{
  "complete_registered_paired_measurement": true,
  "original_baseline_viability": {
    "passed": false,
    "errors": 60,
    "denominator": 100,
    "user_override": "Environment accepted and product work authorized; viability score unchanged."
  },
  "outcome_distributions_including_refusals_and_blocking": {
    "original": {
      "refusal_unverified": 4,
      "task_error": 161,
      "correct_completed": 36,
      "silent_wrong": 31,
      "comparison_unverified": 18
    },
    "candidate": {
      "refusal_unverified": 24,
      "task_error": 93,
      "correct_completed": 57,
      "unnecessary_clarification": 1,
      "comparison_unverified": 21,
      "silent_wrong": 54
    }
  },
  "targets": [
    {
      "target": "zero_curated_security_cohort_grain_unit_calendar_violations",
      "status": "passed",
      "evidence": [
        "tests-v18/final-curated-and-matrix-binding.json"
      ],
      "measurements": {
        "checks_run": 568,
        "passed": 568,
        "failed": 0,
        "skipped": 0
      },
      "limit": "Finite critical suite; not proof of arbitrary question meaning."
    },
    {
      "target": "zero_unclassified_errors_in_both_arms",
      "status": "failed",
      "measurements": {
        "original": 21,
        "candidate": 0,
        "postrun_audited_classes_do_not_replace_original_scores": true
      }
    },
    {
      "target": "zero_fixture_caused_errors_in_both_arms",
      "status": "passed",
      "evidence": [
        "phase0/fixture-self-check.json",
        "phase0/e5-classification-audit.json",
        "phase0/final-v18-heldout/terminal-and-environment-audit.json"
      ],
      "measurements": {
        "fixture_caused_errors": {
          "original": 0,
          "candidate": 0
        }
      },
      "limit": "Generated absent aliases and wrong-connection SQL are product defects. The authorized physical source fields exist."
    },
    {
      "target": "zero_environment_caused_errors_in_both_arms",
      "status": "failed",
      "evidence": [
        "phase0/final-v18-heldout/terminal-and-environment-audit.json",
        "phase0/final-v18-heldout/provider-and-job-records-original.json",
        "phase0/final-v18-heldout/provider-and-job-records-candidate.json"
      ],
      "measurements": {
        "configuration_caused_errors": {
          "original": 0,
          "candidate": 0
        },
        "provider_availability_failed_calls": {
          "original": 0,
          "candidate": 20
        },
        "missing_api_key_errors": {
          "original": 0,
          "candidate": 0
        }
      },
      "limit": "The strict registered target counts every failed external-generation boundary call. An exception does not by itself prove a remote outage; configuration remains valid."
    },
    {
      "target": "candidate_first_turn_execution_errors_at_most_15_percent",
      "status": "failed",
      "measurements": {
        "original_errors": 60,
        "candidate_errors": 35,
        "denominator": 100,
        "candidate_rate": 0.35
      }
    },
    {
      "target": "candidate_precision_at_least_80_percent_and_lower95_above_original",
      "status": "failed",
      "measurements": {
        "original_correct": 36,
        "original_denominator": 67,
        "candidate_correct": 57,
        "candidate_denominator": 111,
        "candidate_precision": 0.5135135135135135,
        "paired_difference_ci95": [
          -0.23182397959183676,
          0.18543292456335936
        ],
        "undefined_bootstrap_draws": 0,
        "interval_conditions_on_defined_draws": false
      }
    },
    {
      "target": "silent_wrong_relative_reduction_at_least_50_percent_lower95_at_least_30_percent",
      "status": "failed",
      "measurements": {
        "original_wrong": 31,
        "candidate_wrong": 54,
        "denominator_per_arm": 250,
        "relative_reduction": -0.7419354838709677,
        "paired_relative_reduction_ci95": [
          -2.3529411764705883,
          0.022727272727272728
        ],
        "undefined_bootstrap_draws": 0,
        "interval_conditions_on_defined_draws": false,
        "priority": "secondary under the user amendment"
      }
    },
    {
      "target": "candidate_correct_supported_first_turns_at_least_40_percent",
      "status": "failed",
      "measurements": {
        "original_correct": 14,
        "candidate_correct": 18,
        "denominator": 100
      }
    },
    {
      "target": "candidate_unnecessary_clarification_at_most_5_percent_of_successful_parents",
      "status": "passed",
      "measurements": {
        "original_count": 0,
        "original_denominator": 62,
        "candidate_count": 0,
        "candidate_denominator": 106,
        "candidate_rate": 0.0
      }
    },
    {
      "target": "candidate_full_conversation_success_at_least_30_percent",
      "status": "failed",
      "measurements": {
        "original_successes": 1,
        "candidate_successes": 4,
        "denominator": 50
      }
    },
    {
      "target": "zero_paired_correct_to_error_or_wrong_regressions",
      "status": "failed",
      "measurements": {
        "count": 7,
        "attempts": [
          "independent11:quote_only_system_activity:first-count:base:repeat1",
          "independent11:quote_days_before_activation:first-count:base:repeat1",
          "independent11:customer_referrals_with_reward_event:t4:base:repeat1",
          "independent11:organization_wallet_share_unknown:t1:base:repeat1",
          "independent11:organization_wallet_share_unknown:t3:base:repeat1",
          "independent11:activity_duration_missing_due_time:t3:base:repeat1",
          "independent11:deal_stage_timestamp_order:t1:base:repeat1"
        ]
      }
    },
    {
      "target": "candidate_jobs_reaching_deadline_at_most_1_percent",
      "status": "passed",
      "measurements": {
        "original_deadlines": 0,
        "original_submitted": 250,
        "candidate_deadlines": 0,
        "candidate_submitted": 250
      }
    },
    {
      "target": "zero_unsupported_numeric_claims_on_curated_narrative_suite",
      "status": "passed",
      "evidence": [
        "tests-v18/final-curated-and-matrix-binding.json"
      ],
      "measurements": {
        "checks_run": 12,
        "passed": 12,
        "failed": 0,
        "skipped": 0
      },
      "limit": "Twelve curated metadata-backed narrative controls. Query meaning and legacy metadata-free callers are outside this finite claim."
    },
    {
      "target": "all_required_result_surfaces_agree_on_checked_facts_and_scope",
      "status": "unverified",
      "evidence": [
        "results-v8/v18-native-acceptance-input.json",
        "results-v8/v18-ui-verification-source-binding.json"
      ],
      "measurements": {
        "native_cases_passed": 6,
        "native_cases_failed": 0,
        "native_checked_members": 45,
        "native_checked_total_gbp": "2454.50",
        "native_count": 45,
        "native_final_reads_without_new_work": 30,
        "observed_modes": [
          "API",
          "table",
          "headline",
          "snapshot CSV",
          "authorized same-vault full-details share API",
          "share DOM",
          "share reload"
        ],
        "unverified_native_modes": [
          "Compose",
          "refreshed CSV after source mutation",
          "mixed-currency browser output"
        ]
      },
      "limit": "The checked surfaces agree. Full coverage of every required result surface is not established by the six native cases."
    },
    {
      "target": "deterministic_adapter_p95_regression_at_most_10_percent",
      "status": "failed",
      "measurements": {
        "timing_purpose": "uncontended_acceptance",
        "paired_configuration": {
          "equal": true,
          "excluded_replica_identity_keys": [
            "LORE_BI_FEDERATION_CT_CONNECTION_ID",
            "LORE_BI_FEDERATION_GOLD_CONNECTION_ID",
            "LORE_BI_FEDERATION_VAULT_ID",
            "LORE_DB_PATH"
          ],
          "shared_configuration_sha256": "58fc02223a1f8d38dec9a5e0726dd70850c138b4c2e55d76cfff496ae25407aa"
        },
        "workload_sha256": "193aed08832bb0feb808c0e0c3bd4817b7e09a8b3ff7a13a140f46aa46ad963e",
        "response_supplement_sha256": "56e8a72041ecf76af6324fd7f0f1e9b0ca7ef430d50ca79ace464513aa7789eb",
        "arms": {
          "original": {
            "attempts": 90,
            "p95_ms": 1549.4488747790456,
            "successful_results": 54,
            "completed_fixed_queries": 36,
            "terminal_errors_or_refusals": 36,
            "fixed_provider_proven_for_every_job": true,
            "all_external_calls": {
              "valid": true,
              "failure_codes": [],
              "sql_external_calls": 159,
              "auxiliary_external_calls": 87,
              "persisted_call_records": 246,
              "usage_ledger_calls": 87,
              "guard_violations": 0
            }
          },
          "candidate": {
            "attempts": 90,
            "p95_ms": 1732.0084376260638,
            "successful_results": 69,
            "completed_fixed_queries": 66,
            "terminal_errors_or_refusals": 21,
            "fixed_provider_proven_for_every_job": true,
            "all_external_calls": {
              "valid": true,
              "failure_codes": [],
              "sql_external_calls": 111,
              "auxiliary_external_calls": 69,
              "persisted_call_records": 180,
              "usage_ledger_calls": 69,
              "guard_violations": 0
            }
          }
        },
        "candidate_to_original_p95_ratio": 1.1178222565575464,
        "all_fixed_queries_completed": false,
        "target": "failed",
        "numerical_p95_ratio_at_most_1_10": false,
        "fixed_provider_verified": true,
        "non_answers_without_persisted_llm_calls": 57,
        "missing_job_context_association": "Keep raw null job_id. Associate fixed case and complete provider interval within the client's real-monotonic attempt window.",
        "workload_conformity_checked_outside_timing": true,
        "request_row_limit": 10000,
        "enrichment_policy": "Only additive client/client_type columns beside a requested user_id. All requested fields, order, membership, values and types remain exact. No truncated result completes the fixed query.",
        "poll_interval_ms": 250,
        "timing_scope": "Client monotonic submission through terminal persisted result read. Includes 250 ms polling resolution and HTTP overhead; not worker-only execution time.",
        "all_timings_retained": true,
        "timings_filtered_for_correctness": false,
        "no_quality_claim_from_deterministic_provider": true
      },
      "limit": "The fixed method retains every timing and reports query conformity separately."
    }
  ],
  "all_targets_passed": false,
  "report_input_sha256": {
    "/tmp/lore-goal3-eval/final-v18-paired.json": "de0d429c89684009885e19f0ff8e68e964089697ed6a42786edb97462eb8cce1",
    "/tmp/lore-goal3-eval/final-v18-evidence-checks.json": "e718bcd7e148ed2002e0d471039339ea9fa778cc04d5fe2e6f20132fd5bda7d0",
    "/tmp/lore-goal3-eval/latency-final-run/summary.json": "d4b2aa34fd5dc18c622f1e89f7e8fbd40d9405bf69cdb7312080ed1aea43dab7"
  }
}