{
 "tm": "1RB0LF_1RC1RB_0RD0RC_1LE1LF_1LD---_0LB1LA",
 "bucket_reason": "Priority call: the follow stage found no halt to phase 1,857 and predicts the next excursion (phase 3,321) benign. A proof would need a grammar for the merge wave, with no template close.",
 "question": "#37 stays in sync to phase 1,857, and its next excursion (phase 3,321) is predicted benign: park it, spend one round deriving the merge-wave rule, or first confirm phase 3,321 on the real orbit?",
 "options": [
  {
   "id": "park",
   "label": "Park #37 behind the other machines",
   "consequence": "No compute. The C simulator, the checkpoint at phase 1,857, the fast-forward predictions and maturation rule M stay in notes.md and deepen.json for a later round."
  },
  {
   "id": "grammar",
   "label": "One ~2 h round on the merge-wave rule",
   "consequence": "The agent maps what one wave does to the lowest runs as a function of run-length residues, using synthetic layouts and the fast-forward simulator. Exit: a class of low layouts that the cascade maps into itself with a1 >= 6 gives a proof plan for the sync; otherwise park."
  },
  {
   "id": "confirm",
   "label": "Confirm the phase-3,321 excursion on the real orbit (~1 h)",
   "consequence": "One light C process resumes the real orbit from the phase-1,857 checkpoint through phase 3,321 and its cascade. Exit: compare with the fast-forward prediction. A match strengthens M; a mismatch or an off-sync is reported."
  }
 ],
 "recommended": "park",
 "rationale": [
  "No halt to phase 1,857; 4/4 excursions benign; the 3-run in question was absorbed by a merge.",
  "Fast-forward is exact on the past excursion and predicts the next 3 excursions benign.",
  "A proof needs a wave grammar on unbounded run words; no busycoq template is near."
 ],
 "evidence": [
  {
   "kind": "fact",
   "text": "Real orbit followed to phase 1,857 (n ~ 2^1860) with the C port: excursions at phases 91, 119, 215 and 487, all benign, with 14, 77, 93 and 700 cells in the run that became the bottom.",
   "href": "work/follow/real2.log"
  },
  {
   "kind": "fact",
   "text": "C port checks: 160/160 phases equal to Python; random instances 90/90 vs Python and 90/90 vs exact fastphase; fast-path rules 3,000/3,000 each.",
   "href": "work/follow/cmp_c_vs_py_160.log"
  },
  {
   "kind": "fact",
   "text": "Fast-forward from phases 311 and 411 reproduces the real excursion at phase 487 and its cascade (45/45 phases). From phases 811, 1,011 and 1,207 it predicts phase 3,321 as benign, with a1 = 555.",
   "href": "work/follow/ffwd/next_d1196_r4.txt"
  },
  {
   "kind": "fact",
   "text": "Fig 5: frozen runs are vertical stripes, merge waves are diagonal bands, and short runs (orange) are absorbed by the next wave.",
   "href": "img/follow_lowruns_real.png"
  }
 ],
 "effort_if_yes": "park: none. grammar: one ~2 h round, light compute, no Rocq. confirm: about 1 h of one light C process from work/follow/ckpt_follow_end.txt, plus ~15 min of agent time.",
 "updated": "2026-10-10",
 "answer": {
  "option": "park",
  "note": "Dennis: go with recommendations (2026-10-11)",
  "date": "2026-10-11"
 }
}