{
 "tm": "1RB0LC_1RC0RA_1RD0RF_0LE---_1LB1LA_1LD1RA",
 "bucket_reason": "Confidence is still med after corner_rules_2. Every corner event on H* orbits is ruled now and the closure target is met, but the clause for short right ends (block seeds) is wider than what the machine produces, so H*c2 is not yet an inductive invariant.",
 "question": "Fund one stage to replace #16's seed clause by closed forms of the short (S) transitions, so that H*c2 becomes inductive, before proof.md?",
 "options": [
  {
   "id": "seed_clause",
   "label": "One seed-clause stage",
   "consequence": "About 2 h, 1-2 light processes, no Coq. Derive the S transitions (collapse completion by cascade and edge-mode eats; Absorb starts) in closed form from the macro rules, state the short right ends they produce as linear relations, replace the seed clause of H*c2 with them, and rerun the closure tests including random short seeds drawn from the new clause. Also rule or exclude V1 with d = 1 and Z nonempty. Exits: closed with 0 leaks -> proof.md; a seed family that needs non-linear relations -> park."
  },
  {
   "id": "proof_now",
   "label": "Write proof.md now",
   "consequence": "Start the hub-to-hub induction on H*c2 with the macro rules as lemmas (Conv/Transfer from TM5's P/P_S/P_S'). The seed clause and the V1 d = 1, Z nonempty case stay as open lemmas."
  },
  {
   "id": "park",
   "label": "Park",
   "consequence": "Status becomes parked at corner_rules_2. macro_sim3, the CORNER2 rules and the closure tests stay in scripts/ and notes.md."
  }
 ],
 "recommended": "seed_clause",
 "rationale": [
  "Closure target met: 10,000/10,000 orbits at both scales by validated rules; 0 halts.",
  "Left: 29/10^4 random short H* seeds leave; seeds the machine makes stay in.",
  "S transitions are chains of validated rules, so the fix is linear."
 ],
 "evidence": [
  {
   "kind": "fact",
   "text": "CORNER2 rules by exact simulation: 11 families 40/40 each; negative controls 0/110.",
   "href": "scripts/val_corner2_13.json"
  },
  {
   "kind": "fact",
   "text": "No margin, 40 traversals: 10,000/10,000 orbits ruled and in H*c2 at levels <= 5,459 (and <= 1e30); 0 halts.",
   "href": "scripts/hubinv_Hc3_1002_smax5459.txt"
  },
  {
   "kind": "fact",
   "text": "Random short H* seeds (runs >= 1,000, levels <= 1e30): 9,971/10,000 stay in H*c2; 27 P-mode ends with P even, 2 band k = 2.",
   "href": "scripts/hc3_seeds_905_c1000.txt"
  },
  {
   "kind": "fact",
   "text": "End-to-end vs accel_sim through the new rules: 30 orbits, 238 right ends in order, 0 mismatches.",
   "href": "scripts/e2e3_1002.txt"
  }
 ],
 "effort_if_yes": "One stage: about 2 h of agent time and 1-2 light processes (macro_sim3, exact simulation, closure tests). No Coq.",
 "updated": "2026-10-12",
 "answer": {
  "option": "seed_clause",
  "note": "Dennis: b (fund all five), 2026-10-12",
  "date": "2026-10-12"
 }
}