Beyond Blockchain: Convergent Consensus
summary
Claims a CRDT-style consensus called Convergent Proof of Stake that merges signed per-peer orderings (Merkle-tree beliefs), solves history bloat with de‑duplication and content‑addressed storage, and confirms commits in three rounds of gossip (sub‑second in practice). It pairs with a CVM and Convex Lisp runtime, reports million transactions per second locally, enforces memory accounting and a token fee, and aims for Byzantine fault tolerance (assumes <1/3 malicious).
AI-written summary · about
chapters (25)
- 0:00Intro
- 2:03Some Requirements
- 3:23Convergence example
- 5:05Three requirements for a CRDT merge function
- 5:47Some examples of CRDTS
- 6:29Ideally we want universal state transitions
- 7:52Voting Example
- 8:44One solution is Blockchain
- 10:28Not a good solution
- 12:30What happens if somebody adds a new block?
- 13:35Fault Tolerance
- 14:04What if blocks conflict?
- 15:54Simultaneous confirmation
- 17:13We can now define a merge function!
- 18:55One big problem?
- 19:44Data structures to the rescue
- 21:04Optimising Beliefs
- 23:31Convergent Proof of Stake is close to optimal
- 25:09Convex Virtual Machine
- 26:01Structure of Global State
- 27:23Data Types
- 27:54Every Account is Lisp Machine!
- 29:37Smart Contracts - Vote Counting Actor
- 32:27Smart Contracts - Usage from User Account
- 33:32A little necessary economics
related talks
47mCant we just agree
43mAll Together Now: An Introduction to Distributed Consensus
02h 44mBlockchain consensus mechanisms
23mBlockchain
23mPerishable Proof of Transactions - New Consensus for Day to Day dApps
56mA sledgehammer to crack a nut: why blockchain is not (always) a good idea
56mConsensus Through Herding
31mProofless Consensus & Client-side Validation