Consensus with Asymmetric Trust
summary
Argues that consensus and replicated storage work when each node declares its own fail‑prone sets, formalizing a asymmetric quorum system that generalizes traditional Byzantine quorums, and gives concrete register and broadcast protocols that simply replace symmetric quorums with per‑node quorums. Proves existence iff a global Q3 property holds, distinguishes “wise” vs “naive” nodes and the need for a guild for liveness, and shows Ripple/Stellar style configs can stall in practice if quorum assumptions are misconfigured.
AI-written summary · about
chapters (14)
- 0:00Introduction and history
- 3:49Understanding distributed trust
- 5:25Fundamentals of threshold trust
- 8:20Safety and liveness properties
- 10:54Read-write register abstraction
- 13:06Consistency and concurrency
- 17:33Byzantine fault-tolerant protocols
- 22:29Generalized quorum systems
- 26:36Asymmetric trust systems
- 34:05Blockchain consensus models
- 38:41Stellar and federated quorums
- 45:20Defining asymmetric quorums
- 50:48Generalizing protocols for trust
- 56:54Discussion and conclusions
related talks
34mWhat properties should asymmetric quorum systems satisfy?
56mA sledgehammer to crack a nut: why blockchain is not (always) a good idea
43mWhat We Talk About When We Talk About Distributed Systems
18mAdoB: Bridging Benign and Byzantine Consensus with Atomic Distributed Objects
43mAll Together Now: An Introduction to Distributed Consensus
22mLiberating Distributed Consensus
42mBeyond Blockchain: Convergent Consensus
48mDistributed Consensus Algorithms