{
  "analysis": "side32_static_calibration_budget_stress",
  "confirmatory": true,
  "comparison": "16 versus 256 instruction-off observations per edge",
  "bootstrap": {
    "unit": "task; three component draws averaged within task",
    "task_clusters": 40,
    "replicates": 20000,
    "seed": 20260829,
    "interval": "percentile 95%"
  },
  "metrics": {
    "classification_error": {
      "budget16_mean": 0.271875,
      "budget256_mean": 0.303125,
      "budget256_improvement": -0.03125,
      "paired_task_bootstrap_95ci": [
        -0.059375,
        -0.003125
      ],
      "higher_is_better": false
    },
    "classification_error_auc": {
      "budget16_mean": 0.29671006944444445,
      "budget256_mean": 0.2908246527777778,
      "budget256_improvement": 0.005885416666666668,
      "paired_task_bootstrap_95ci": [
        -0.01811653645833334,
        0.029548828124999982
      ],
      "higher_is_better": false
    },
    "stable_zero_fraction": {
      "budget16_mean": 0.2333333333333333,
      "budget256_mean": 0.2,
      "budget256_improvement": -0.03333333333333334,
      "paired_task_bootstrap_95ci": [
        -0.09166666666666667,
        0.025
      ],
      "higher_is_better": true
    },
    "local_scalar_reads_to_target": {
      "budget16_mean": 9347072.0,
      "budget256_mean": 9573717.333333334,
      "budget256_improvement": -226645.3333333334,
      "paired_task_bootstrap_95ci": [
        -430080.0,
        -10922.666666666744
      ],
      "higher_is_better": false
    }
  },
  "decision_rule": "sampling-limited only if final error improves by at least 0.02 and the paired 95% interval excludes zero",
  "sampling_limited": false,
  "sources": {
    "budget16": "results/coupled_ladder/p2_confirm_side32.json",
    "budget256": "results/coupled_ladder/p4_constant256_side32.json"
  }
}