# cl1-01: Fast Finality

## Related Talks

- [cl2-01](https://hackmd.io/@willcorcoran/H1rNmtF0el): Fast Finality Roadmap
- [cl1-02](https://hackmd.io/@willcorcoran/BywxmtF0gx): Fast Synchronous Finality
- [cl2-03](https://hackmd.io/@willcorcoran/HJY5QYKRge): Fast Finality (whiteboard session 1)
- [cl3-03](https://hackmd.io/@willcorcoran/SJfkVtKRxx): Fast Finality (whiteboard session 2)

## Summary
The Fast Finality Workshop at Jesus College Cambridge focused on advancing Ethereum's protocol consensus for faster economic finality, emphasizing Fast Synchronous Finality (FSF) and related architectures. Participants reviewed key protocols like 3SF (three-slot finality with dynamic availability, optimal resilience but higher complexity), 2SF (two-slot finality via 25% DA adversary threshold, faster but reduced DA resilience), decoupled BFT + RLMD/Goldfish (low block time with committees, trade-offs in asynchrony resilience and complexity), BFT with no DA (simpler, lower thresholds but no DA), and BFT with DA as fallback (questionable practicality). Discussions highlighted property priorities such as block time, voting phases, thresholds (finality 67-80%, BFT adv. 20-33%, DA adv. 25-50%, block production >0-40%), synchronous safety, DA decentralization/resilience to bribery/DoS, short asynchrony resilience, accountable liveness, and protocol complexity (low to very high). Part 2 dove into two-round (or one-round) finality protocols including Kudzu (80% fast path, 60% slow path), Alpenglow (with Turbine for equivocation detection, crash tolerance), Minimmit (80% finality, 40% notarization, potential deadlocks), Hydrangea, Live-Minimmit (guaranteed progress with 40%), and Live-Simplex (one-round with equivocation detection, 60% unlock from deadlocks). Debates centered on dynamic availability vs. complexity, client diversity risks (e.g., 40% thresholds for liveness), geographic latency impacts, slashing (double-vote only, no surround), and user needs for fast on-chain finality (e.g., bridges, exchanges). Objectives included consensus on priorities and narrowing architectures, with calls for user research, specs, and champions for efforts like shorter slots/epochs.

## Key Takeaways
- **Protocol Trade-offs**: 3SF offers high DA resilience (50%) and synchronous safety but medium complexity and high block time; 2SF reduces phases to 2 but lowers DA adv. to 25%; decoupled variants enable low block time via committees but compromise asynchrony or add complexity (e.g., Goldfish lacks short asynchrony resilience).
- **Threshold Dynamics**: Finality often at 67-80%, BFT adv. 20-33%, DA adv. 25-50%; lower thresholds improve liveness (e.g., >0% block production) but risk deadlocks or reduced safety; accountable liveness via inactivity leak possible in most, ensuring recovery post-asynchrony.
- **One/Two-Round Protocols**: Aim for 1-2 slot finality; fast paths (80-95%) for low latency, slow paths (60%) for resilience; equivocation detection (e.g., Turbine, headers) prevents splits; deadlocks possible under asynchrony, unlocked with high participation (60-80%).
- **Dynamic Availability Debate**: DA enables chain progress under low participation but adds complexity and client challenges (e.g., handling multiple chains); room split 50/50 on favoring DA; alternatives like 40% liveness thresholds simplify but limit apocalypse resilience.
- **Practical Considerations**: Geographic decentralization affects round overlaps; client bugs (e.g., Flashbots case) highlight need for diversity; FSF suffices for many apps, economic finality key for bridges/high-value transfers; prioritize fast enough UX over micro-optimizations (e.g., 12 vs. 18s not transformative).
- **Open Challenges**: Balancing latency, reorg resilience, asynchrony; defining "too much" reorg in decoupled setups; engineering for low thresholds (e.g., attestations limit block size); long-term vs. short-term roadmaps (e.g., ZK proving timelines); user research on preferences (stated vs. revealed).

## Speakers
- [Luca Zanolini](https://x.com/luca_zanolini)
- [Yann Vonlanthen](https://x.com/yannvon)
- [Roberto Saltini](https://x.com/robsaltini)
- [Ellie Davidson](https://x.com/ellierdavidson)
- [Francesco d'Amato](https://x.com/fradamt)

## Resources
- [Comparison Table](https://drive.google.com/file/d/1f_-IxuMo-g1hexviriD3yCvbvGN6oRhi/view?usp=drive_link)
- [Slides](https://drive.google.com/file/d/1Ib551tYRZGE7fsr7Jas8HNpL3Pyh_Hca/view?usp=drive_link)
- [Video](https://youtu.be/wdT7B3iBdes)

> back to: [Index Page](https://hackmd.io/H4S9OqgnRUWFzavoOhy4aw)

*Note: Summaries were generated in part with the help of AI, so names and terms may not be 100% accurate.*