{
 "selection": {
  "total": 195,
  "sequential": [
   {
    "criterion": "a component compared with a control (ledger type is a layer comparison)",
    "removed": 35,
    "remaining": 160
   },
   {
    "criterion": "not a structural row (control cannot do the task at all) and not a verdict traced to a fixed harness bug",
    "removed": 4,
    "remaining": 156
   },
   {
    "criterion": "the control is a no-harness control (bare, told-bare, diligence instruction, fresh/alone), not a rival system",
    "removed": 66,
    "remaining": 90
   },
   {
    "criterion": "a bounded metric with a recorded control score (headroom computable on a 0-100 scale)",
    "removed": 60,
    "remaining": 30
   },
   {
    "criterion": "a standard error is recorded",
    "removed": 9,
    "remaining": 21
   },
   {
    "criterion": "the harnessed model is Claude Opus 5.5",
    "removed": 6,
    "remaining": 15
   }
  ],
  "failing_only_one": [
   {
    "criterion": "a component compared with a control (ledger type is a layer comparison)",
    "rows_failing_only_this": [
     "L119",
     "L146"
    ]
   },
   {
    "criterion": "not a structural row (control cannot do the task at all) and not a verdict traced to a fixed harness bug",
    "rows_failing_only_this": [
     "L005",
     "L100"
    ]
   },
   {
    "criterion": "the control is a no-harness control (bare, told-bare, diligence instruction, fresh/alone), not a rival system",
    "rows_failing_only_this": [
     "L052",
     "L053",
     "L056",
     "L112",
     "L125",
     "L135",
     "L142",
     "L143",
     "L148",
     "L157",
     "L158",
     "L164"
    ]
   },
   {
    "criterion": "a bounded metric with a recorded control score (headroom computable on a 0-100 scale)",
    "rows_failing_only_this": [
     "L009",
     "L014",
     "L020",
     "L023",
     "L024",
     "L028",
     "L048",
     "L062",
     "L073",
     "L083",
     "L084",
     "L128",
     "L129",
     "L138",
     "L141",
     "L154",
     "L162",
     "L165",
     "L175",
     "L176",
     "L180",
     "L181",
     "L184",
     "L185",
     "L186",
     "L187",
     "L188",
     "L193",
     "L194",
     "L195"
    ]
   },
   {
    "criterion": "a standard error is recorded",
    "rows_failing_only_this": [
     "L097",
     "L102"
    ]
   },
   {
    "criterion": "the harnessed model is Claude Opus 5.5",
    "rows_failing_only_this": [
     "L001",
     "L002",
     "L006",
     "L007",
     "L054",
     "L058"
    ]
   }
  ]
 },
 "reproduction": {
  "k": 0.539,
  "k_ci": [
   0.427,
   0.675
  ],
  "zero": {
   "aic": 107.1,
   "tau": 7.09,
   "params": []
  },
  "constant": {
   "aic": 96.82,
   "tau": 4.16,
   "params": [
    5.705
   ]
  },
  "headroom": {
   "aic": 77.16,
   "tau": 0.98,
   "params": [
    0.539
   ]
  },
  "weaker_k": 0.383,
  "weaker_tau": 11.45,
  "all21_k": 0.453,
  "published_bf": {
   "lpd_head": -51.95,
   "lpd_zero": -54.52,
   "bf": 13.1,
   "note": "both scored with tau fitted on the weaker rows under the headroom model"
  }
 },
 "models_opus15": {
  "M0": {
   "params": [],
   "tau": 7.085,
   "nll": 52.55,
   "aic": 107.1,
   "bic": 107.81,
   "loo_lpd": -53.211
  },
  "M1": {
   "params": [
    5.7053
   ],
   "tau": 4.16,
   "nll": 46.411,
   "aic": 96.82,
   "bic": 98.24,
   "loo_lpd": -48.3
  },
  "M2": {
   "params": [
    0.5392
   ],
   "tau": 0.98,
   "nll": 36.578,
   "aic": 77.16,
   "bic": 78.57,
   "loo_lpd": -38.58
  },
  "M2b": {
   "params": [
    0.6414,
    1.5926
   ],
   "tau": 0.0,
   "nll": 34.788,
   "aic": 75.58,
   "bic": 77.7,
   "loo_lpd": -41.142
  },
  "M3": {
   "params": [
    7.3286
   ],
   "tau": 2.639,
   "nll": 42.695,
   "aic": 89.39,
   "bic": 90.81,
   "loo_lpd": -45.374
  },
  "M5": {
   "params": [
    2.4956
   ],
   "tau": 1.887,
   "nll": 39.577,
   "aic": 83.15,
   "bic": 84.57,
   "loo_lpd": -42.077
  }
 },
 "models_weaker6": {
  "M0": {
   "params": [],
   "tau": 15.327,
   "nll": 25.242,
   "aic": 52.48,
   "bic": 52.28,
   "loo_lpd": -26.239
  },
  "M1": {
   "params": [
    8.1201
   ],
   "tau": 12.927,
   "nll": 24.339,
   "aic": 52.68,
   "bic": 52.26,
   "loo_lpd": -30.494
  },
  "M2": {
   "params": [
    0.3834
   ],
   "tau": 11.455,
   "nll": 23.696,
   "aic": 51.39,
   "bic": 50.98,
   "loo_lpd": -31.762
  },
  "M2b": {
   "params": [
    0.7663,
    10.7617
   ],
   "tau": 10.882,
   "nll": 23.399,
   "aic": 52.8,
   "bic": 52.17,
   "loo_lpd": -34.87
  },
  "M3": {
   "params": [
    8.1201
   ],
   "tau": 12.927,
   "nll": 24.339,
   "aic": 52.68,
   "bic": 52.26,
   "loo_lpd": -32.262
  },
  "M5": {
   "params": [
    2.4542
   ],
   "tau": 12.61,
   "nll": 24.195,
   "aic": 52.39,
   "bic": 51.97,
   "loo_lpd": -32.969
  }
 },
 "models_all21": {
  "M0": {
   "params": [],
   "tau": 9.667,
   "nll": 79.826,
   "aic": 161.65,
   "bic": 162.7,
   "loo_lpd": -81.0
  },
  "M1": {
   "params": [
    6.3857
   ],
   "tau": 6.733,
   "nll": 74.485,
   "aic": 152.97,
   "bic": 155.06,
   "loo_lpd": -78.206
  },
  "M2": {
   "params": [
    0.4529
   ],
   "tau": 4.411,
   "nll": 69.903,
   "aic": 143.81,
   "bic": 145.89,
   "loo_lpd": -78.883
  },
  "M2b": {
   "params": [
    0.4785,
    0.5041
   ],
   "tau": 4.422,
   "nll": 69.878,
   "aic": 145.76,
   "bic": 148.89,
   "loo_lpd": -80.597
  },
  "M3": {
   "params": [
    7.7262
   ],
   "tau": 5.724,
   "nll": 72.905,
   "aic": 149.81,
   "bic": 151.9,
   "loo_lpd": -78.932
  },
  "M5": {
   "params": [
    2.4729
   ],
   "tau": 4.939,
   "nll": 71.413,
   "aic": 146.83,
   "bic": 148.92,
   "loo_lpd": -80.15
  }
 },
 "cross_weaker_to_opus": {
  "models": {
   "M0": {
    "params": [],
    "tau": 15.327,
    "lpd_own_tau": -57.172,
    "lpd_head_tau": -54.519,
    "lpd_tau_refit": -52.55
   },
   "M1": {
    "params": [
     8.1201
    ],
    "tau": 12.927,
    "lpd_own_tau": -54.229,
    "lpd_head_tau": -52.91,
    "lpd_tau_refit": -47.797
   },
   "M2": {
    "params": [
     0.3834
    ],
    "tau": 11.455,
    "lpd_own_tau": -51.945,
    "lpd_head_tau": -51.945,
    "lpd_tau_refit": -40.273
   },
   "M2b": {
    "params": [
     0.7663,
     10.7617
    ],
    "tau": 10.882,
    "lpd_own_tau": -54.565,
    "lpd_head_tau": -54.888,
    "lpd_tau_refit": -53.586
   },
   "M3": {
    "params": [
     8.1201
    ],
    "tau": 12.927,
    "lpd_own_tau": -53.664,
    "lpd_head_tau": -52.197,
    "lpd_tau_refit": -42.894
   },
   "M5": {
    "params": [
     2.4542
    ],
    "tau": 12.61,
    "lpd_own_tau": -53.061,
    "lpd_head_tau": -51.847,
    "lpd_tau_refit": -39.588
   }
  },
  "bf_m2_vs": {
   "M0": {
    "own_tau": 186.23,
    "shared_tau_w": 13.12,
    "tau_refit": 214700.65
   },
   "M1": {
    "own_tau": 9.82,
    "shared_tau_w": 2.62,
    "tau_refit": 1851.96
   },
   "M2b": {
    "own_tau": 13.74,
    "shared_tau_w": 18.97,
    "tau_refit": 605009.84
   },
   "M3": {
    "own_tau": 5.58,
    "shared_tau_w": 1.29,
    "tau_refit": 13.75
   },
   "M5": {
    "own_tau": 3.05,
    "shared_tau_w": 0.91,
    "tau_refit": 0.5
   }
  }
 },
 "cross_opus_to_weaker": {
  "models": {
   "M0": {
    "params": [],
    "tau": 7.085,
    "lpd_own_tau": -28.663,
    "lpd_head_tau": -28.663,
    "lpd_tau_refit": -25.242
   },
   "M1": {
    "params": [
     5.7053
    ],
    "tau": 4.16,
    "lpd_own_tau": -30.14,
    "lpd_head_tau": -30.14,
    "lpd_tau_refit": -24.431
   },
   "M2": {
    "params": [
     0.5392
    ],
    "tau": 0.98,
    "lpd_own_tau": -37.603,
    "lpd_head_tau": -37.603,
    "lpd_tau_refit": -24.003
   },
   "M2b": {
    "params": [
     0.6414,
     1.5926
    ],
    "tau": 0.0,
    "lpd_own_tau": -40.764,
    "lpd_head_tau": -40.764,
    "lpd_tau_refit": -24.123
   },
   "M3": {
    "params": [
     7.3286
    ],
    "tau": 2.639,
    "lpd_own_tau": -32.943,
    "lpd_head_tau": -32.943,
    "lpd_tau_refit": -24.349
   },
   "M5": {
    "params": [
     2.4956
    ],
    "tau": 1.887,
    "lpd_own_tau": -34.318,
    "lpd_head_tau": -34.318,
    "lpd_tau_refit": -24.196
   }
  },
  "bf_m2_vs": {
   "M0": {
    "own_tau": 0.0,
    "shared_tau_w": 0.0,
    "tau_refit": 3.45
   },
   "M1": {
    "own_tau": 0.0,
    "shared_tau_w": 0.0,
    "tau_refit": 1.53
   },
   "M2b": {
    "own_tau": 23.59,
    "shared_tau_w": 23.59,
    "tau_refit": 1.13
   },
   "M3": {
    "own_tau": 0.01,
    "shared_tau_w": 0.01,
    "tau_refit": 1.41
   },
   "M5": {
    "own_tau": 0.04,
    "shared_tau_w": 0.04,
    "tau_refit": 1.21
   }
  }
 },
 "per_row": [
  {
   "id": "L010",
   "h": 0.0,
   "y": -1.1,
   "s": 0.842,
   "mu_M1": 5.7053,
   "ll_M1": -3.65,
   "mu_M2": 0.0,
   "ll_M2": -1.5376,
   "mu_M3": 0.0,
   "ll_M3": -2.0168,
   "M2-M1": 2.1123,
   "M2-M3": 0.4791
  },
  {
   "id": "L003",
   "h": 2.3,
   "y": 0.0,
   "s": 2.3469,
   "mu_M1": 5.7053,
   "ll_M1": -3.196,
   "mu_M2": 1.2401,
   "ll_M2": -1.9713,
   "mu_M3": 2.3,
   "ll_M3": -2.3928,
   "M2-M1": 1.2247,
   "M2-M3": 0.4215
  },
  {
   "id": "L016",
   "h": 5.0,
   "y": -1.6667,
   "s": 4.2763,
   "mu_M1": 5.7053,
   "ll_M1": -3.4685,
   "mu_M2": 2.6958,
   "ll_M2": -2.892,
   "mu_M3": 5.0,
   "ll_M3": -3.4134,
   "M2-M1": 0.5764,
   "M2-M3": 0.5214
  },
  {
   "id": "L004",
   "h": 8.0,
   "y": 5.75,
   "s": 2.6378,
   "mu_M1": 5.7053,
   "ll_M1": -2.5134,
   "mu_M2": 4.3134,
   "ll_M2": -2.0839,
   "mu_M3": 7.3286,
   "ll_M3": -2.3252,
   "M2-M1": 0.4296,
   "M2-M3": 0.2413
  },
  {
   "id": "L136",
   "h": 9.7,
   "y": 2.68,
   "s": 1.176,
   "mu_M1": 5.7053,
   "ll_M1": -2.6277,
   "mu_M2": 5.2299,
   "ll_M2": -2.7318,
   "mu_M3": 7.3286,
   "ll_M3": -3.2741,
   "M2-M1": -0.1041,
   "M2-M3": 0.5423
  },
  {
   "id": "L011",
   "h": 10.0,
   "y": 0.0,
   "s": 10.607,
   "mu_M1": 5.7053,
   "ll_M1": -3.4774,
   "mu_M2": 5.3917,
   "ll_M2": -3.4128,
   "mu_M3": 7.3286,
   "ll_M3": -3.5353,
   "M2-M1": 0.0646,
   "M2-M3": 0.1225
  },
  {
   "id": "L163",
   "h": 10.1,
   "y": 4.3,
   "s": 1.346,
   "mu_M1": 5.7053,
   "ll_M1": -2.4459,
   "mu_M2": 5.4456,
   "ll_M2": -1.6655,
   "mu_M3": 7.3286,
   "ll_M3": -2.5275,
   "M2-M1": 0.7803,
   "M2-M3": 0.862
  },
  {
   "id": "L012",
   "h": 11.0,
   "y": 1.4,
   "s": 4.031,
   "mu_M1": 5.7053,
   "ll_M1": -2.9517,
   "mu_M2": 5.9309,
   "ll_M2": -2.9381,
   "mu_M3": 7.3286,
   "ll_M3": -3.2483,
   "M2-M1": 0.0136,
   "M2-M3": 0.3102
  },
  {
   "id": "L099",
   "h": 11.0,
   "y": 10.8,
   "s": 4.4643,
   "mu_M1": 5.7053,
   "ll_M1": -3.0761,
   "mu_M2": 5.9309,
   "ll_M2": -3.006,
   "mu_M3": 7.3286,
   "ll_M3": -2.789,
   "M2-M1": 0.0701,
   "M2-M3": -0.2171
  },
  {
   "id": "L147",
   "h": 13.1,
   "y": 9.3,
   "s": 3.6224,
   "mu_M1": 5.7053,
   "ll_M1": -2.8389,
   "mu_M2": 7.0631,
   "ll_M2": -2.4191,
   "mu_M3": 7.3286,
   "ll_M3": -2.5157,
   "M2-M1": 0.4199,
   "M2-M3": 0.0967
  },
  {
   "id": "L174",
   "h": 15.1,
   "y": 6.7,
   "s": 1.7109,
   "mu_M1": 5.7053,
   "ll_M1": -2.447,
   "mu_M2": 8.1415,
   "ll_M2": -1.8651,
   "mu_M3": 7.3286,
   "ll_M3": -2.0848,
   "M2-M1": 0.5819,
   "M2-M3": 0.2197
  },
  {
   "id": "L126",
   "h": 15.6,
   "y": 10.8,
   "s": 2.1939,
   "mu_M1": 5.7053,
   "ll_M1": -3.0539,
   "mu_M2": 8.411,
   "ll_M2": -2.2898,
   "mu_M3": 7.3286,
   "ll_M3": -2.6636,
   "M2-M1": 0.7641,
   "M2-M3": 0.3738
  },
  {
   "id": "L124",
   "h": 16.7,
   "y": 11.9,
   "s": 2.7296,
   "mu_M1": 5.7053,
   "ll_M1": -3.2985,
   "mu_M2": 9.0041,
   "ll_M2": -2.4822,
   "mu_M3": 7.3286,
   "ll_M3": -2.9779,
   "M2-M1": 0.8163,
   "M2-M3": 0.4957
  },
  {
   "id": "L166",
   "h": 19.3,
   "y": 13.2,
   "s": 2.7769,
   "mu_M1": 5.7053,
   "ll_M1": -3.6514,
   "mu_M2": 10.406,
   "ll_M2": -2.4491,
   "mu_M3": 7.3286,
   "ll_M3": -3.4365,
   "M2-M1": 1.2023,
   "M2-M3": 0.9874
  },
  {
   "id": "L057",
   "h": 25.8,
   "y": 15.2,
   "s": 6.5851,
   "mu_M1": 5.7053,
   "ll_M1": -3.7146,
   "mu_M2": 13.9106,
   "ll_M2": -2.8335,
   "mu_M3": 7.3286,
   "ll_M3": -3.4938,
   "M2-M1": 0.8812,
   "M2-M3": 0.6603
  }
 ],
 "sensitivity_se_only_rows": {
  "imputed_s_0.5": {
   "rows": [
    "L097",
    "L102"
   ],
   "M1": {
    "params": [
     4.775
    ],
    "tau": 4.308,
    "aic": 109.29
   },
   "M2": {
    "params": [
     0.518
    ],
    "tau": 0.0,
    "aic": 78.88
   },
   "M3": {
    "params": [
     7.094
    ],
    "tau": 2.012,
    "aic": 96.53
   }
  },
  "imputed_s_1.0": {
   "rows": [
    "L097",
    "L102"
   ],
   "M1": {
    "params": [
     4.802
    ],
    "tau": 4.286,
    "aic": 109.28
   },
   "M2": {
    "params": [
     0.522
    ],
    "tau": 0.369,
    "aic": 81.61
   },
   "M3": {
    "params": [
     7.134
    ],
    "tau": 2.102,
    "aic": 96.83
   }
  },
  "imputed_s_2.0": {
   "rows": [
    "L097",
    "L102"
   ],
   "M1": {
    "params": [
     4.898
    ],
    "tau": 4.214,
    "aic": 109.23
   },
   "M2": {
    "params": [
     0.533
    ],
    "tau": 0.775,
    "aic": 83.95
   },
   "M3": {
    "params": [
     7.203
    ],
    "tau": 2.273,
    "aic": 97.67
   }
  }
 },
 "D1_profile_ci": {
  "opus15": {
   "n": 15,
   "k": 0.539,
   "k_ci95": [
    0.427,
    0.675
   ],
   "tau": 0.98,
   "grid": [
    -0.5,
    1.5
   ],
   "interval_contiguous": true
  },
  "weaker6": {
   "n": 6,
   "k": 0.383,
   "k_ci95": [
    -0.058,
    0.837
   ],
   "tau": 11.455,
   "grid": [
    -0.5,
    1.5
   ],
   "interval_contiguous": true
  },
  "all21": {
   "n": 21,
   "k": 0.453,
   "k_ci95": [
    0.299,
    0.609
   ],
   "tau": 4.411,
   "grid": [
    -0.5,
    1.5
   ],
   "interval_contiguous": true
  },
  "opus15_matches_reproduction": true,
  "method": "profile likelihood of M2 (mu = k h) with log tau profiled out by Nelder-Mead at each grid value of k; the 95% set is {k : nll_profile(k) - nll_min <= 1.92}; identical to profile_k in part 1, grid widened only if the set touches the grid edge"
 },
 "D2_forecast_weaker_to_opus": {
  "summary": {
   "M0": {
    "name": "zero",
    "params_src": [],
    "tau_src": 15.327,
    "mae": 6.32,
    "rmse": 8.067,
    "mean_abs_err_over_s": 2.33,
    "mean_abs_err_over_sqrt_s2_tau2": 0.401,
    "mean_err_bias": 5.951,
    "max_abs_err": 15.2
   },
   "M1": {
    "name": "constant",
    "params_src": [
     8.1201
    ],
    "tau_src": 12.927,
    "mae": 5.166,
    "rmse": 5.863,
    "mean_abs_err_over_s": 2.317,
    "mean_abs_err_over_sqrt_s2_tau2": 0.379,
    "mean_err_bias": -2.169,
    "max_abs_err": 9.787
   },
   "M2": {
    "name": "headroom",
    "params_src": [
     0.3834
    ],
    "tau_src": 11.455,
    "mae": 3.304,
    "rmse": 3.862,
    "mean_abs_err_over_s": 1.073,
    "mean_abs_err_over_sqrt_s2_tau2": 0.269,
    "mean_err_bias": 1.537,
    "max_abs_err": 6.583
   },
   "M2b": {
    "name": "headroom+shift",
    "params_src": [
     0.7663,
     10.7617
    ],
    "tau_src": 10.882,
    "mae": 7.889,
    "rmse": 8.337,
    "mean_abs_err_over_s": 3.575,
    "mean_abs_err_over_sqrt_s2_tau2": 0.692,
    "mean_err_bias": 7.889,
    "max_abs_err": 13.132
   },
   "M3": {
    "name": "capped constant",
    "params_src": [
     8.1201
    ],
    "tau_src": 12.927,
    "mae": 4.021,
    "rmse": 4.614,
    "mean_abs_err_over_s": 1.457,
    "mean_abs_err_over_sqrt_s2_tau2": 0.291,
    "mean_err_bias": -1.024,
    "max_abs_err": 8.12
   }
  },
  "rowwise_best": {
   "candidates": [
    "M0",
    "M1",
    "M2",
    "M3"
   ],
   "strictly_best_rows": {
    "M0": 4,
    "M1": 0,
    "M2": 4,
    "M3": 6
   },
   "tied_best_rows": {
    "M0": 1,
    "M1": 0,
    "M2": 1,
    "M3": 1
   },
   "n_rows_with_tie": 1
  },
  "m2_vs_other_rowwise": {
   "M0": {
    "m2_closer": 10,
    "other_closer": 4,
    "tie": 1
   },
   "M1": {
    "m2_closer": 9,
    "other_closer": 6,
    "tie": 0
   },
   "M2b": {
    "m2_closer": 14,
    "other_closer": 1,
    "tie": 0
   },
   "M3": {
    "m2_closer": 8,
    "other_closer": 6,
    "tie": 1
   }
  },
  "per_row": [
   {
    "id": "L003",
    "model": "opus-5.5",
    "h": 2.3,
    "y": 0.0,
    "s": 2.3469,
    "mu_M0": 0.0,
    "err_M0": 0.0,
    "mu_M1": 8.1201,
    "err_M1": -8.1201,
    "mu_M2": 0.8817,
    "err_M2": -0.8817,
    "mu_M2b": -8.9991,
    "err_M2b": 8.9991,
    "mu_M3": 2.3,
    "err_M3": -2.3,
    "best_of_M0_M1_M2_M3": "M0"
   },
   {
    "id": "L004",
    "model": "opus-5.5",
    "h": 8.0,
    "y": 5.75,
    "s": 2.6378,
    "mu_M0": 0.0,
    "err_M0": 5.75,
    "mu_M1": 8.1201,
    "err_M1": -2.3701,
    "mu_M2": 3.0669,
    "err_M2": 2.6831,
    "mu_M2b": -4.6309,
    "err_M2b": 10.3809,
    "mu_M3": 8.0,
    "err_M3": -2.25,
    "best_of_M0_M1_M2_M3": "M3"
   },
   {
    "id": "L010",
    "model": "opus-5.5",
    "h": 0.0,
    "y": -1.1,
    "s": 0.842,
    "mu_M0": 0.0,
    "err_M0": -1.1,
    "mu_M1": 8.1201,
    "err_M1": -9.2201,
    "mu_M2": 0.0,
    "err_M2": -1.1,
    "mu_M2b": -10.7617,
    "err_M2b": 9.6617,
    "mu_M3": 0.0,
    "err_M3": -1.1,
    "best_of_M0_M1_M2_M3": "M0/M2/M3"
   },
   {
    "id": "L011",
    "model": "opus-5.5",
    "h": 10.0,
    "y": 0.0,
    "s": 10.607,
    "mu_M0": 0.0,
    "err_M0": 0.0,
    "mu_M1": 8.1201,
    "err_M1": -8.1201,
    "mu_M2": 3.8336,
    "err_M2": -3.8336,
    "mu_M2b": -3.0983,
    "err_M2b": 3.0983,
    "mu_M3": 8.1201,
    "err_M3": -8.1201,
    "best_of_M0_M1_M2_M3": "M0"
   },
   {
    "id": "L012",
    "model": "opus-5.5",
    "h": 11.0,
    "y": 1.4,
    "s": 4.031,
    "mu_M0": 0.0,
    "err_M0": 1.4,
    "mu_M1": 8.1201,
    "err_M1": -6.7201,
    "mu_M2": 4.217,
    "err_M2": -2.817,
    "mu_M2b": -2.3319,
    "err_M2b": 3.7319,
    "mu_M3": 8.1201,
    "err_M3": -6.7201,
    "best_of_M0_M1_M2_M3": "M0"
   },
   {
    "id": "L016",
    "model": "opus-5.5",
    "h": 5.0,
    "y": -1.6667,
    "s": 4.2763,
    "mu_M0": 0.0,
    "err_M0": -1.6667,
    "mu_M1": 8.1201,
    "err_M1": -9.7868,
    "mu_M2": 1.9168,
    "err_M2": -3.5835,
    "mu_M2b": -6.93,
    "err_M2b": 5.2633,
    "mu_M3": 5.0,
    "err_M3": -6.6667,
    "best_of_M0_M1_M2_M3": "M0"
   },
   {
    "id": "L057",
    "model": "opus-5.5",
    "h": 25.8,
    "y": 15.2,
    "s": 6.5851,
    "mu_M0": 0.0,
    "err_M0": 15.2,
    "mu_M1": 8.1201,
    "err_M1": 7.0799,
    "mu_M2": 9.8907,
    "err_M2": 5.3093,
    "mu_M2b": 9.0099,
    "err_M2b": 6.1901,
    "mu_M3": 8.1201,
    "err_M3": 7.0799,
    "best_of_M0_M1_M2_M3": "M2"
   },
   {
    "id": "L099",
    "model": "opus-5.5",
    "h": 11.0,
    "y": 10.8,
    "s": 4.4643,
    "mu_M0": 0.0,
    "err_M0": 10.8,
    "mu_M1": 8.1201,
    "err_M1": 2.6799,
    "mu_M2": 4.217,
    "err_M2": 6.583,
    "mu_M2b": -2.3319,
    "err_M2b": 13.1319,
    "mu_M3": 8.1201,
    "err_M3": 2.6799,
    "best_of_M0_M1_M2_M3": "M3"
   },
   {
    "id": "L124",
    "model": "opus-5.5",
    "h": 16.7,
    "y": 11.9,
    "s": 2.7296,
    "mu_M0": 0.0,
    "err_M0": 11.9,
    "mu_M1": 8.1201,
    "err_M1": 3.7799,
    "mu_M2": 6.4021,
    "err_M2": 5.4979,
    "mu_M2b": 2.0362,
    "err_M2b": 9.8638,
    "mu_M3": 8.1201,
    "err_M3": 3.7799,
    "best_of_M0_M1_M2_M3": "M3"
   },
   {
    "id": "L126",
    "model": "opus-5.5",
    "h": 15.6,
    "y": 10.8,
    "s": 2.1939,
    "mu_M0": 0.0,
    "err_M0": 10.8,
    "mu_M1": 8.1201,
    "err_M1": 2.6799,
    "mu_M2": 5.9804,
    "err_M2": 4.8196,
    "mu_M2b": 1.1933,
    "err_M2b": 9.6067,
    "mu_M3": 8.1201,
    "err_M3": 2.6799,
    "best_of_M0_M1_M2_M3": "M3"
   },
   {
    "id": "L136",
    "model": "opus-5.5",
    "h": 9.7,
    "y": 2.68,
    "s": 1.176,
    "mu_M0": 0.0,
    "err_M0": 2.68,
    "mu_M1": 8.1201,
    "err_M1": -5.4401,
    "mu_M2": 3.7186,
    "err_M2": -1.0386,
    "mu_M2b": -3.3282,
    "err_M2b": 6.0082,
    "mu_M3": 8.1201,
    "err_M3": -5.4401,
    "best_of_M0_M1_M2_M3": "M2"
   },
   {
    "id": "L147",
    "model": "opus-5.5",
    "h": 13.1,
    "y": 9.3,
    "s": 3.6224,
    "mu_M0": 0.0,
    "err_M0": 9.3,
    "mu_M1": 8.1201,
    "err_M1": 1.1799,
    "mu_M2": 5.022,
    "err_M2": 4.278,
    "mu_M2b": -0.7226,
    "err_M2b": 10.0226,
    "mu_M3": 8.1201,
    "err_M3": 1.1799,
    "best_of_M0_M1_M2_M3": "M3"
   },
   {
    "id": "L163",
    "model": "opus-5.5",
    "h": 10.1,
    "y": 4.3,
    "s": 1.346,
    "mu_M0": 0.0,
    "err_M0": 4.3,
    "mu_M1": 8.1201,
    "err_M1": -3.8201,
    "mu_M2": 3.8719,
    "err_M2": 0.4281,
    "mu_M2b": -3.0216,
    "err_M2b": 7.3216,
    "mu_M3": 8.1201,
    "err_M3": -3.8201,
    "best_of_M0_M1_M2_M3": "M2"
   },
   {
    "id": "L166",
    "model": "opus-5.5",
    "h": 19.3,
    "y": 13.2,
    "s": 2.7769,
    "mu_M0": 0.0,
    "err_M0": 13.2,
    "mu_M1": 8.1201,
    "err_M1": 5.0799,
    "mu_M2": 7.3988,
    "err_M2": 5.8012,
    "mu_M2b": 4.0287,
    "err_M2b": 9.1713,
    "mu_M3": 8.1201,
    "err_M3": 5.0799,
    "best_of_M0_M1_M2_M3": "M3"
   },
   {
    "id": "L174",
    "model": "opus-5.5",
    "h": 15.1,
    "y": 6.7,
    "s": 1.7109,
    "mu_M0": 0.0,
    "err_M0": 6.7,
    "mu_M1": 8.1201,
    "err_M1": -1.4201,
    "mu_M2": 5.7887,
    "err_M2": 0.9113,
    "mu_M2b": 0.8101,
    "err_M2b": 5.8899,
    "mu_M3": 8.1201,
    "err_M3": -1.4201,
    "best_of_M0_M1_M2_M3": "M2"
   }
  ]
 },
 "D3_forecast_opus_to_weaker": {
  "summary": {
   "M0": {
    "name": "zero",
    "params_src": [],
    "tau_src": 7.085,
    "mae": 14.312,
    "rmse": 16.486,
    "mean_abs_err_over_s": 2.967,
    "mean_abs_err_over_sqrt_s2_tau2": 1.648,
    "mean_err_bias": 8.345,
    "max_abs_err": 29.23
   },
   "M1": {
    "name": "constant",
    "params_src": [
     5.7053
    ],
    "tau_src": 4.16,
    "mae": 10.508,
    "rmse": 14.461,
    "mean_abs_err_over_s": 2.049,
    "mean_abs_err_over_sqrt_s2_tau2": 1.577,
    "mean_err_bias": 2.64,
    "max_abs_err": 23.605
   },
   "M2": {
    "name": "headroom",
    "params_src": [
     0.5392
    ],
    "tau_src": 0.98,
    "mae": 9.277,
    "rmse": 13.302,
    "mean_abs_err_over_s": 2.044,
    "mean_abs_err_over_sqrt_s2_tau2": 1.991,
    "mean_err_bias": -4.981,
    "max_abs_err": 29.438
   },
   "M2b": {
    "name": "headroom+shift",
    "params_src": [
     0.6414,
     1.5926
    ],
    "tau_src": 0.0,
    "mae": 9.296,
    "rmse": 13.535,
    "mean_abs_err_over_s": 2.106,
    "mean_abs_err_over_sqrt_s2_tau2": 2.106,
    "mean_err_bias": -5.915,
    "max_abs_err": 30.033
   },
   "M3": {
    "name": "capped constant",
    "params_src": [
     7.3286
    ],
    "tau_src": 2.639,
    "mae": 10.112,
    "rmse": 14.254,
    "mean_abs_err_over_s": 1.955,
    "mean_abs_err_over_sqrt_s2_tau2": 1.733,
    "mean_err_bias": 1.016,
    "max_abs_err": 25.229
   }
  },
  "rowwise_best": {
   "candidates": [
    "M0",
    "M1",
    "M2",
    "M3"
   ],
   "strictly_best_rows": {
    "M0": 1,
    "M1": 2,
    "M2": 2,
    "M3": 1
   },
   "tied_best_rows": {
    "M0": 0,
    "M1": 0,
    "M2": 0,
    "M3": 0
   },
   "n_rows_with_tie": 0
  },
  "m2_vs_other_rowwise": {
   "M0": {
    "m2_closer": 4,
    "other_closer": 2,
    "tie": 0
   },
   "M1": {
    "m2_closer": 2,
    "other_closer": 4,
    "tie": 0
   },
   "M2b": {
    "m2_closer": 4,
    "other_closer": 2,
    "tie": 0
   },
   "M3": {
    "m2_closer": 3,
    "other_closer": 3,
    "tie": 0
   }
  },
  "per_row": [
   {
    "id": "L001",
    "model": "haiku",
    "h": 33.4,
    "y": 8.9,
    "s": 3.1257,
    "mu_M0": 0.0,
    "err_M0": 8.9,
    "mu_M1": 5.7053,
    "err_M1": 3.1947,
    "mu_M2": 18.0083,
    "err_M2": -9.1083,
    "mu_M2b": 19.829,
    "err_M2b": -10.929,
    "mu_M3": 7.3286,
    "err_M3": 1.5714,
    "best_of_M0_M1_M2_M3": "M3"
   },
   {
    "id": "L002",
    "model": "haiku",
    "h": 19.9,
    "y": 6.5,
    "s": 4.0491,
    "mu_M0": 0.0,
    "err_M0": 6.5,
    "mu_M1": 5.7053,
    "err_M1": 0.7947,
    "mu_M2": 10.7295,
    "err_M2": -4.2295,
    "mu_M2b": 11.1706,
    "err_M2b": -4.6706,
    "mu_M3": 7.3286,
    "err_M3": -0.8286,
    "best_of_M0_M1_M2_M3": "M1"
   },
   {
    "id": "L006",
    "model": "haiku",
    "h": 36.9,
    "y": 29.23,
    "s": 5.8878,
    "mu_M0": 0.0,
    "err_M0": 29.23,
    "mu_M1": 5.7053,
    "err_M1": 23.5247,
    "mu_M2": 19.8954,
    "err_M2": 9.3346,
    "mu_M2b": 22.0738,
    "err_M2b": 7.1562,
    "mu_M3": 7.3286,
    "err_M3": 21.9014,
    "best_of_M0_M1_M2_M3": "M2"
   },
   {
    "id": "L007",
    "model": "haiku",
    "h": 27.6,
    "y": 17.24,
    "s": 5.2781,
    "mu_M0": 0.0,
    "err_M0": 17.24,
    "mu_M1": 5.7053,
    "err_M1": 11.5347,
    "mu_M2": 14.8811,
    "err_M2": 2.3589,
    "mu_M2b": 16.1091,
    "err_M2b": 1.1309,
    "mu_M3": 7.3286,
    "err_M3": 9.9114,
    "best_of_M0_M1_M2_M3": "M2"
   },
   {
    "id": "L054",
    "model": "opus-4.8",
    "h": 9.1,
    "y": 6.1,
    "s": 4.1161,
    "mu_M0": 0.0,
    "err_M0": 6.1,
    "mu_M1": 5.7053,
    "err_M1": 0.3947,
    "mu_M2": 4.9064,
    "err_M2": 1.1936,
    "mu_M2b": 4.2438,
    "err_M2b": 1.8562,
    "mu_M3": 7.3286,
    "err_M3": -1.2286,
    "best_of_M0_M1_M2_M3": "M1"
   },
   {
    "id": "L058",
    "model": "haiku",
    "h": 21.4,
    "y": -17.9,
    "s": 4.9205,
    "mu_M0": 0.0,
    "err_M0": -17.9,
    "mu_M1": 5.7053,
    "err_M1": -23.6053,
    "mu_M2": 11.5382,
    "err_M2": -29.4382,
    "mu_M2b": 12.1326,
    "err_M2b": -30.0326,
    "mu_M3": 7.3286,
    "err_M3": -25.2286,
    "best_of_M0_M1_M2_M3": "M0"
   }
  ]
 },
 "D4_loo_opus15": {
  "summary": {
   "M0": {
    "name": "zero",
    "loo_mae": 6.32,
    "loo_rmse": 8.067,
    "loo_mean_abs_err_over_s": 2.33,
    "insample_mae": 6.32,
    "insample_rmse": 8.067,
    "loo_lpd": -53.211
   },
   "M1": {
    "name": "constant",
    "loo_mae": 5.142,
    "loo_rmse": 5.799,
    "loo_mean_abs_err_over_s": 2.089,
    "insample_mae": 4.822,
    "insample_rmse": 5.452,
    "loo_lpd": -48.3
   },
   "M2": {
    "name": "headroom",
    "loo_mae": 2.836,
    "loo_rmse": 3.173,
    "loo_mean_abs_err_over_s": 1.02,
    "insample_mae": 2.645,
    "insample_rmse": 3.007,
    "loo_lpd": -38.58
   },
   "M2b": {
    "name": "headroom+shift",
    "loo_mae": 2.616,
    "loo_rmse": 2.998,
    "loo_mean_abs_err_over_s": 1.023,
    "insample_mae": 2.305,
    "insample_rmse": 2.778,
    "loo_lpd": -41.142
   },
   "M3": {
    "name": "capped constant",
    "loo_mae": 4.329,
    "loo_rmse": 4.905,
    "loo_mean_abs_err_over_s": 1.554,
    "insample_mae": 4.029,
    "insample_rmse": 4.61,
    "loo_lpd": -45.374
   }
  },
  "rowwise_strictly_best": {
   "M0": 5,
   "M1": 1,
   "M2": 5,
   "M3": 3
  },
  "per_row": [
   {
    "id": "L003",
    "h": 2.3,
    "y": 0.0,
    "s": 2.3469,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 0.0,
    "loo_mu_M1": 6.1751,
    "loo_err_M1": -6.1751,
    "loo_mu_M2": 1.2491,
    "loo_err_M2": -1.2491,
    "loo_mu_M2b": -0.1268,
    "loo_err_M2b": 0.1268,
    "loo_mu_M3": 2.3,
    "loo_err_M3": -2.3,
    "loo_best_of_M0_M1_M2_M3": "M0"
   },
   {
    "id": "L004",
    "h": 8.0,
    "y": 5.75,
    "s": 2.6378,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 5.75,
    "loo_mu_M1": 5.7323,
    "loo_err_M1": 0.0177,
    "loo_mu_M2": 4.2813,
    "loo_err_M2": 1.4687,
    "loo_mu_M2b": 3.4573,
    "loo_err_M2b": 2.2927,
    "loo_mu_M3": 7.5418,
    "loo_err_M3": -1.7918,
    "loo_best_of_M0_M1_M2_M3": "M1"
   },
   {
    "id": "L010",
    "h": 0.0,
    "y": -1.1,
    "s": 0.842,
    "loo_mu_M0": 0.0,
    "loo_err_M0": -1.1,
    "loo_mu_M1": 6.4017,
    "loo_err_M1": -7.5017,
    "loo_mu_M2": 0.0,
    "loo_err_M2": -1.1,
    "loo_mu_M2b": -3.7447,
    "loo_err_M2b": 2.6447,
    "loo_mu_M3": 0.0,
    "loo_err_M3": -1.1,
    "loo_best_of_M0_M1_M2_M3": "M0/M2/M3"
   },
   {
    "id": "L011",
    "h": 10.0,
    "y": 0.0,
    "s": 10.607,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 0.0,
    "loo_mu_M1": 5.7904,
    "loo_err_M1": -5.7904,
    "loo_mu_M2": 5.4112,
    "loo_err_M2": -5.4112,
    "loo_mu_M2b": 4.8332,
    "loo_err_M2b": -4.8332,
    "loo_mu_M3": 7.4164,
    "loo_err_M3": -7.4164,
    "loo_best_of_M0_M1_M2_M3": "M0"
   },
   {
    "id": "L012",
    "h": 11.0,
    "y": 1.4,
    "s": 4.031,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 1.4,
    "loo_mu_M1": 5.9608,
    "loo_err_M1": -4.5608,
    "loo_mu_M2": 6.0558,
    "loo_err_M2": -4.6558,
    "loo_mu_M2b": 5.5441,
    "loo_err_M2b": -4.1441,
    "loo_mu_M3": 7.6791,
    "loo_err_M3": -6.2791,
    "loo_best_of_M0_M1_M2_M3": "M0"
   },
   {
    "id": "L016",
    "h": 5.0,
    "y": -1.6667,
    "s": 4.2763,
    "loo_mu_M0": 0.0,
    "loo_err_M0": -1.6667,
    "loo_mu_M1": 6.0839,
    "loo_err_M1": -7.7506,
    "loo_mu_M2": 2.7179,
    "loo_err_M2": -4.3845,
    "loo_mu_M2b": 1.6652,
    "loo_err_M2b": -3.3318,
    "loo_mu_M3": 5.0,
    "loo_err_M3": -6.6667,
    "loo_best_of_M0_M1_M2_M3": "M0"
   },
   {
    "id": "L057",
    "h": 25.8,
    "y": 15.2,
    "s": 6.5851,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 15.2,
    "loo_mu_M1": 5.4138,
    "loo_err_M1": 9.7862,
    "loo_mu_M2": 13.8425,
    "loo_err_M2": 1.3575,
    "loo_mu_M2b": 14.9411,
    "loo_err_M2b": 0.2589,
    "loo_mu_M3": 7.1127,
    "loo_err_M3": 8.0873,
    "loo_best_of_M0_M1_M2_M3": "M2"
   },
   {
    "id": "L099",
    "h": 11.0,
    "y": 10.8,
    "s": 4.4643,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 10.8,
    "loo_mu_M1": 5.4519,
    "loo_err_M1": 5.3481,
    "loo_mu_M2": 5.8208,
    "loo_err_M2": 4.9792,
    "loo_mu_M2b": 5.3752,
    "loo_err_M2b": 5.4248,
    "loo_mu_M3": 7.1599,
    "loo_err_M3": 3.6401,
    "loo_best_of_M0_M1_M2_M3": "M3"
   },
   {
    "id": "L124",
    "h": 16.7,
    "y": 11.9,
    "s": 2.7296,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 11.9,
    "loo_mu_M1": 5.197,
    "loo_err_M1": 6.703,
    "loo_mu_M2": 8.5857,
    "loo_err_M2": 3.3143,
    "loo_mu_M2b": 8.8021,
    "loo_err_M2b": 3.0979,
    "loo_mu_M3": 6.8218,
    "loo_err_M3": 5.0782,
    "loo_best_of_M0_M1_M2_M3": "M2"
   },
   {
    "id": "L126",
    "h": 15.6,
    "y": 10.8,
    "s": 2.1939,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 10.8,
    "loo_mu_M1": 5.2403,
    "loo_err_M1": 5.5597,
    "loo_mu_M2": 7.9385,
    "loo_err_M2": 2.8615,
    "loo_mu_M2b": 8.0433,
    "loo_err_M2b": 2.7567,
    "loo_mu_M3": 6.8582,
    "loo_err_M3": 3.9418,
    "loo_best_of_M0_M1_M2_M3": "M2"
   },
   {
    "id": "L136",
    "h": 9.7,
    "y": 2.68,
    "s": 1.176,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 2.68,
    "loo_mu_M1": 6.0409,
    "loo_err_M1": -3.3609,
    "loo_mu_M2": 5.542,
    "loo_err_M2": -2.862,
    "loo_mu_M2b": 5.1118,
    "loo_err_M2b": -2.4318,
    "loo_mu_M3": 8.3183,
    "loo_err_M3": -5.6383,
    "loo_best_of_M0_M1_M2_M3": "M0"
   },
   {
    "id": "L147",
    "h": 13.1,
    "y": 9.3,
    "s": 3.6224,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 9.3,
    "loo_mu_M1": 5.492,
    "loo_err_M1": 3.808,
    "loo_mu_M2": 6.9651,
    "loo_err_M2": 2.3349,
    "loo_mu_M2b": 6.7235,
    "loo_err_M2b": 2.5765,
    "loo_mu_M3": 7.2153,
    "loo_err_M3": 2.0847,
    "loo_best_of_M0_M1_M2_M3": "M3"
   },
   {
    "id": "L163",
    "h": 10.1,
    "y": 4.3,
    "s": 1.346,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 4.3,
    "loo_mu_M1": 5.8821,
    "loo_err_M1": -1.5821,
    "loo_mu_M2": 5.6539,
    "loo_err_M2": -1.3539,
    "loo_mu_M2b": 5.0062,
    "loo_err_M2b": -0.7062,
    "loo_mu_M3": 7.8497,
    "loo_err_M3": -3.5497,
    "loo_best_of_M0_M1_M2_M3": "M2"
   },
   {
    "id": "L166",
    "h": 19.3,
    "y": 13.2,
    "s": 2.7769,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 13.2,
    "loo_mu_M1": 5.0757,
    "loo_err_M1": 8.1243,
    "loo_mu_M2": 9.8661,
    "loo_err_M2": 3.3339,
    "loo_mu_M2b": 10.388,
    "loo_err_M2b": 2.812,
    "loo_mu_M3": 6.6427,
    "loo_err_M3": 6.5573,
    "loo_best_of_M0_M1_M2_M3": "M2"
   },
   {
    "id": "L174",
    "h": 15.1,
    "y": 6.7,
    "s": 1.7109,
    "loo_mu_M0": 0.0,
    "loo_err_M0": 6.7,
    "loo_mu_M1": 5.6449,
    "loo_err_M1": 1.0551,
    "loo_mu_M2": 8.57,
    "loo_err_M2": -1.87,
    "loo_mu_M2b": 8.4958,
    "loo_err_M2b": -1.7958,
    "loo_mu_M3": 7.4972,
    "loo_err_M3": -0.7972,
    "loo_best_of_M0_M1_M2_M3": "M3"
   }
  ]
 },
 "D5_rows": {
  "opus15": [
   {
    "id": "L010",
    "model": "opus-5.5",
    "team": "longrun",
    "outcome": "O3",
    "component": "living",
    "contrast": "living - bare (AppWorld probes H0-H41, item-paired)",
    "ctype": "carried_w2s",
    "instrument": "hourly probes AppWorld (H0-H24)",
    "control_score": 100.0,
    "harnessed_score": 99.0,
    "h": 0.0,
    "delta": -1.1,
    "s": 0.842,
    "r_delta_over_h": null,
    "ci_lo": -3.3,
    "ci_hi": 0.0,
    "p": 0.71,
    "n": 30,
    "se_method": "ci",
    "verdict": "null",
    "prereg": "Y",
    "flags": "ceiling,prereg",
    "source": "probe record H000-H041 (ci_fill)",
    "note": "ledger -0.001 was the pooled headroom lift, not AppWorld; 0 vs 4 discordant",
    "metric_scale_max": 1.0
   },
   {
    "id": "L003",
    "model": "opus-5.5",
    "team": "evolved engine",
    "outcome": "O2",
    "component": "lineage",
    "contrast": "lineage - bare_notice",
    "ctype": "layer_vs_prior",
    "instrument": "AppWorld challenge k=11 single crash survival",
    "control_score": 97.7,
    "harnessed_score": 97.7,
    "h": 2.3,
    "delta": 0.0,
    "s": 2.3469,
    "r_delta_over_h": 0.0,
    "ci_lo": -4.6,
    "ci_hi": 4.6,
    "p": 1.0,
    "n": 87,
    "se_method": "ci",
    "verdict": "neg",
    "prereg": "Y",
    "flags": "prereg,ceiling",
    "source": "evolved engine preregistered boundary run record",
    "note": "K1 fired: a notice is enough",
    "metric_scale_max": 100.0
   },
   {
    "id": "L016",
    "model": "opus-5.5",
    "team": "longrun",
    "outcome": "O3",
    "component": "living",
    "contrast": "living - bare",
    "ctype": "carried_w2s",
    "instrument": "fresh AppWorld offline (60)",
    "control_score": 95.0,
    "harnessed_score": 93.3333,
    "h": 5.0,
    "delta": -1.6667,
    "s": 4.2763,
    "r_delta_over_h": -0.3333,
    "ci_lo": null,
    "ci_hi": null,
    "p": null,
    "n": 60,
    "se_method": "prop",
    "verdict": "null",
    "prereg": "?",
    "flags": "",
    "source": "program record O3 H24",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L004",
    "model": "opus-5.5",
    "team": "evolved engine",
    "outcome": "O2",
    "component": "lineage",
    "contrast": "lineage - bare",
    "ctype": "layer_vs_bare",
    "instrument": "AppWorld challenge k=11 single crash survival",
    "control_score": 92.0,
    "harnessed_score": 97.7,
    "h": 8.0,
    "delta": 5.75,
    "s": 2.6378,
    "r_delta_over_h": 0.7188,
    "ci_lo": 1.15,
    "ci_hi": 11.49,
    "p": 0.06,
    "n": 87,
    "se_method": "ci",
    "verdict": "pos",
    "prereg": "Y",
    "flags": "prereg",
    "source": "same",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L136",
    "model": "opus-5.5",
    "team": "companion",
    "outcome": "O12",
    "component": "vp3.1",
    "contrast": "vp3.1 - prior",
    "ctype": "layer_vs_prior",
    "instrument": "vp3.1 X+Y",
    "control_score": 90.3,
    "harnessed_score": 93.0,
    "h": 9.7,
    "delta": 2.68,
    "s": 1.176,
    "r_delta_over_h": 0.2763,
    "ci_lo": 0.45,
    "ci_hi": 5.06,
    "p": null,
    "n": 224,
    "se_method": "ci",
    "verdict": "pos",
    "prereg": "?",
    "flags": "",
    "source": "companion-vp3 run record",
    "note": "floor fails -0.60 twice",
    "metric_scale_max": 100.0
   },
   {
    "id": "L011",
    "model": "opus-5.5",
    "team": "longrun",
    "outcome": "O3",
    "component": "living",
    "contrast": "living - bare",
    "ctype": "carried_w2s",
    "instrument": "hourly probes code (H0-H24)",
    "control_score": 90.0,
    "harnessed_score": 90.0,
    "h": 10.0,
    "delta": 0.0,
    "s": 10.607,
    "r_delta_over_h": 0.0,
    "ci_lo": null,
    "ci_hi": null,
    "p": null,
    "n": 16,
    "se_method": "prop",
    "verdict": "null",
    "prereg": "Y",
    "flags": "ceiling,prereg",
    "source": "program record O3 H24",
    "note": "0 discordant item-hours; degenerate CI; unpaired 0.884 vs 0.913",
    "metric_scale_max": 1.0
   },
   {
    "id": "L163",
    "model": "opus-5.5",
    "team": "appworld",
    "outcome": "O13",
    "component": "layer+sensor",
    "contrast": "layer+sensor - bare",
    "ctype": "layer_vs_bare",
    "instrument": "test_challenge stream drift+52 crashes",
    "control_score": 89.9,
    "harnessed_score": 94.2,
    "h": 10.1,
    "delta": 4.3,
    "s": 1.346,
    "r_delta_over_h": 0.4257,
    "ci_lo": null,
    "ci_hi": null,
    "p": 0.0014,
    "n": null,
    "se_method": "p",
    "verdict": "pos",
    "prereg": "Y",
    "flags": "prereg",
    "source": "appworld run record",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L012",
    "model": "opus-5.5",
    "team": "longrun",
    "outcome": "O3",
    "component": "living",
    "contrast": "living - bare",
    "ctype": "carried_w2s",
    "instrument": "hourly probes tau2 pass^1 (H0-H24)",
    "control_score": 89.0,
    "harnessed_score": 89.0,
    "h": 11.0,
    "delta": 1.4,
    "s": 4.031,
    "r_delta_over_h": 0.1273,
    "ci_lo": -5.8,
    "ci_hi": 10.0,
    "p": 0.91,
    "n": 15,
    "se_method": "ci",
    "verdict": "null",
    "prereg": "Y",
    "flags": "ceiling,prereg",
    "source": "program record O3 H24",
    "note": "16 vs 11 discordant",
    "metric_scale_max": 1.0
   },
   {
    "id": "L099",
    "model": "opus-5.5",
    "team": "long-horizon",
    "outcome": "O11",
    "component": "sensor",
    "contrast": "sensor - bare",
    "ctype": "layer_vs_bare",
    "instrument": "UD-drift wave 2",
    "control_score": 89.0,
    "harnessed_score": 100.0,
    "h": 11.0,
    "delta": 10.8,
    "s": 4.4643,
    "r_delta_over_h": 0.9818,
    "ci_lo": 3.3,
    "ci_hi": 20.8,
    "p": 0.00024,
    "n": 120,
    "se_method": "ci",
    "verdict": "pos",
    "prereg": "?",
    "flags": "",
    "source": "program record O11",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L147",
    "model": "opus-5.5",
    "team": "companion",
    "outcome": "O12",
    "component": "vp2",
    "contrast": "vp2 - prior",
    "ctype": "layer_vs_prior",
    "instrument": "v6 silent shift FINAL",
    "control_score": 86.9,
    "harnessed_score": 96.2,
    "h": 13.1,
    "delta": 9.3,
    "s": 3.6225,
    "r_delta_over_h": 0.7099,
    "ci_lo": 2.7,
    "ci_hi": 16.9,
    "p": null,
    "n": null,
    "se_method": "ci",
    "verdict": "pos",
    "prereg": "?",
    "flags": "",
    "source": "companion run record",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L174",
    "model": "opus-5.5",
    "team": "appworld",
    "outcome": "APW",
    "component": "Opus+layer",
    "contrast": "Opus+layer - Opus alone",
    "ctype": "layer_vs_bare",
    "instrument": "durable test_challenge stream (crash+drift+swap)",
    "control_score": 84.9,
    "harnessed_score": 91.6,
    "h": 15.1,
    "delta": 6.7,
    "s": 1.7109,
    "r_delta_over_h": 0.4437,
    "ci_lo": null,
    "ci_hi": null,
    "p": 9e-05,
    "n": null,
    "se_method": "p",
    "verdict": "pos",
    "prereg": "?",
    "flags": "audited_cross",
    "source": "appworld run record",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L126",
    "model": "opus-5.5",
    "team": "companion",
    "outcome": "O12",
    "component": "vp3",
    "contrast": "vp3 - prior",
    "ctype": "layer_vs_prior",
    "instrument": "vp3 pooled",
    "control_score": 84.4,
    "harnessed_score": 95.2,
    "h": 15.6,
    "delta": 10.8,
    "s": 2.1939,
    "r_delta_over_h": 0.6923,
    "ci_lo": 6.6,
    "ci_hi": 15.2,
    "p": null,
    "n": null,
    "se_method": "ci",
    "verdict": "pos",
    "prereg": "Y",
    "flags": "prereg",
    "source": "companion-vp3 run record",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L124",
    "model": "opus-5.5",
    "team": "companion",
    "outcome": "O12",
    "component": "vp3",
    "contrast": "vp3 - prior",
    "ctype": "layer_vs_prior",
    "instrument": "vp3 A2",
    "control_score": 83.3,
    "harnessed_score": 95.2,
    "h": 16.7,
    "delta": 11.9,
    "s": 2.7296,
    "r_delta_over_h": 0.7126,
    "ci_lo": 6.9,
    "ci_hi": 17.6,
    "p": null,
    "n": null,
    "se_method": "ci",
    "verdict": "pos",
    "prereg": "Y",
    "flags": "prereg",
    "source": "companion-vp3 run record",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L166",
    "model": "opus-5.5",
    "team": "appworld",
    "outcome": "O13",
    "component": "v3",
    "contrast": "v3 - drifted bare",
    "ctype": "layer_vs_bare",
    "instrument": "amazon-cents v3 (212 tasks; seen in dev)",
    "control_score": 80.7,
    "harnessed_score": 93.9,
    "h": 19.3,
    "delta": 13.2,
    "s": 2.7769,
    "r_delta_over_h": 0.6839,
    "ci_lo": null,
    "ci_hi": null,
    "p": 2e-06,
    "n": null,
    "se_method": "p",
    "verdict": "pos",
    "prereg": "?",
    "flags": "posthoc",
    "source": "appworld run record",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L057",
    "model": "opus-5.5",
    "team": "model succession",
    "outcome": "O6",
    "component": "a5",
    "contrast": "a5 - fresh",
    "ctype": "layer_vs_bare",
    "instrument": "Haiku->Opus a5",
    "control_score": 74.2,
    "harnessed_score": 89.4,
    "h": 25.8,
    "delta": 15.2,
    "s": 6.5851,
    "r_delta_over_h": 0.5891,
    "ci_lo": null,
    "ci_hi": null,
    "p": null,
    "n": 66,
    "se_method": "prop",
    "verdict": "pos",
    "prereg": "?",
    "flags": "",
    "source": "program record O6",
    "note": "",
    "metric_scale_max": 100.0
   }
  ],
  "weaker6": [
   {
    "id": "L054",
    "model": "opus-4.8",
    "team": "model succession",
    "outcome": "O6",
    "component": "a5",
    "contrast": "a5 - Opus 4.8 alone",
    "ctype": "layer_vs_bare",
    "instrument": "a5 DOWN, unseen shard 1 (66/pair)",
    "control_score": 90.9,
    "harnessed_score": 97.0,
    "h": 9.1,
    "delta": 6.1,
    "s": 4.1161,
    "r_delta_over_h": 0.6703,
    "ci_lo": null,
    "ci_hi": null,
    "p": null,
    "n": 66,
    "se_method": "prop",
    "verdict": "pos",
    "prereg": "Y",
    "flags": "prereg",
    "source": "program record O6",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L002",
    "model": "haiku",
    "team": "appworld",
    "outcome": "O1",
    "component": "Haiku+Opus-compiled store",
    "contrast": "Haiku+Opus-compiled store - Haiku bare",
    "ctype": "carried_s2w",
    "instrument": "AppWorld test_normal TGC",
    "control_score": 80.1,
    "harnessed_score": 86.6,
    "h": 19.9,
    "delta": 6.5,
    "s": 4.0491,
    "r_delta_over_h": 0.3266,
    "ci_lo": null,
    "ci_hi": null,
    "p": null,
    "n": 168,
    "se_method": "prop",
    "verdict": "pos",
    "prereg": "?",
    "flags": "audited_cross",
    "source": "program record O1",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L058",
    "model": "haiku",
    "team": "model succession",
    "outcome": "O6",
    "component": "online store",
    "contrast": "online store - fresh",
    "ctype": "carried_self",
    "instrument": "168-task stream, online store",
    "control_score": 78.6,
    "harnessed_score": 60.7,
    "h": 21.4,
    "delta": -17.9,
    "s": 4.9205,
    "r_delta_over_h": -0.8364,
    "ci_lo": null,
    "ci_hi": null,
    "p": null,
    "n": 168,
    "se_method": "prop",
    "verdict": "neg",
    "prereg": "?",
    "flags": "",
    "source": "program record O6 FINAL",
    "note": "613 admitted, 0 rejected",
    "metric_scale_max": 100.0
   },
   {
    "id": "L007",
    "model": "haiku",
    "team": "evolved engine",
    "outcome": "O2",
    "component": "lineage",
    "contrast": "lineage - bare_notice",
    "ctype": "layer_vs_prior",
    "instrument": "AppWorld fresh_normal k=14 single crash survival",
    "control_score": 72.4,
    "harnessed_score": 89.7,
    "h": 27.6,
    "delta": 17.24,
    "s": 5.2781,
    "r_delta_over_h": 0.6246,
    "ci_lo": 6.9,
    "ci_hi": 27.59,
    "p": 0.006,
    "n": 58,
    "se_method": "ci",
    "verdict": "pos",
    "prereg": "Y",
    "flags": "prereg",
    "source": "same",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L001",
    "model": "haiku",
    "team": "appworld",
    "outcome": "O1",
    "component": "Haiku+Opus-compiled store",
    "contrast": "Haiku+Opus-compiled store - Haiku bare",
    "ctype": "carried_s2w",
    "instrument": "AppWorld test_challenge TGC",
    "control_score": 66.6,
    "harnessed_score": 75.5,
    "h": 33.4,
    "delta": 8.9,
    "s": 3.1257,
    "r_delta_over_h": 0.2665,
    "ci_lo": null,
    "ci_hi": null,
    "p": null,
    "n": 417,
    "se_method": "prop",
    "verdict": "pos",
    "prereg": "?",
    "flags": "audited_cross",
    "source": "program record O1 (audit READY, Astra)",
    "note": "",
    "metric_scale_max": 100.0
   },
   {
    "id": "L006",
    "model": "haiku",
    "team": "evolved engine",
    "outcome": "O2",
    "component": "lineage",
    "contrast": "lineage - bare_notice",
    "ctype": "layer_vs_prior",
    "instrument": "AppWorld challenge k=14 single crash survival",
    "control_score": 63.1,
    "harnessed_score": 92.3,
    "h": 36.9,
    "delta": 29.23,
    "s": 5.8878,
    "r_delta_over_h": 0.7921,
    "ci_lo": 18.46,
    "ci_hi": 41.54,
    "p": 2e-05,
    "n": 65,
    "se_method": "ci",
    "verdict": "pos",
    "prereg": "Y",
    "flags": "prereg",
    "source": "same",
    "note": "",
    "metric_scale_max": 100.0
   }
  ],
  "note": "sorted by h; control/harnessed/delta/s/h on a 0-100 scale (x 100/mx); r = delta/h (None where h = 0); component = left side of the ledger contrast"
 },
 "headline": {
  "opus15": {
   "k": 0.539,
   "k_ci95": [
    0.427,
    0.675
   ],
   "aic": {
    "headroom": 77.16,
    "constant": 96.82,
    "zero": 107.1,
    "capped": 89.39,
    "smooth_cap": 83.15
   }
  },
  "weaker6": {
   "k": 0.383,
   "k_ci95": [
    -0.058,
    0.837
   ],
   "tau": 11.45
  },
  "all21": {
   "k": 0.453,
   "k_ci95": [
    0.299,
    0.609
   ]
  },
  "forecast_bf_own_tau": {
   "vs_zero": 186.23,
   "vs_constant": 9.82,
   "vs_capped": 5.58,
   "vs_smooth_cap": 3.05
  },
  "forecast_bf_shared_tau_published": 13.1,
  "forecast_error_weaker_to_opus": {
   "zero": {
    "mae": 6.32,
    "rmse": 8.067,
    "mean_abs_err_over_s": 2.33,
    "strictly_best_rows": 4
   },
   "constant": {
    "mae": 5.166,
    "rmse": 5.863,
    "mean_abs_err_over_s": 2.317,
    "strictly_best_rows": 0
   },
   "headroom": {
    "mae": 3.304,
    "rmse": 3.862,
    "mean_abs_err_over_s": 1.073,
    "strictly_best_rows": 4
   },
   "capped constant": {
    "mae": 4.021,
    "rmse": 4.614,
    "mean_abs_err_over_s": 1.457,
    "strictly_best_rows": 6
   }
  }
 },
 "reproduction_check": {
  "source": "the synthesis paper's re-test of the law",
  "leaves_compared": 508,
  "mismatches": [],
  "exact": true
 }
}