{
  "checks_failed": 0,
  "checks_indeterminate": 1,
  "checks_passed": 95,
  "checks_total": 96,
  "coarsening_precision": {
    "ambiguous_items": [
      {
        "correct": 0,
        "exported_confidence": 0.5,
        "possible_coarse_levels": [
          0.3333333333333333,
          0.6666666666666666
        ],
        "task_id": "mmlu-abstract_algebra-1"
      },
      {
        "correct": 1,
        "exported_confidence": 0.8333,
        "possible_coarse_levels": [
          0.6666666666666666,
          1.0
        ],
        "task_id": "mmlu-elementary_mathematics-12"
      },
      {
        "correct": 0,
        "exported_confidence": 0.5,
        "possible_coarse_levels": [
          0.3333333333333333,
          0.6666666666666666
        ],
        "task_id": "triviaqa-tc_133"
      }
    ],
    "maximum_auc": 0.779320987654321,
    "minimum_auc": 0.7569444444444444,
    "published_value_within_rounding_bounds": true,
    "reason": "Rounded confidence intervals cross coarsening boundaries; original precision is unavailable"
  },
  "comparisons": [
    {
      "absolute_difference": 0,
      "field": "arms.swarm_round0.n",
      "passed": true,
      "published": 66,
      "recomputed": 66,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_round0.accuracy",
      "passed": true,
      "published": 0.8485,
      "recomputed": 0.8485,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_round0.mean_confidence",
      "passed": true,
      "published": 0.8939,
      "recomputed": 0.8939,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_round0.overconfidence",
      "passed": true,
      "published": 0.0455,
      "recomputed": 0.0455,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 9.999999999999593e-05,
      "field": "arms.swarm_round0.ece",
      "passed": true,
      "published": 0.0556,
      "recomputed": 0.0555,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_round0.mce",
      "passed": true,
      "published": 0.0833,
      "recomputed": 0.0833,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_round0.brier",
      "passed": true,
      "published": 0.0993,
      "recomputed": 0.0993,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_round0.auc",
      "passed": true,
      "published": 0.7812,
      "recomputed": 0.7812,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0,
      "field": "arms.swarm_postdebate.n",
      "passed": true,
      "published": 66,
      "recomputed": 66,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_postdebate.accuracy",
      "passed": true,
      "published": 0.8182,
      "recomputed": 0.8182,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_postdebate.mean_confidence",
      "passed": true,
      "published": 0.9293,
      "recomputed": 0.9293,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_postdebate.overconfidence",
      "passed": true,
      "published": 0.1111,
      "recomputed": 0.1111,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_postdebate.ece",
      "passed": true,
      "published": 0.1111,
      "recomputed": 0.1111,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_postdebate.mce",
      "passed": true,
      "published": 0.3333,
      "recomputed": 0.3333,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_postdebate.brier",
      "passed": true,
      "published": 0.1279,
      "recomputed": 0.1279,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.swarm_postdebate.auc",
      "passed": true,
      "published": 0.7407,
      "recomputed": 0.7407,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0,
      "field": "arms.baseline_first3.n",
      "passed": true,
      "published": 66,
      "recomputed": 66,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_first3.accuracy",
      "passed": true,
      "published": 0.8182,
      "recomputed": 0.8182,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_first3.mean_confidence",
      "passed": true,
      "published": 0.8939,
      "recomputed": 0.8939,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_first3.overconfidence",
      "passed": true,
      "published": 0.0758,
      "recomputed": 0.0758,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_first3.ece",
      "passed": true,
      "published": 0.0758,
      "recomputed": 0.0758,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_first3.mce",
      "passed": true,
      "published": 0.3333,
      "recomputed": 0.3333,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_first3.brier",
      "passed": true,
      "published": 0.1162,
      "recomputed": 0.1162,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_first3.auc",
      "passed": true,
      "published": 0.7639,
      "recomputed": 0.7639,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0,
      "field": "arms.baseline_full.n",
      "passed": true,
      "published": 66,
      "recomputed": 66,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_full.accuracy",
      "passed": true,
      "published": 0.8182,
      "recomputed": 0.8182,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_full.mean_confidence",
      "passed": true,
      "published": 0.8702,
      "recomputed": 0.8702,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_full.overconfidence",
      "passed": true,
      "published": 0.052,
      "recomputed": 0.052,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_full.ece",
      "passed": true,
      "published": 0.109,
      "recomputed": 0.109,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_full.mce",
      "passed": true,
      "published": 0.4286,
      "recomputed": 0.4286,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_full.brier",
      "passed": true,
      "published": 0.1,
      "recomputed": 0.1,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "arms.baseline_full.auc",
      "passed": true,
      "published": 0.8295,
      "recomputed": 0.8295,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.007000000000000006,
      "field": "baseline_full_auc_coarsened_to_3_levels",
      "passed": null,
      "published": 0.7639,
      "recomputed": 0.7569,
      "rounding_bounds": [
        0.7569444444444444,
        0.779320987654321
      ],
      "status": "indeterminate_original_precision_required",
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0,
      "field": "round0_unanimity.unanimous_items",
      "passed": true,
      "published": 49,
      "recomputed": 49,
      "tolerance": 0
    },
    {
      "absolute_difference": 0,
      "field": "round0_unanimity.n",
      "passed": true,
      "published": 66,
      "recomputed": 66,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "round0_unanimity.share",
      "passed": true,
      "published": 0.7424,
      "recomputed": 0.7424,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0,
      "field": "accuracy_discordant_items_swarm_vs_baseline",
      "passed": true,
      "published": 2,
      "recomputed": 2,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.overconf_postdebate_minus_round0.mean",
      "passed": true,
      "published": 0.0658,
      "recomputed": 0.0658,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.overconf_postdebate_minus_round0.lo",
      "passed": true,
      "published": 0.0202,
      "recomputed": 0.0202,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.overconf_postdebate_minus_round0.hi",
      "passed": true,
      "published": 0.1162,
      "recomputed": 0.1162,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.conf_postdebate_minus_baseline3.mean",
      "passed": true,
      "published": 0.0357,
      "recomputed": 0.0357,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 9.99999999999994e-05,
      "field": "deltas_ci95.conf_postdebate_minus_baseline3.lo",
      "passed": true,
      "published": -0.0152,
      "recomputed": -0.0151,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.conf_postdebate_minus_baseline3.hi",
      "passed": true,
      "published": 0.0859,
      "recomputed": 0.0859,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.overconf_postdebate_minus_baseline3.mean",
      "passed": true,
      "published": 0.0358,
      "recomputed": 0.0358,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.overconf_postdebate_minus_baseline3.lo",
      "passed": true,
      "published": -0.0202,
      "recomputed": -0.0202,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.overconf_postdebate_minus_baseline3.hi",
      "passed": true,
      "published": 0.0909,
      "recomputed": 0.0909,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.conf_postdebate_minus_baselinefull.mean",
      "passed": true,
      "published": 0.0594,
      "recomputed": 0.0594,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.conf_postdebate_minus_baselinefull.lo",
      "passed": true,
      "published": 0.0158,
      "recomputed": 0.0158,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.conf_postdebate_minus_baselinefull.hi",
      "passed": true,
      "published": 0.1058,
      "recomputed": 0.1058,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.auc_postdebate_minus_baselinefull.mean",
      "passed": true,
      "published": -0.0889,
      "recomputed": -0.0889,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.auc_postdebate_minus_baselinefull.lo",
      "passed": true,
      "published": -0.276,
      "recomputed": -0.276,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.auc_postdebate_minus_baselinefull.hi",
      "passed": true,
      "published": 0.08,
      "recomputed": 0.08,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.auc_postdebate_minus_baseline3.mean",
      "passed": true,
      "published": -0.0234,
      "recomputed": -0.0234,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.auc_postdebate_minus_baseline3.lo",
      "passed": true,
      "published": -0.2419,
      "recomputed": -0.2419,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.auc_postdebate_minus_baseline3.hi",
      "passed": true,
      "published": 0.196,
      "recomputed": 0.196,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.acc_postdebate_minus_baselinefull.mean",
      "passed": true,
      "published": -0.0,
      "recomputed": -0.0,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.acc_postdebate_minus_baselinefull.lo",
      "passed": true,
      "published": -0.0455,
      "recomputed": -0.0455,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.acc_postdebate_minus_baselinefull.hi",
      "passed": true,
      "published": 0.0455,
      "recomputed": 0.0455,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.acc_postdebate_minus_round0.mean",
      "passed": true,
      "published": -0.0304,
      "recomputed": -0.0304,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.acc_postdebate_minus_round0.lo",
      "passed": true,
      "published": -0.0758,
      "recomputed": -0.0758,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "deltas_ci95.acc_postdebate_minus_round0.hi",
      "passed": true,
      "published": 0.0,
      "recomputed": 0.0,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0,
      "field": "per_source_postdebate_vs_baseline.fresh.n",
      "passed": true,
      "published": 2,
      "recomputed": 2,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.fresh.swarm_postdebate.accuracy",
      "passed": true,
      "published": 0.0,
      "recomputed": 0.0,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.fresh.swarm_postdebate.overconfidence",
      "passed": true,
      "published": 0.6667,
      "recomputed": 0.6667,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.fresh.baseline_full.accuracy",
      "passed": true,
      "published": 0.0,
      "recomputed": 0.0,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 9.999999999998899e-05,
      "field": "per_source_postdebate_vs_baseline.fresh.baseline_full.overconfidence",
      "passed": true,
      "published": 0.4464,
      "recomputed": 0.4465,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0,
      "field": "per_source_postdebate_vs_baseline.gsm8k.n",
      "passed": true,
      "published": 20,
      "recomputed": 20,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.gsm8k.swarm_postdebate.accuracy",
      "passed": true,
      "published": 0.9,
      "recomputed": 0.9,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.gsm8k.swarm_postdebate.overconfidence",
      "passed": true,
      "published": 0.0667,
      "recomputed": 0.0667,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.gsm8k.baseline_full.accuracy",
      "passed": true,
      "published": 0.95,
      "recomputed": 0.95,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.gsm8k.baseline_full.overconfidence",
      "passed": true,
      "published": -0.0205,
      "recomputed": -0.0205,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0,
      "field": "per_source_postdebate_vs_baseline.mc.n",
      "passed": true,
      "published": 1,
      "recomputed": 1,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.mc.swarm_postdebate.accuracy",
      "passed": true,
      "published": 1.0,
      "recomputed": 1.0,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.mc.swarm_postdebate.overconfidence",
      "passed": true,
      "published": -0.3333,
      "recomputed": -0.3333,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.mc.baseline_full.accuracy",
      "passed": true,
      "published": 1.0,
      "recomputed": 1.0,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.mc.baseline_full.overconfidence",
      "passed": true,
      "published": -0.2857,
      "recomputed": -0.2857,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0,
      "field": "per_source_postdebate_vs_baseline.mmlu.n",
      "passed": true,
      "published": 20,
      "recomputed": 20,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.mmlu.swarm_postdebate.accuracy",
      "passed": true,
      "published": 1.0,
      "recomputed": 1.0,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.mmlu.swarm_postdebate.overconfidence",
      "passed": true,
      "published": -0.05,
      "recomputed": -0.05,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.mmlu.baseline_full.accuracy",
      "passed": true,
      "published": 0.95,
      "recomputed": 0.95,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.mmlu.baseline_full.overconfidence",
      "passed": true,
      "published": -0.0738,
      "recomputed": -0.0738,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0,
      "field": "per_source_postdebate_vs_baseline.priv.n",
      "passed": true,
      "published": 1,
      "recomputed": 1,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.priv.swarm_postdebate.accuracy",
      "passed": true,
      "published": 1.0,
      "recomputed": 1.0,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.priv.swarm_postdebate.overconfidence",
      "passed": true,
      "published": -0.3333,
      "recomputed": -0.3333,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.priv.baseline_full.accuracy",
      "passed": true,
      "published": 1.0,
      "recomputed": 1.0,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.priv.baseline_full.overconfidence",
      "passed": true,
      "published": -0.2857,
      "recomputed": -0.2857,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0,
      "field": "per_source_postdebate_vs_baseline.reason.n",
      "passed": true,
      "published": 2,
      "recomputed": 2,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.reason.swarm_postdebate.accuracy",
      "passed": true,
      "published": 1.0,
      "recomputed": 1.0,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.reason.swarm_postdebate.overconfidence",
      "passed": true,
      "published": 0.0,
      "recomputed": 0.0,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.reason.baseline_full.accuracy",
      "passed": true,
      "published": 1.0,
      "recomputed": 1.0,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.00010000000000001674,
      "field": "per_source_postdebate_vs_baseline.reason.baseline_full.overconfidence",
      "passed": true,
      "published": -0.1964,
      "recomputed": -0.1965,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0,
      "field": "per_source_postdebate_vs_baseline.triviaqa.n",
      "passed": true,
      "published": 20,
      "recomputed": 20,
      "tolerance": 0
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.triviaqa.swarm_postdebate.accuracy",
      "passed": true,
      "published": 0.6,
      "recomputed": 0.6,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.triviaqa.swarm_postdebate.overconfidence",
      "passed": true,
      "published": 0.3167,
      "recomputed": 0.3167,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.triviaqa.baseline_full.accuracy",
      "passed": true,
      "published": 0.6,
      "recomputed": 0.6,
      "tolerance": 0.0002
    },
    {
      "absolute_difference": 0.0,
      "field": "per_source_postdebate_vs_baseline.triviaqa.baseline_full.overconfidence",
      "passed": true,
      "published": 0.2696,
      "recomputed": 0.2696,
      "tolerance": 0.0002
    }
  ],
  "format_version": 1,
  "input_sha256": {
    "analysis_config": "22097b082e5e24b62756ca4748cfc0f7d1e36074cee034e5b6d9954a499db136",
    "per_item_csv": "c05aa524f260b99e815bfaefab4278bc19c6c3e6712c6ae29e91a42186d85ed2",
    "published_summary": "5103851807b3ec574176428c8e25ff843ca0ddb06941b131aa56bbf7677689ac"
  },
  "limitations": [
    "Correctness flags are accepted from the public CSV, not independently re-scored against a frozen truth key.",
    "CSV confidence values are rounded to four decimals and answer strings may be truncated to 120 characters; original response precision/text cannot be recovered.",
    "Original per-agent messages, events.jsonl and checkpoint.jsonl are absent from this allowlisted bundle; no raw-event reconstruction or model rerun is claimed.",
    "Bootstrap draws follow the documented Python random.Random convention and seed; no fresh experimental observations are generated.",
    "Agreement checks arithmetic over published evidence. It does not validate a causal interpretation or transfer empirical support to metaphysical claims."
  ],
  "maximum_absolute_difference": 0.007000000000000006,
  "n": 66,
  "numeric_coverage": {
    "published_fields": [
      "accuracy_discordant_items_swarm_vs_baseline",
      "arms.baseline_first3.accuracy",
      "arms.baseline_first3.auc",
      "arms.baseline_first3.brier",
      "arms.baseline_first3.ece",
      "arms.baseline_first3.mce",
      "arms.baseline_first3.mean_confidence",
      "arms.baseline_first3.n",
      "arms.baseline_first3.overconfidence",
      "arms.baseline_full.accuracy",
      "arms.baseline_full.auc",
      "arms.baseline_full.brier",
      "arms.baseline_full.ece",
      "arms.baseline_full.mce",
      "arms.baseline_full.mean_confidence",
      "arms.baseline_full.n",
      "arms.baseline_full.overconfidence",
      "arms.swarm_postdebate.accuracy",
      "arms.swarm_postdebate.auc",
      "arms.swarm_postdebate.brier",
      "arms.swarm_postdebate.ece",
      "arms.swarm_postdebate.mce",
      "arms.swarm_postdebate.mean_confidence",
      "arms.swarm_postdebate.n",
      "arms.swarm_postdebate.overconfidence",
      "arms.swarm_round0.accuracy",
      "arms.swarm_round0.auc",
      "arms.swarm_round0.brier",
      "arms.swarm_round0.ece",
      "arms.swarm_round0.mce",
      "arms.swarm_round0.mean_confidence",
      "arms.swarm_round0.n",
      "arms.swarm_round0.overconfidence",
      "baseline_full_auc_coarsened_to_3_levels",
      "bootstrap.resamples",
      "bootstrap.seed",
      "deltas_ci95.acc_postdebate_minus_baselinefull.hi",
      "deltas_ci95.acc_postdebate_minus_baselinefull.lo",
      "deltas_ci95.acc_postdebate_minus_baselinefull.mean",
      "deltas_ci95.acc_postdebate_minus_round0.hi",
      "deltas_ci95.acc_postdebate_minus_round0.lo",
      "deltas_ci95.acc_postdebate_minus_round0.mean",
      "deltas_ci95.auc_postdebate_minus_baseline3.hi",
      "deltas_ci95.auc_postdebate_minus_baseline3.lo",
      "deltas_ci95.auc_postdebate_minus_baseline3.mean",
      "deltas_ci95.auc_postdebate_minus_baselinefull.hi",
      "deltas_ci95.auc_postdebate_minus_baselinefull.lo",
      "deltas_ci95.auc_postdebate_minus_baselinefull.mean",
      "deltas_ci95.conf_postdebate_minus_baseline3.hi",
      "deltas_ci95.conf_postdebate_minus_baseline3.lo",
      "deltas_ci95.conf_postdebate_minus_baseline3.mean",
      "deltas_ci95.conf_postdebate_minus_baselinefull.hi",
      "deltas_ci95.conf_postdebate_minus_baselinefull.lo",
      "deltas_ci95.conf_postdebate_minus_baselinefull.mean",
      "deltas_ci95.overconf_postdebate_minus_baseline3.hi",
      "deltas_ci95.overconf_postdebate_minus_baseline3.lo",
      "deltas_ci95.overconf_postdebate_minus_baseline3.mean",
      "deltas_ci95.overconf_postdebate_minus_round0.hi",
      "deltas_ci95.overconf_postdebate_minus_round0.lo",
      "deltas_ci95.overconf_postdebate_minus_round0.mean",
      "per_source_postdebate_vs_baseline.fresh.baseline_full.accuracy",
      "per_source_postdebate_vs_baseline.fresh.baseline_full.overconfidence",
      "per_source_postdebate_vs_baseline.fresh.n",
      "per_source_postdebate_vs_baseline.fresh.swarm_postdebate.accuracy",
      "per_source_postdebate_vs_baseline.fresh.swarm_postdebate.overconfidence",
      "per_source_postdebate_vs_baseline.gsm8k.baseline_full.accuracy",
      "per_source_postdebate_vs_baseline.gsm8k.baseline_full.overconfidence",
      "per_source_postdebate_vs_baseline.gsm8k.n",
      "per_source_postdebate_vs_baseline.gsm8k.swarm_postdebate.accuracy",
      "per_source_postdebate_vs_baseline.gsm8k.swarm_postdebate.overconfidence",
      "per_source_postdebate_vs_baseline.mc.baseline_full.accuracy",
      "per_source_postdebate_vs_baseline.mc.baseline_full.overconfidence",
      "per_source_postdebate_vs_baseline.mc.n",
      "per_source_postdebate_vs_baseline.mc.swarm_postdebate.accuracy",
      "per_source_postdebate_vs_baseline.mc.swarm_postdebate.overconfidence",
      "per_source_postdebate_vs_baseline.mmlu.baseline_full.accuracy",
      "per_source_postdebate_vs_baseline.mmlu.baseline_full.overconfidence",
      "per_source_postdebate_vs_baseline.mmlu.n",
      "per_source_postdebate_vs_baseline.mmlu.swarm_postdebate.accuracy",
      "per_source_postdebate_vs_baseline.mmlu.swarm_postdebate.overconfidence",
      "per_source_postdebate_vs_baseline.priv.baseline_full.accuracy",
      "per_source_postdebate_vs_baseline.priv.baseline_full.overconfidence",
      "per_source_postdebate_vs_baseline.priv.n",
      "per_source_postdebate_vs_baseline.priv.swarm_postdebate.accuracy",
      "per_source_postdebate_vs_baseline.priv.swarm_postdebate.overconfidence",
      "per_source_postdebate_vs_baseline.reason.baseline_full.accuracy",
      "per_source_postdebate_vs_baseline.reason.baseline_full.overconfidence",
      "per_source_postdebate_vs_baseline.reason.n",
      "per_source_postdebate_vs_baseline.reason.swarm_postdebate.accuracy",
      "per_source_postdebate_vs_baseline.reason.swarm_postdebate.overconfidence",
      "per_source_postdebate_vs_baseline.triviaqa.baseline_full.accuracy",
      "per_source_postdebate_vs_baseline.triviaqa.baseline_full.overconfidence",
      "per_source_postdebate_vs_baseline.triviaqa.n",
      "per_source_postdebate_vs_baseline.triviaqa.swarm_postdebate.accuracy",
      "per_source_postdebate_vs_baseline.triviaqa.swarm_postdebate.overconfidence",
      "round0_unanimity.n",
      "round0_unanimity.share",
      "round0_unanimity.unanimous_items"
    ],
    "unchecked_fields": [],
    "validated_configuration_fields": [
      "bootstrap.resamples",
      "bootstrap.seed"
    ]
  },
  "recomputed": {
    "accuracy_discordant_items_swarm_vs_baseline": 2,
    "arms": {
      "baseline_first3": {
        "accuracy": 0.8182,
        "auc": 0.7639,
        "brier": 0.1162,
        "ece": 0.0758,
        "mce": 0.3333,
        "mean_confidence": 0.8939,
        "n": 66,
        "overconfidence": 0.0758
      },
      "baseline_full": {
        "accuracy": 0.8182,
        "auc": 0.8295,
        "brier": 0.1,
        "ece": 0.109,
        "mce": 0.4286,
        "mean_confidence": 0.8702,
        "n": 66,
        "overconfidence": 0.052
      },
      "swarm_postdebate": {
        "accuracy": 0.8182,
        "auc": 0.7407,
        "brier": 0.1279,
        "ece": 0.1111,
        "mce": 0.3333,
        "mean_confidence": 0.9293,
        "n": 66,
        "overconfidence": 0.1111
      },
      "swarm_round0": {
        "accuracy": 0.8485,
        "auc": 0.7812,
        "brier": 0.0993,
        "ece": 0.0555,
        "mce": 0.0833,
        "mean_confidence": 0.8939,
        "n": 66,
        "overconfidence": 0.0455
      }
    },
    "baseline_full_auc_coarsened_to_3_levels": 0.7569,
    "deltas_ci95": {
      "acc_postdebate_minus_baselinefull": {
        "hi": 0.0455,
        "lo": -0.0455,
        "mean": -0.0
      },
      "acc_postdebate_minus_round0": {
        "hi": 0.0,
        "lo": -0.0758,
        "mean": -0.0304
      },
      "auc_postdebate_minus_baseline3": {
        "hi": 0.196,
        "lo": -0.2419,
        "mean": -0.0234
      },
      "auc_postdebate_minus_baselinefull": {
        "hi": 0.08,
        "lo": -0.276,
        "mean": -0.0889
      },
      "conf_postdebate_minus_baseline3": {
        "hi": 0.0859,
        "lo": -0.0151,
        "mean": 0.0357
      },
      "conf_postdebate_minus_baselinefull": {
        "hi": 0.1058,
        "lo": 0.0158,
        "mean": 0.0594
      },
      "overconf_postdebate_minus_baseline3": {
        "hi": 0.0909,
        "lo": -0.0202,
        "mean": 0.0358
      },
      "overconf_postdebate_minus_round0": {
        "hi": 0.1162,
        "lo": 0.0202,
        "mean": 0.0658
      }
    },
    "per_source_postdebate_vs_baseline": {
      "fresh": {
        "baseline_full": {
          "accuracy": 0.0,
          "overconfidence": 0.4465
        },
        "n": 2,
        "swarm_postdebate": {
          "accuracy": 0.0,
          "overconfidence": 0.6667
        }
      },
      "gsm8k": {
        "baseline_full": {
          "accuracy": 0.95,
          "overconfidence": -0.0205
        },
        "n": 20,
        "swarm_postdebate": {
          "accuracy": 0.9,
          "overconfidence": 0.0667
        }
      },
      "mc": {
        "baseline_full": {
          "accuracy": 1.0,
          "overconfidence": -0.2857
        },
        "n": 1,
        "swarm_postdebate": {
          "accuracy": 1.0,
          "overconfidence": -0.3333
        }
      },
      "mmlu": {
        "baseline_full": {
          "accuracy": 0.95,
          "overconfidence": -0.0738
        },
        "n": 20,
        "swarm_postdebate": {
          "accuracy": 1.0,
          "overconfidence": -0.05
        }
      },
      "priv": {
        "baseline_full": {
          "accuracy": 1.0,
          "overconfidence": -0.2857
        },
        "n": 1,
        "swarm_postdebate": {
          "accuracy": 1.0,
          "overconfidence": -0.3333
        }
      },
      "reason": {
        "baseline_full": {
          "accuracy": 1.0,
          "overconfidence": -0.1965
        },
        "n": 2,
        "swarm_postdebate": {
          "accuracy": 1.0,
          "overconfidence": 0.0
        }
      },
      "triviaqa": {
        "baseline_full": {
          "accuracy": 0.6,
          "overconfidence": 0.2696
        },
        "n": 20,
        "swarm_postdebate": {
          "accuracy": 0.6,
          "overconfidence": 0.3167
        }
      }
    },
    "round0_unanimity": {
      "n": 66,
      "share": 0.7424,
      "unanimous_items": 49
    }
  },
  "scope": "Reaggregation of public scored rows; not a raw-event or truth-key audit",
  "status": "partial_precision"
}