{
 "tm": "1RB1RA_1LC1RE_---0LD_1LE1LD_0RA0LF_1LF0LB",
 "bucket_reason": "Medium confidence at observe: the mechanism is exact (mirror twin of #5), but there is no non-halting argument, because the shared edge has no known erosion bound. Proof work waits on Dennis.",
 "question": "#26 is #5 mirrored with renamed states (same edge at every bounce): park it with #5 as one cryptid candidate, or fund the shared Rocq rule reduction now?",
 "options": [
  {
   "id": "park_with_5",
   "label": "Park with #5 as one cryptid candidate",
   "consequence": "Status stays CRYPTID_CANDIDATE with the rules and the twin map on file. Any future erosion bound for #5 transfers to #26 through the mirror map, giving 2 kills from 1 proof. No agent time now."
  },
  {
   "id": "pair_reduction",
   "label": "One stage (at most 2 h): Rocq rule reduction for both twins",
   "consequence": "One .v proves the 7 local rules, the bounce and the halting door for #5 and #26, as AntiHydra.v does: halting reduces to the edge eating the block. This is the cryptid-protocol deliverable #5 still owes. It is not a kill: there is no nonhalt lemma."
  },
  {
   "id": "erosion_research",
   "label": "Fund a research stage on the shared erosion bound",
   "consequence": "Tries #5's best remaining route: a two-scale computer-checked abstraction (exact last D cells plus a regular summary of the deep part). It is a long shot; the #5 research measured D > 64 and maybe 1e12 states."
  }
 ],
 "recommended": "park_with_5",
 "rationale": [
  "Tables identical under mirror + renaming; edges agree at all 102,240 bounces to 1e10.",
  "So #5's 2026-10-10 cryptid call applies unchanged: no known erosion bound.",
  "The owed rule reduction is best done once, in one file for both twins."
 ],
 "evidence": [
  {
   "kind": "figure",
   "text": "Fig 2: #26 beside #5 mirrored, one row per bounce; red marks the only differing cell (the wall).",
   "href": "img/twin_vs_5_mirrored.png"
  },
  {
   "kind": "fact",
   "text": "Twin map: #26 at 0^inf <E 1^n w 0^inf (step t) <-> #5 at 0^inf rev(w) 1^(n+1) B> 0^inf (step t+2k), k = 2..102,241, 102,240/102,240.",
   "href": "work/lockstep_1e10.txt"
  },
  {
   "kind": "fact",
   "text": "Rules: walks, bounce, T/Ca/Fo/E1/E2/P1/P2 and the halting door, 60/60 each; no-00 invariant 0 violations to 1e9.",
   "href": "work/validate_rules.json"
  },
  {
   "kind": "link",
   "text": "#5 erosion research: why no deterministic bound was found",
   "href": "../1RB1RA_0LC0RF_1LD1LC_1RE1LB_---0RA_1RF0RD/research/erosion.md"
  }
 ],
 "effort_if_yes": "park_with_5: none. pair_reduction: one stage (at most 2 h), new file without a nonhalt claim. erosion_research: open-ended, likely several stages with a low chance.",
 "updated": "2026-10-10",
 "answer": {
  "option": "park_with_5",
  "note": "auto-applied: Claude's recommendation, under Dennis's standing instruction 'do your recommendations and continuously orchestrate' (2026-10-10, while he was at lunch). Override any time.",
  "date": "2026-10-10"
 }
}