# CAIN-42 BFT IP report (2026-10-05)

Engineering prioritisation, not patentability. Score classes: 5 DEVELOPING, 3 MATERIAL. The MATERIAL ones sit at the bottom
of that band (23-25/50) and are held down by design-around resistance (1/5). Every candidate has related prior
art and EASY design-arounds.

| id | candidate | score/50 | class | design-around | protection |
|---|---|---|---|---|---|
| BFT-IP-0001 | Quorum-committed authorization snapshots with local revocation epoch | 25 | MATERIAL | EASY | TRADE_SECRET (implementation) + DEFENSIVE_PUBLICATION candidate |
| BFT-IP-0002 | Fail-closed consensus stage: permit requires a gateway-verified QC over the exact decision commitment | 25 | MATERIAL | EASY | OPEN_PROTOCOL (commitment + QC format) / TRADE_SECRET (pipeline) |
| BFT-IP-0005 | Evidence posture: proof objects that state what consensus did NOT evaluate + public zero-CAIN verifier + signed hourly proofs | 23 | MATERIAL | EASY | OPEN_PROTOCOL (verifier + formats) |
| BFT-IP-0004 | Model-checked vote-history fast path | 20 | DEVELOPING | EASY | DEFENSIVE_PUBLICATION / OPEN |
| BFT-IP-0003 | Safety-asymmetric consensus deferral (non-permits ordered after reply) | 18 | DEVELOPING | EASY | DEFENSIVE_PUBLICATION |
| BFT-IP-0007 | Group commit with per-decision RFC 6962 inclusion proofs | 18 | DEVELOPING | EASY | OPEN |
| BFT-IP-0006 | Consensus-ordered quarantine + hash-chained Byzantine evidence | 17 | DEVELOPING | EASY | OPEN |
| BFT-IP-0008 | Hybrid Ed25519 + ML-DSA-65 quorum certificates | 14 | DEVELOPING | EASY | OPEN |

{
 "assumptions": "a team of 1-2 senior engineers with distributed-systems experience, Python, existing libraries (cryptography, FastAPI); ranges are engineering judgement, not measured, and exclude sales/ops",
 "rows": [
  {
   "mechanism": "PBFT engine at the tested level (BFT-TV-01..20, 39 suites)",
   "engineer_months": "6-12",
   "expertise": "BFT protocols",
   "dominant_burden": "fault injection + liveness debugging (live soak found 4+ liveness bugs)"
  },
  {
   "mechanism": "BFT-IP-0002 fail-closed consensus stage + QC verification",
   "engineer_months": "0.5-1.5",
   "expertise": "applied crypto",
   "dominant_burden": "integration + failure semantics"
  },
  {
   "mechanism": "BFT-IP-0001 snapshot data plane",
   "engineer_months": "1-3",
   "expertise": "authorization systems",
   "dominant_burden": "revocation/TTL semantics, anchoring, multi-worker consistency"
  },
  {
   "mechanism": "multi-region operations (2 clusters, WireGuard, hourly proofs)",
   "engineer_months": "2-4",
   "expertise": "SRE",
   "dominant_burden": "operations; both clusters currently NOT_OPERATIONAL"
  },
  {
   "mechanism": "evidence program (public verifiers, signed claims, bundles)",
   "engineer_months": "1-3",
   "expertise": "security engineering",
   "dominant_burden": "keeping claims truthful (stale-claim bug found 2026-10-05)"
  }
 ],
 "context": "docs/ip/valuation (2026-10-02/04 internal estimate) put whole-repository rebuild cost around USD 0.65M; that is an earlier session's estimate with its own assumptions and is cited, not re-derived, here."
}