The Constellation Debate Part 2, with Brennan Watt (Anza)
Deep dive into Constellation: how multi‑proposer ordering, attesters, and AlpenGlow aim to reduce MEV, tighten latency, and strengthen censorship resistance.
Key Takeaways
- Constellation moves ordering on‑protocol with multiple proposers and attesters, making priority‑fee bidding the main arbiter and reducing single‑leader MEV capture.
- Attesters observe 50ms payload slices, attest to signed shreds, and require roughly half honest attestations; validators dedupe by proposer IDs and can reject blocks missing attestations.
- AlpenGlow targets faster market ticks (200ms slots, two‑slot leader spans) to shorten censorship windows and finality, trading lower latency for higher bandwidth and coding overhead.
- Design aligns economics: stake‑weighted fee sharing, shared cost to deter infinite self‑bidding, and rewards for fast, reliable proposers to incentivize honest performance.
- Replication and DA: proposers replicate data to many nodes with erasure coding; DA payments apply per replication while validators dedupe and execute transactions once.
- Rollout is phased: enable deterministic in‑protocol ordering and AlpenGlow first, gather community feedback (SIMD), then incrementally add IBRL and multi‑proposer features.
Original Source
The Constellation Debate Part 2, with Brennan Watt (Anza)
Visit Source