{
 "tm": "1RB1LF_0RC0RB_0LD1LA_1LC0LE_0LA---_1LA1LF",
 "bucket_reason": "Medium confidence at observe: the rules are closed and validated, but the non-halting invariant is not found yet, so proof work waits for Dennis's call.",
 "question": "Fund one 2 h stage to find an invariant on (type, N, m) that keeps every overflow off the halting points, start proof.md with it open, or park?",
 "options": [
  {
   "id": "hunt",
   "label": "One stage (at most 2 h): find and test the invariant, then write proof.md",
   "consequence": "Derive the danger points of dirty orbits from k's bits (R/N/H landings). Propose an inequality on (type, N, m) and the dirty state and check it on 17M abstract cycles plus forced R3 resets; it must be closed under std -> pk -> std and R3. If one survives, write proof.md. Stop rule: no candidate closed under the big steps in 2 h means park."
  },
  {
   "id": "proof_md",
   "label": "Start proof.md now, with the invariant as an open lemma",
   "consequence": "proof.md gets the validated rule table (R1-R4, big steps, phi^j) and the template mapping. The invariant is marked open, so a skeptic would call it fatal and nothing can be formalized past the rules."
  },
  {
   "id": "park",
   "label": "Park (STUCK at observe)",
   "consequence": "Keeps rules.py, validate_rules.py, the 17M-cycle run and the danger-distance data on file. Stops work here, and leaves 349 (via TM518) open."
  }
 ],
 "recommended": "hunt",
 "rationale": [
  "Rules closed and validated; abstract = direct on all 1,002 overflows to 1e10.",
  "Halting only at rare odd landings; 0 hits in 17M cycles, gap >= 0.29 m and not shrinking.",
  "That margin suggests an inequality invariant (not a cryptid); it would also carry 349 via TM518."
 ],
 "evidence": [
  {
   "kind": "figure",
   "text": "Fig 2: counter word at each overflow; the encoding flips std/packed every cycle, dirty rows are rare",
   "href": "img/counter_rows.png"
  },
  {
   "kind": "fact",
   "text": "17,000,000 abstract cycles from blank (m = 4.25e7, about 5e22 steps): 0 halt/nested overflows, 1 dirty reset (cycle 27), 4 all-ones resets.",
   "href": "work/abstract_startA_17e6.out"
  },
  {
   "kind": "fact",
   "text": "Closest approach over 6M cycles: H points 0.287 m, N points 0.376 m before the overflow; R points down to distance 2.",
   "href": "work/danger_distance_6e6.out"
  },
  {
   "kind": "link",
   "text": "notes.md: rules, danger-point analysis, template comparison",
   "href": "notes.md"
  }
 ],
 "effort_if_yes": "One stage (about 2 h) for the invariant and proof.md, then about 2-3 stages of Rocq (packed counter lemmas, dirty-orbit lemma, invariant) if it holds.",
 "updated": "2026-10-10",
 "answer": {
  "option": "hunt",
  "note": "Dennis: go with recommendations (2026-10-11)",
  "date": "2026-10-11"
 }
}