FOCIL Breakout Room, February 25th, 2025

[soispoke]

Schedule

• Tuesday February 25th, 2025 @ 2pm UTC
• Duration 60 min
• Recurring meeting already on Ethereum Calendar
  Zoom Link

Agenda

• FOCIl and scaling https://x.com/adietrichs/status/1892951240524403089

• Francesco's doc on delayed execution and FOCIL: https://hackmd.io/UX7Vhsv8RTy8I49Uxez3Ng

• Discussions about IL building rules on Discord and here: https://ethereum-magicians.org/t/eip-7805-committee-based-fork-choice-enforced-inclusion-lists-focil/21578/6?u=soispoke

• "Wrong fork domain used" issue

[github-actions]

Discourse Topic ID: 22980

• Action: Created
• URL: https://ethereum-magicians.org/t/22980

[Marchhill]

I'd also like to discuss a few fixes I made in a PR to the spec and get some review. The most important one is the payload being valid if it doesn't satisfy the IL.
For the IL building rules I have made another PR on top of this to propose some changes

[poojaranjan]

Recording:
https://www.youtube.com/watch?v=UW9vA3FIYn8

[jihoonsong]

Research

FOCIL and Scaling

• There was a thread on solo stakers and censorsip resistance on Twitter. We want to scale Ethereum without compromising verifiability and censorship resistance. While Ethereum’s censorship resistance relies on solo stakers who locally build blocks, they currently forgo potential profits from MEV-boost to uphold censorship resistance. FOCIL eliminates this trade-off between profits and values by allowing solo stakers to continue influencing block content decisions without sacrificing revenue. Consequently, FOCIL opens the door to more effective scaling solutions such as deeper exploration into PBS and APS.

Compatibilities with Other Proposals

• Francesco suggested a way to make FOCIL be compatible with Delayed Execution. It enables “dry-run” IL validation during the static validation phase. It adds a bitfield over the IL committee to a block and a proposer marks the bits corresponding to the ILs considered during block construction. During the static validation phase, attesters check whether the ILs specified in the bitfield form a superset of the ILs they collected. They do not verify that all IL transactions are included; they focus solely on whether the bitfield matches or exceeds their own IL set. In the next slot, the proposer confirms whether the head block contains all valid transactions from the ILs specified in the bitfield. If it does, the proposer extends the head block; otherwise, they extend the parent block. Attesters follow the same process to determine whether to vote for the head block or its parent.

• It's similar to an approach to make FOCIL be compatible with ePBS.

Adding Randomness to IL Building Rules to Optimize Throughput

• Marc proposed an IL building approach that leverages randomness to reduce overlap among ILs and optimize throughput. In this approach, each IL committee member is assigned an ID from 0 to f, and the member with ID a is instructed to favor transactions whose hashes begin with a. More details can be found in the Ethereum Magicians post and this PR.

• Terence suggested waiting until we have a better understanding of how client diversity plays out before making any premature optimizations. In the meantime, we could further study this topic. However, Terence also agreed that the EIP should provide clearer recommendations on the properties each client can reference when implementing IL construction.

• Mark’s proposal could be a viable solution if we decide to prioritize throughput, but we could use more research before making any changes. In the meantime, we can update the EIP to include clearer recommendations on which properties to consider when implementing IL construction.

The Interop Between Prysm and Lodestar

• Prysm and Lodestar are currently failing to interop due to an invalid signature issue, possibly related to signing over the wrong fork version or another cause. We are investigating the problem, and once it's resolved, we expect to have the first interop between CLs.

Implementation Updates

• Geth has fixed bugs to avoid marking non-IL-compliant blocks as invalid and let them be reorged instead, and to support the transition from Electra to Fulu and subsequently to eip7805 fork.
• Nethermind has opened a PR to flesh out engine API changes in the EIP, has reported a bug in Geth where non-IL-compliant blocks were marked as invalid instead of remaining valid and being reorged, and fixed it in Nethermind's implementation as well.
• Lodestar has rebased from Electra to Fulu in preparation for interop with Prysm, primarily has been testing, and is working on activating eip7805 fork in the same epoch as Fulu.
• Teku is working on interop between two Teku nodes using stub ELs.
• Prysm has no update.
• Reth has no update.
• Lighthouse was absent during the call.
• Katya is developing a metrics dashboard with Prysm and maybe have it ready to share during the next breakout session.

Consensus Hong Kong

• Jihoon delivered an ELI5 presentation on FOCIL at Consensus Hong Kong. The recording is available here, starting at approximately 28:13.

Links

• A thread on FOCIL and Scaling
• IL Tx Scoring Function in FOCIL
• epbs/DAS/FOCIL all-in-1 fork-choice
• FOCIL talk at Consensus Hong Kong
• Adding randomness to IL building rules
• A PR to the EIP that expands on EL changes
• A PR to the EIP that adds properties to consider for IL construction