{
 "analysis_final": true,
 "confirmatory": {
  "H1": {
   "ci95_bootstrap": [
    -0.071429,
    0.142857
   ],
   "cohens_dz": 0.1512,
   "control": "fixed",
   "effective_n_sign_test": 4,
   "family": "e144_director",
   "holm": {
    "holm_log10_p_adj": -0.1627,
    "holm_significant": false,
    "log10_p": -0.1627,
    "p_display": "0.688",
    "rank": 3
   },
   "log10_p": -0.1627,
   "losses": 2,
   "mean_diff": 0.02381,
   "n_experiments": 7,
   "n_required": 20,
   "p_display": "0.687",
   "practical_threshold": 0.05,
   "required_experiments": 20,
   "ties": 3,
   "treatment": "director",
   "underpowered": true,
   "verdict": "UNDERPOWERED",
   "wins": 2
  },
  "H2": {
   "ci95_bootstrap": [
    0.0,
    0.004371
   ],
   "cohens_dz": 0.4472,
   "control": "theta0",
   "effective_n_sign_test": 1,
   "family": "e144_meta",
   "holm": {
    "holm_log10_p_adj": 0.0,
    "holm_significant": false,
    "log10_p": -0.301,
    "p_display": "0.5",
    "rank": 2
   },
   "log10_p": -0.301,
   "losses": 0,
   "mean_diff": 0.001457,
   "n_experiments": 5,
   "n_required": 12,
   "p_display": "0.5",
   "practical_threshold": 0.01,
   "required_experiments": 12,
   "ties": 4,
   "treatment": "meta",
   "underpowered": true,
   "verdict": "UNDERPOWERED",
   "wins": 1
  },
  "H3": {
   "ci95_bootstrap": [
    0.014799,
    0.025333
   ],
   "cohens_dz": 2.6085,
   "control": "constant",
   "effective_n_sign_test": 7,
   "family": "e144_selfmodel",
   "holm": {
    "holm_log10_p_adj": -1.6301,
    "holm_significant": true,
    "log10_p": -2.1072,
    "p_display": "0.00781",
    "rank": 1
   },
   "log10_p": -2.1072,
   "losses": 0,
   "mean_diff": 0.019979,
   "n_experiments": 7,
   "n_required": 20,
   "p_display": "0.00781",
   "practical_threshold": 0.005,
   "required_experiments": 20,
   "ties": 0,
   "treatment": "calibrated",
   "underpowered": true,
   "verdict": "UNDERPOWERED",
   "wins": 7
  }
 },
 "cutoff": "2026-11-22T13:57:37Z",
 "cutoff_reason": "OWNER-DIRECTED EARLY STOP at the owner's explicit request; the preregistered stopping rule (20/12/20 verified or campaign start + 10 h) was NOT met. All confirmatory tests are reported UNDERPOWERED and no hypothesis is called supported or contradicted. Remaining campaign rows are post-cutoff and are not merged.",
 "environment": "E142 constructed world families; each instance is regenerated from (family, level, seed) in experiments.jsonl and the published study salt",
 "experiments_by_family_state": {
  "e144_director": {
   "VERIFICATION_PENDING": 1,
   "VERIFIED": 7
  },
  "e144_meta": {
   "VERIFICATION_PENDING": 1,
   "VERIFIED": 5
  },
  "e144_selfmodel": {
   "VERIFICATION_PENDING": 1,
   "VERIFIED": 7
  }
 },
 "exported_at": "2026-11-22T13:58:03Z",
 "organism_suite": {
  "exploratory": true,
  "gates": {
   "FAIL": 2,
   "PARTIAL": 3
  },
  "math_discovery": {
   "path": "organism/E144_MATH.json",
   "preregistration": "docs/evolution/E144_PREREGISTRATION_AMENDMENT_2.md",
   "summaries": {
    "exact_discovery_minus_random": "NOT SUPPORTED",
    "prediction_discovery_minus_prior": "SUPPORTED"
   }
  },
  "path": "organism/E144_ORGANISM.json",
  "preregistration": "docs/evolution/E144_PREREGISTRATION_AMENDMENT_1.md",
  "summaries": {
   "O1": {
    "goal_aligned_minus_no_model": "SUPPORTED",
    "goal_aligned_minus_scratch": "NOT SUPPORTED (treatment worse)",
    "prediction_aligned_minus_frequency": "NOT SUPPORTED",
    "prediction_aligned_minus_naive": "NOT SUPPORTED",
    "prediction_aligned_minus_scratch": "NOT SUPPORTED (treatment worse)"
   },
   "O2": {
    "score_capability_minus_no_world_model": "SUPPORTED",
    "score_capability_minus_random": "SUPPORTED",
    "score_capability_minus_retrieval": "SUPPORTED",
    "score_capability_minus_scratch": "NOT SUPPORTED (treatment worse)"
   },
   "O3": {
    "sealed_full_minus_none": "NOT SUPPORTED",
    "sealed_full_minus_reset": "NOT SUPPORTED (treatment worse)",
    "sealed_full_minus_retrieval": "SUPPORTED"
   },
   "O4": {
    "exact_discovery_minus_random": "SUPPORTED",
    "exact_discovery_minus_random_matched": "NOT SUPPORTED (treatment worse)",
    "prediction_discovery_minus_prior": "SUPPORTED",
    "prediction_discovery_minus_random_matched": "NOT SUPPORTED (treatment worse)"
   },
   "O5": {
    "heldout_collective_minus_single": "NOT SUPPORTED (treatment worse)"
   }
  }
 },
 "post_cutoff": {
  "H1": {
   "ci95_bootstrap": null,
   "cohens_dz": null,
   "control": "fixed",
   "effective_n_sign_test": 0,
   "family": "e144_director",
   "holm": {
    "holm_log10_p_adj": 0.0,
    "holm_significant": false,
    "log10_p": 0.0,
    "p_display": "1",
    "rank": 1
   },
   "log10_p": 0.0,
   "losses": 0,
   "mean_diff": null,
   "n_experiments": 0,
   "n_required": 20,
   "p_display": "1",
   "practical_threshold": 0.05,
   "required_experiments": 20,
   "ties": 0,
   "treatment": "director",
   "underpowered": true,
   "verdict": "UNDERPOWERED",
   "wins": 0
  },
  "H2": {
   "ci95_bootstrap": null,
   "cohens_dz": null,
   "control": "theta0",
   "effective_n_sign_test": 0,
   "family": "e144_meta",
   "holm": {
    "holm_log10_p_adj": 0.0,
    "holm_significant": false,
    "log10_p": 0.0,
    "p_display": "1",
    "rank": 2
   },
   "log10_p": 0.0,
   "losses": 0,
   "mean_diff": null,
   "n_experiments": 0,
   "n_required": 12,
   "p_display": "1",
   "practical_threshold": 0.01,
   "required_experiments": 12,
   "ties": 0,
   "treatment": "meta",
   "underpowered": true,
   "verdict": "UNDERPOWERED",
   "wins": 0
  },
  "H3": {
   "ci95_bootstrap": null,
   "cohens_dz": null,
   "control": "constant",
   "effective_n_sign_test": 0,
   "family": "e144_selfmodel",
   "holm": {
    "holm_log10_p_adj": 0.0,
    "holm_significant": false,
    "log10_p": 0.0,
    "p_display": "1",
    "rank": 3
   },
   "log10_p": 0.0,
   "losses": 0,
   "mean_diff": null,
   "n_experiments": 0,
   "n_required": 20,
   "p_display": "1",
   "practical_threshold": 0.005,
   "required_experiments": 20,
   "ties": 0,
   "treatment": "calibrated",
   "underpowered": true,
   "verdict": "UNDERPOWERED",
   "wins": 0
  }
 },
 "schema": "cain42.e144.analysis.v1",
 "secondary": {
  "H1_counts": {
   "correct": {
    "director": 17,
    "fixed": 9,
    "random": 12
   },
   "false_discoveries": {
    "director": 10,
    "fixed": 4,
    "random": 7
   },
   "probes_used": {
    "director": 672,
    "fixed": 672,
    "random": 672
   },
   "true_effects": 28
  },
  "H1b": {
   "ci95_bootstrap": [
    -0.047619,
    0.107143
   ],
   "cohens_dz": 0.207,
   "control": "random",
   "effective_n_sign_test": 4,
   "family": "e144_director",
   "log10_p": -0.1627,
   "losses": 2,
   "mean_diff": 0.02381,
   "n_experiments": 7,
   "p_display": "0.687",
   "ties": 3,
   "treatment": "director",
   "wins": 2
  },
  "H2_per_generation_sealed_mean": {
   "0": 0.699259,
   "1": 0.699259,
   "2": 0.700716
  },
  "H2_promotions": {
   "meta": 1,
   "meta_random": 1
  },
  "H2_test_time_episodes": {
   "meta": 2260,
   "meta_random": 2190,
   "theta0": 2400
  },
  "H2_train_episodes": {
   "meta": 2970,
   "meta_random": 3030
  },
  "H2b": {
   "ci95_bootstrap": [
    0.0,
    0.005883
   ],
   "cohens_dz": 0.7303,
   "control": "meta_random",
   "effective_n_sign_test": 2,
   "family": "e144_meta",
   "log10_p": -0.6021,
   "losses": 0,
   "mean_diff": 0.002932,
   "n_experiments": 5,
   "p_display": "0.25",
   "ties": 3,
   "treatment": "meta",
   "wins": 2
  },
  "H2c": {
   "ci95_bootstrap": [
    0.002164,
    0.010044
   ],
   "cohens_dz": 1.2557,
   "control": "G0",
   "effective_n_sign_test": 4,
   "family": "e144_meta",
   "log10_p": -1.2041,
   "losses": 0,
   "mean_diff": 0.006189,
   "n_experiments": 5,
   "p_display": "0.0625",
   "ties": 1,
   "treatment": "meta",
   "wins": 4
  },
  "H3_mean_brier": {
   "calibrated": 0.045389,
   "constant": 0.065368,
   "raw": 0.062477
  },
  "H3_selective_half_neg_brier": {
   "calibrated": -0.046556,
   "constant": -0.082448,
   "raw": -0.075797
  },
  "H3b": {
   "ci95_bootstrap": [
    0.008098,
    0.026485
   ],
   "cohens_dz": 1.2833,
   "control": "raw",
   "effective_n_sign_test": 7,
   "family": "e144_selfmodel",
   "log10_p": -1.2041,
   "losses": 1,
   "mean_diff": 0.017089,
   "n_experiments": 7,
   "p_display": "0.0625",
   "ties": 0,
   "treatment": "calibrated",
   "wins": 6
  }
 },
 "source_revision": "9acd6a2545b122928c9d7e998612cdcefc2c8715",
 "status": {
  "agi": "NOT ESTABLISHED",
  "independent_reproductions": 0,
  "model_calls": 0,
  "neural_weights_modified": false,
  "official_benchmarks_run_in_e144": 0,
  "paid_api_usd": 0.0,
  "rsi": "EXPERIMENTAL"
 }
}
