Stochastic reasoning in decentralised systems
summary
Argues that choosing committees by a fair beacon with ε = 10⁻¹⁸ per trial yields one billion trials → δ ≈ 10⁻⁹ overall, enabling deterministic reasoning about consensus. Gives concrete committee sizes for 1%, 5%, 10% bad-node rates, describes a two-level consensus plus handover protocol and "hooks" (forward pointers) to make forks stochastically impossible. Models and a small blockchain prototype are checked with CSP and FDR, and random beacons are assumed bulletproof random beacons for fairness.
AI-written summary · about
chapters (9)
- 0:00Stochastic certainty intro
- 4:35Stochastic blockchain proof
- 12:29Byzantine agreement assumptions
- 18:07Committee selection choices
- 23:32Blockchain probabilistic forks
- 26:40Two-level consensus protocols
- 28:06Verification with consensus machines
- 30:35Decentralized systems book
- 33:45Q&A
related talks
02h 44mBlockchain consensus mechanisms
29mBuilding Probabilistic BFT Protocols Through Subsampled Voting
23mProduct Driven R&D Strategy and Report
42mBeyond Blockchain: Convergent Consensus
40mConsensus Algorithms in Distributed Systems
47mCant we just agree
56mConsensus Through Herding
32mAnalysis of Nakamoto Consensus