{
  "actual_utc": "2026-10-09T23:23:03.108689+00:00",
  "scope": "Post-hoc descriptive diagnostics, no new inference or primary-estimand change; overlapping error categories are not mutually exclusive",
  "conditions": [
    {
      "model": "qwen2.5-coder:0.5b",
      "condition": "single-contract",
      "observed_trials": 360,
      "empty_final_plans": 0,
      "duplicate_nonrecovery_final_plans": 48,
      "premature_nonfinal_claim_events": 0,
      "trials_with_premature_nonfinal_claim": 0,
      "correct_not_accepted": 0,
      "correct_accepted": 59,
      "accepted_final_artifacts": 223,
      "acceptance_coverage_observed": 0.6194444444444445,
      "conditional_accepted_error_observed": 0.7354260089686099,
      "trials_with_correct_installed_plan_at_any_recorded_stage": 59,
      "trials_with_prior_correct_plan_but_wrong_final": 0,
      "changed_installations_from_correct_to_wrong": 0,
      "stage_grade_scope": "Frozen external oracle, post hoc; readonly proposals excluded; recovery length omission retained. Descriptive transitions do not identify hidden reasoning or causal mediation.",
      "model_returns": 768,
      "roles": {
        "single": 768
      },
      "returned_finish_reasons": {
        "stop": 768
      },
      "measured_prompt_tokens": 607595,
      "backend_output_tokens": 35745,
      "backend_cached_prompt_tokens_where_available": 351932,
      "returns_without_cached_count": 0
    },
    {
      "model": "qwen2.5-coder:0.5b",
      "condition": "single-open",
      "observed_trials": 360,
      "empty_final_plans": 0,
      "duplicate_nonrecovery_final_plans": 42,
      "premature_nonfinal_claim_events": 0,
      "trials_with_premature_nonfinal_claim": 0,
      "correct_not_accepted": 0,
      "correct_accepted": 62,
      "accepted_final_artifacts": 357,
      "acceptance_coverage_observed": 0.9916666666666667,
      "conditional_accepted_error_observed": 0.8263305322128851,
      "trials_with_correct_installed_plan_at_any_recorded_stage": 62,
      "trials_with_prior_correct_plan_but_wrong_final": 0,
      "changed_installations_from_correct_to_wrong": 0,
      "stage_grade_scope": "Frozen external oracle, post hoc; readonly proposals excluded; recovery length omission retained. Descriptive transitions do not identify hidden reasoning or causal mediation.",
      "model_returns": 369,
      "roles": {
        "single": 369
      },
      "returned_finish_reasons": {
        "stop": 369
      },
      "measured_prompt_tokens": 160684,
      "backend_output_tokens": 17610,
      "backend_cached_prompt_tokens_where_available": 39767,
      "returns_without_cached_count": 0
    },
    {
      "model": "qwen2.5-coder:0.5b",
      "condition": "staged-contract",
      "observed_trials": 359,
      "empty_final_plans": 0,
      "duplicate_nonrecovery_final_plans": 28,
      "premature_nonfinal_claim_events": 336,
      "trials_with_premature_nonfinal_claim": 215,
      "correct_not_accepted": 52,
      "correct_accepted": 11,
      "accepted_final_artifacts": 106,
      "acceptance_coverage_observed": 0.29526462395543174,
      "conditional_accepted_error_observed": 0.8962264150943396,
      "trials_with_correct_installed_plan_at_any_recorded_stage": 63,
      "trials_with_prior_correct_plan_but_wrong_final": 0,
      "changed_installations_from_correct_to_wrong": 0,
      "stage_grade_scope": "Frozen external oracle, post hoc; readonly proposals excluded; recovery length omission retained. Descriptive transitions do not identify hidden reasoning or causal mediation.",
      "model_returns": 1436,
      "roles": {
        "planner": 359,
        "worker": 718,
        "reviewer": 359
      },
      "returned_finish_reasons": {
        "stop": 1436
      },
      "measured_prompt_tokens": 1018495,
      "backend_output_tokens": 63757,
      "backend_cached_prompt_tokens_where_available": 385098,
      "returns_without_cached_count": 0
    },
    {
      "model": "qwen2.5-coder:0.5b",
      "condition": "staged-open",
      "observed_trials": 360,
      "empty_final_plans": 0,
      "duplicate_nonrecovery_final_plans": 28,
      "premature_nonfinal_claim_events": 325,
      "trials_with_premature_nonfinal_claim": 194,
      "correct_not_accepted": 59,
      "correct_accepted": 3,
      "accepted_final_artifacts": 135,
      "acceptance_coverage_observed": 0.375,
      "conditional_accepted_error_observed": 0.9777777777777777,
      "trials_with_correct_installed_plan_at_any_recorded_stage": 63,
      "trials_with_prior_correct_plan_but_wrong_final": 1,
      "changed_installations_from_correct_to_wrong": 1,
      "stage_grade_scope": "Frozen external oracle, post hoc; readonly proposals excluded; recovery length omission retained. Descriptive transitions do not identify hidden reasoning or causal mediation.",
      "model_returns": 1440,
      "roles": {
        "planner": 360,
        "worker": 720,
        "reviewer": 360
      },
      "returned_finish_reasons": {
        "stop": 1440
      },
      "measured_prompt_tokens": 1014558,
      "backend_output_tokens": 63866,
      "backend_cached_prompt_tokens_where_available": 385605,
      "returns_without_cached_count": 0
    },
    {
      "model": "qwen2.5:0.5b",
      "condition": "single-contract",
      "observed_trials": 359,
      "empty_final_plans": 315,
      "duplicate_nonrecovery_final_plans": 1,
      "premature_nonfinal_claim_events": 0,
      "trials_with_premature_nonfinal_claim": 0,
      "correct_not_accepted": 1,
      "correct_accepted": 0,
      "accepted_final_artifacts": 12,
      "acceptance_coverage_observed": 0.033426183844011144,
      "conditional_accepted_error_observed": 1.0,
      "trials_with_correct_installed_plan_at_any_recorded_stage": 1,
      "trials_with_prior_correct_plan_but_wrong_final": 0,
      "changed_installations_from_correct_to_wrong": 0,
      "stage_grade_scope": "Frozen external oracle, post hoc; readonly proposals excluded; recovery length omission retained. Descriptive transitions do not identify hidden reasoning or causal mediation.",
      "model_returns": 1188,
      "roles": {
        "single": 1188
      },
      "returned_finish_reasons": {
        "stop": 1188
      },
      "measured_prompt_tokens": 1010011,
      "backend_output_tokens": 18585,
      "backend_cached_prompt_tokens_where_available": 651159,
      "returns_without_cached_count": 0
    },
    {
      "model": "qwen2.5:0.5b",
      "condition": "single-open",
      "observed_trials": 360,
      "empty_final_plans": 296,
      "duplicate_nonrecovery_final_plans": 7,
      "premature_nonfinal_claim_events": 0,
      "trials_with_premature_nonfinal_claim": 0,
      "correct_not_accepted": 2,
      "correct_accepted": 2,
      "accepted_final_artifacts": 266,
      "acceptance_coverage_observed": 0.7388888888888889,
      "conditional_accepted_error_observed": 0.9924812030075187,
      "trials_with_correct_installed_plan_at_any_recorded_stage": 4,
      "trials_with_prior_correct_plan_but_wrong_final": 0,
      "changed_installations_from_correct_to_wrong": 0,
      "stage_grade_scope": "Frozen external oracle, post hoc; readonly proposals excluded; recovery length omission retained. Descriptive transitions do not identify hidden reasoning or causal mediation.",
      "model_returns": 476,
      "roles": {
        "single": 476
      },
      "returned_finish_reasons": {
        "stop": 476
      },
      "measured_prompt_tokens": 249573,
      "backend_output_tokens": 9030,
      "backend_cached_prompt_tokens_where_available": 99213,
      "returns_without_cached_count": 0
    },
    {
      "model": "qwen2.5:0.5b",
      "condition": "staged-contract",
      "observed_trials": 359,
      "empty_final_plans": 333,
      "duplicate_nonrecovery_final_plans": 2,
      "premature_nonfinal_claim_events": 39,
      "trials_with_premature_nonfinal_claim": 39,
      "correct_not_accepted": 0,
      "correct_accepted": 0,
      "accepted_final_artifacts": 0,
      "acceptance_coverage_observed": 0.0,
      "conditional_accepted_error_observed": null,
      "trials_with_correct_installed_plan_at_any_recorded_stage": 0,
      "trials_with_prior_correct_plan_but_wrong_final": 0,
      "changed_installations_from_correct_to_wrong": 0,
      "stage_grade_scope": "Frozen external oracle, post hoc; readonly proposals excluded; recovery length omission retained. Descriptive transitions do not identify hidden reasoning or causal mediation.",
      "model_returns": 1436,
      "roles": {
        "planner": 359,
        "worker": 718,
        "reviewer": 359
      },
      "returned_finish_reasons": {
        "stop": 1436
      },
      "measured_prompt_tokens": 937440,
      "backend_output_tokens": 20740,
      "backend_cached_prompt_tokens_where_available": 380592,
      "returns_without_cached_count": 0
    },
    {
      "model": "qwen2.5:0.5b",
      "condition": "staged-open",
      "observed_trials": 359,
      "empty_final_plans": 297,
      "duplicate_nonrecovery_final_plans": 5,
      "premature_nonfinal_claim_events": 49,
      "trials_with_premature_nonfinal_claim": 40,
      "correct_not_accepted": 0,
      "correct_accepted": 0,
      "accepted_final_artifacts": 42,
      "acceptance_coverage_observed": 0.116991643454039,
      "conditional_accepted_error_observed": 1.0,
      "trials_with_correct_installed_plan_at_any_recorded_stage": 0,
      "trials_with_prior_correct_plan_but_wrong_final": 0,
      "changed_installations_from_correct_to_wrong": 0,
      "stage_grade_scope": "Frozen external oracle, post hoc; readonly proposals excluded; recovery length omission retained. Descriptive transitions do not identify hidden reasoning or causal mediation.",
      "model_returns": 1436,
      "roles": {
        "planner": 359,
        "worker": 718,
        "reviewer": 359
      },
      "returned_finish_reasons": {
        "stop": 1436
      },
      "measured_prompt_tokens": 905169,
      "backend_output_tokens": 22887,
      "backend_cached_prompt_tokens_where_available": 380526,
      "returns_without_cached_count": 0
    }
  ],
  "requirement_errors": [
    {
      "model": "qwen2.5-coder:0.5b",
      "family": "capacity",
      "observed_trials": 240,
      "failed_requirement_counts": {
        "exclusive_groups": 132,
        "optimal_value": 240,
        "unique_budget": 240
      },
      "interpretation": "Overlapping frozen groups; recovery partial groups omit public length, separately audited by strict grammar reference"
    },
    {
      "model": "qwen2.5-coder:0.5b",
      "family": "dependency",
      "observed_trials": 240,
      "failed_requirement_counts": {
        "complete_unique": 3,
        "deadlines": 240,
        "dependency_order": 192
      },
      "interpretation": "Overlapping frozen groups; recovery partial groups omit public length, separately audited by strict grammar reference"
    },
    {
      "model": "qwen2.5-coder:0.5b",
      "family": "freshness",
      "observed_trials": 239,
      "failed_requirement_counts": {
        "ascii_order": 226,
        "membership": 215,
        "unique": 110
      },
      "interpretation": "Overlapping frozen groups; recovery partial groups omit public length, separately audited by strict grammar reference"
    },
    {
      "model": "qwen2.5-coder:0.5b",
      "family": "ranking",
      "observed_trials": 240,
      "failed_requirement_counts": {
        "ascii_tie_order": 239,
        "max_consolidation_topk": 237,
        "size_unique": 135
      },
      "interpretation": "Overlapping frozen groups; recovery partial groups omit public length, separately audited by strict grammar reference"
    },
    {
      "model": "qwen2.5-coder:0.5b",
      "family": "recovery",
      "observed_trials": 240,
      "failed_requirement_counts": {
        "all_transactions_checkpointed": 1,
        "final_values": 1,
        "legal_transitions": 7,
        "required_crashes": 0
      },
      "interpretation": "Overlapping frozen groups; recovery partial groups omit public length, separately audited by strict grammar reference"
    },
    {
      "model": "qwen2.5-coder:0.5b",
      "family": "validation",
      "observed_trials": 240,
      "failed_requirement_counts": {
        "ascii_order": 240,
        "membership": 240,
        "unique": 0
      },
      "interpretation": "Overlapping frozen groups; recovery partial groups omit public length, separately audited by strict grammar reference"
    },
    {
      "model": "qwen2.5:0.5b",
      "family": "capacity",
      "observed_trials": 239,
      "failed_requirement_counts": {
        "exclusive_groups": 16,
        "optimal_value": 238,
        "unique_budget": 35
      },
      "interpretation": "Overlapping frozen groups; recovery partial groups omit public length, separately audited by strict grammar reference"
    },
    {
      "model": "qwen2.5:0.5b",
      "family": "dependency",
      "observed_trials": 240,
      "failed_requirement_counts": {
        "complete_unique": 215,
        "deadlines": 26,
        "dependency_order": 19
      },
      "interpretation": "Overlapping frozen groups; recovery partial groups omit public length, separately audited by strict grammar reference"
    },
    {
      "model": "qwen2.5:0.5b",
      "family": "freshness",
      "observed_trials": 240,
      "failed_requirement_counts": {
        "ascii_order": 240,
        "membership": 240,
        "unique": 0
      },
      "interpretation": "Overlapping frozen groups; recovery partial groups omit public length, separately audited by strict grammar reference"
    },
    {
      "model": "qwen2.5:0.5b",
      "family": "ranking",
      "observed_trials": 240,
      "failed_requirement_counts": {
        "ascii_tie_order": 240,
        "max_consolidation_topk": 240,
        "size_unique": 239
      },
      "interpretation": "Overlapping frozen groups; recovery partial groups omit public length, separately audited by strict grammar reference"
    },
    {
      "model": "qwen2.5:0.5b",
      "family": "recovery",
      "observed_trials": 238,
      "failed_requirement_counts": {
        "all_transactions_checkpointed": 233,
        "final_values": 233,
        "legal_transitions": 0,
        "required_crashes": 173
      },
      "interpretation": "Overlapping frozen groups; recovery partial groups omit public length, separately audited by strict grammar reference"
    },
    {
      "model": "qwen2.5:0.5b",
      "family": "validation",
      "observed_trials": 240,
      "failed_requirement_counts": {
        "ascii_order": 240,
        "membership": 240,
        "unique": 0
      },
      "interpretation": "Overlapping frozen groups; recovery partial groups omit public length, separately audited by strict grammar reference"
    }
  ]
}