_anchor_preservation_rate": 1,
          "forbidden_claim_rate": 0,
          "all_structural_checks_pass_rate": 1,
          "average_token_edit_ratio": 0.5834,
          "average_word_count_delta": 4
        },
        "project_status_update": {
          "case_count": 2,
          "target_repair_rate": 1,
          "required_anchor_preservation_rate": 1,
          "forbidden_claim_rate": 0,
          "all_structural_checks_pass_rate": 1,
          "average_token_edit_ratio": 0.2813,
          "average_word_count_delta": -0.5
        },
        "recommendation": {
          "case_count": 2,
          "target_repair_rate": 1,
          "required_anchor_preservation_rate": 1,
          "forbidden_claim_rate": 0,
          "all_structural_checks_pass_rate": 1,
          "average_token_edit_ratio": 0.3,
          "average_word_count_delta": 6
        },
        "risk_or_delay_explanation": {
          "case_count": 2,
          "target_repair_rate": 1,
          "required_anchor_preservation_rate": 1,
          "forbidden_claim_rate": 0,
          "all_structural_checks_pass_rate": 1,
          "average_token_edit_ratio": 0.5625,
          "average_word_count_delta": 7.5
        },
        "technical_explanation": {
          "case_count": 2,
          "target_repair_rate": 1,
          "required_anchor_preservation_rate": 1,
          "forbidden_claim_rate": 0,
          "all_structural_checks_pass_rate": 1,
          "average_token_edit_ratio": 0.723,
          "average_word_count_delta": 5
        }
      }
    }
  ],
  "case_results_csv": "/data/work-explanation-recovery-benchmark-results.v1.csv",
  "claim_boundary": "These are literal lexical/order checks over authored synthetic text. They do not measure semantic truth, listener comprehension, clarity, learning, transfer, preference or Ptichi product efficacy. A misleading statement can contain the expected witnesses, while a faithful paraphrase can fail them."
}
