{
 "tm": "1RB1LE_0RC1RC_0RD0RB_1RE0RF_1LA0LE_---1RB",
 "bucket_reason": "STUCK at deepen2 with mechanism confidence med. Two rounds of hand rule-closure did not converge, and status.json asks for your call: abstract simulator, decider search, or park.",
 "question": "Park #20 (it stays STUCK), or fund one more stage to build an abstract run-length simulator that generates its full call-stack rule table?",
 "options": [
  {
   "id": "park",
   "label": "Park: keep STUCK, no more agent time now",
   "consequence": "status.json stays STUCK (deepen2) with a re-entry note. Costs 0 agent hours. The two deepen rounds (16 validated transition rules, 3 halting patterns, census and danger scans) stay on file. The budget goes to the family sweep and twins."
  },
  {
   "id": "abstract_sim",
   "label": "One stage: abstract run-length simulator",
   "consequence": "About 2 h of agent time and 1-2 processes. Builds a simulator over run-length words, driven by the validated rules and checked against the real run at every E:R moment, to list every transit family automatically. Even if it works, two things remain: the column grammar with inequalities (no candidate yet) and the non-halting argument. Expect 1-2 more stages before proof.md."
  },
  {
   "id": "decider",
   "label": "One stage: closed-language / decider search",
   "consequence": "About 2 h of agent time and 1-2 processes for a CTL-style closed-language search with larger parameters. Low odds. Small configurations next to reachable shapes halt in every residue class, so a value-free language would have to exclude small blocks by structure alone."
  }
 ],
 "recommended": "park",
 "rationale": "The architecture is TM20's call stack, but the hand closure is spreading rather than converging. Round 2 added 10 rules (731/731 instances pass) and still found 2 parsed and 7 transit classes that first appear in epoch 7. It also found a root that drifts right, which TM20's fixed-root invariant does not cover. The blocker is not the rule list. It is an invariant grammar that keeps every call away from the halting shapes, and there is no candidate for it. Small configurations next to reachable ones often halt (640 of 1,584); medium ones never do (0 of 144). The real run makes a room-2 call in epoch 7, so a proof must show that small blocks arise only at low recursion levels and in fixed shapes. The observer warns this turns cryptid-like if that fails. mxdys's FractalType7 LLM batch did not solve this machine either. Parking costs nothing and frees budget for machines with validated mechanisms. Revisit if an abstract run-length simulator gets built for another fractal (for example, if #16's accelerated simulator generalises), or if the family sweep finds a solved FractalType7 sibling whose grammar covers transit callees.",
 "evidence": [
  {
   "kind": "validation",
   "text": "Deepen round 2: 10 new rules (Call4, Call2 odd/even, ResumeTY and 3 halting patterns), 731/731 random-context instances. Round-1 rules unchanged.",
   "href": "scripts/deepen2_validate.txt"
  },
  {
   "kind": "analysis",
   "text": "Census to 1e9: of 46,375 E:R moments, 45,716 parse as standard frames (40 classes) and 659 are transit forms (86 shapes). 2 parsed and 7 transit classes first appear in epoch 7 (step >= 4.58e8).",
   "href": "scripts/deepen2_census_1e9.txt"
  },
  {
   "kind": "risk",
   "text": "Danger scan with caller X/Y(20,16,5) and 'E> 0^2 1^n 0^2 1^m 0^g 1100 1^5': for small n, m, g, 640 of 1,584 halt, in every residue class. For medium sizes, 0 of 144 halt within 2e7 steps.",
   "href": "scripts/deepen2_danger_small.txt"
  },
  {
   "kind": "sim",
   "text": "The real run makes a room-2 call in epoch 7 (step 458,654,411). Shifting its block sizes by -6..+6 (36 runs) never halts within 5e8 steps.",
   "href": "scripts/deepen2_perturb.txt"
  },
  {
   "kind": "sim",
   "text": "No halt to 1e9 steps (C core) or to 1.99e13 steps (qsim). Span grows like sqrt(5.6 t).",
   "href": "notes.md"
  },
  {
   "kind": "figure",
   "text": "Fig 1, state-E entries to 3e6 steps: the replay staircases to the left of each wide white band.",
   "href": "img/state_ER_3e6.png"
  },
  {
   "kind": "doc",
   "text": "Deepen-2 summary, remaining gaps and the three ways to unstick it.",
   "href": "deepen.json"
  }
 ],
 "effort_if_yes": "None now: 0 agent hours and 0 processes. Status stays STUCK, with a re-entry note.",
 "updated": "2026-10-10",
 "answer": {
  "option": "park",
  "note": "Dennis: go with all recommendations (2026-10-10)",
  "date": "2026-10-10"
 }
}