Fast Confirmation Rule (FCR) #6, April 14, 2026

Protocol Calls & happenings
1

Agenda

:warning: NO CALL ON TUESDAY MARCH 31 — EthCC. NEXT CALL: APRIL 14.

Bi-weekly, Tuesdays @ 12:00 UTC [NEW TIME]

Client Implementation Status

Client Status Contact
Prysm Old algo impl; 25% on latest spec Terence
Lighthouse EPF PoC; running test vectors —
Teku PoC complete; production dev starting Anton
Nimbus Spec reviewed; active in spec PRs (Etan finding edge cases) Etan
Lodestar Spec tests passing; running test vectors; deploying to staking env Nazar
Grandine TBD —

Testing

  • Format: Extended fork choice tests (Altair+)
  • Status: is_one_confirmed tests merged; previous-epoch block confirmation tests nearing completion. Test vectors published — Lighthouse and Lodestar running them successfully.
  • Coverage: Targeting ~90-95%
  • Paper: Readability improvements in progress; no algorithm changes
  • Repo: [consensus-spec-tests TBD]

Call 05 Summary (Mar 17, 2026)

Headline

Spec approaching finalization — three changes landed since Call 04 (refactor, underflow fix, long-standing discrepancy fix), with Etan from Nimbus responsible for catching two of them. Mikhail targeting spec merge in the coming weeks; may happen asynchronously around EthCC. Test vectors published and running on Lighthouse and Lodestar. Website (fastconfirm.it) shipped to positive reception; inbound partner conversations underway. Metrics standardization surfaced as a new workstream. No client devs on the call again.

Client Updates

  • No client developers present this call.
  • John Church (Nethermind) created a dashboard tracking client FCR implementation status — currently shows Lodestar and Lighthouse as having enabled support.

Spec Changes

PR What Status
#28 Adversarial threshold for honest FFG support — last substantial spec change expected before finalization Merged (test fix landed)
Refactor PR Refactor — introduced underflow found by Etan (unsigned integer subtraction). Fix: return zero as safeguard. Merged + follow-up fix merged
Bug fix Long-standing spec discrepancy — present since initial spec. Also caught by Etan. Merged

Spec finalization target: Next couple of weeks. Mikhail to update test format and fix Gloss test failures that broke due to fork choice function updates. Merge may happen asynchronously given EthCC timing.

Spec relationship to consensus-specs: FCR is standalone but extends the fork choice Store with new variables (e.g., canonical chain head at previous slot start) and modifies the safe block hash passed from CL → EL. Backwards/forwards compatible with hard forks. Closest analog: weak subjectivity period spec.

Testing & Research

  • Previous-epoch block confirmation tests in progress; Luca targeting completion by end of week (mid-March).
  • Test vectors published; Lighthouse and Lodestar running them successfully.
  • Testing does not depend on the research paper — algorithm is at final state.
  • Paper: Roberto and Luca improving readability (soft introduction to lemma series). No algorithm changes foreseen.
  • Underflow edge case: returning zero is sufficient; dedicated test can be written later.

GTM & Outreach

  • Website launched (fastconfirm.it) — positive feedback. Call-to-action intake form generating inbound partner conversations.
  • RPC Standards call (Mar 23): Julian to present FCR for awareness. RPC providers don’t need to act yet, but should know about it for when customers ask. Can point downstream customers to the website for security difference explanations.
  • Bridge feedback: RPCs could compete on fastest fast-confirmation delivery (more interesting with ePBS as attestations arrive earlier in the block).
  • EthCC booklets: Being printed for distribution at Kin. Mikhail, Julian, Will to divvy up on-site.

Metrics Standardization (New)

  • Mikhail raised standardizing observability metrics (drawn from Lighthouse implementation) across clients — goal is a common metric set so different clients can be monitored on the same dashboard.
  • Not enforced, but a “should implement” standard so that if a client implements metrics, they match.
  • Will suggested contacting Katya (experienced with client-agnostic metrics from PQ Interop / Pyridos).
  • Key open question: which metrics are most meaningful? Should also factor in what downstream users/partners want to measure.

Action Items

Owner Item
Mikhail Finalize spec; update test format; fix Gloss test failures
Luca Complete previous-epoch block confirmation tests; publish updated test vectors
Luca + Roberto Continue paper readability improvements
Mikhail Draft proposed list of standardized FCR metrics; reach out to Katya
Julian Present FCR at RPC Standards call (Mar 23)
Julian + Will Continue partner intake conversations from website leads
Will Coordinate EthCC booklet printing with Digital Studio (Nuno)
Will Continue async outreach to client teams — attendance remains a priority
All client devs Review latest spec changes (refactor, underflow fix, discrepancy fix)

Agenda for Call 06

A. Client updates (round-robin)
B. Spec status — merged to consensus-specs? Any post-merge issues?
C. Testing — previous-epoch tests complete? Updated test vectors? Coverage assessment
D. Paper status
E. Metrics standardization — proposed metric list? Katya contact made?
F. RPC Standards call debrief (Julian)
G. GTM update — partner conversations, EthCC takeaways
H. Lodestar staking env metrics (Nazar — carried from Call 04)
I. Open discussion

Add agenda items: Comment below with topic

Meeting Time: Tuesday, April 14, 2026 at 12:00 UTC (60 minutes)

GitHub Issue

2

Meeting Summary:

The meeting focused on updates regarding the Fast Confirmation Rule (FCR) implementation and specification. Mikhail provided an overview of recent changes to the FCR spec, including adjustments to restart conditions, epoch validation, and the introduction of a fast confirmation store to reduce dependencies. He noted that the spec is in its final version and not expected to change further. Mercy reported that Gwendon has a working PR for FCR and has validated end-to-end testing against Luster in a live DevNet. Nimbus has already released their FCR implementation, while other clients like Lighthouse and Lodestar are waiting for the spec to be merged before proceeding with their implementations. Barnabas expressed concerns about potential conflicts between EPPS and FCR implementations. Luca mentioned that their technical paper on FCR is complete and they are gathering feedback, with plans to address the comments and release the paper soon. The group discussed the current status of EPPS implementation across different clients and its impact on the timeline for FCR adoption.

Click to expand detailed summary

The team discussed the status of EPPS implementation across different clients. Mikhail shared that he was working on finalizing a confirmation PR for the spec, while Barnabas noted that client implementations were still behind schedule and not ready for interop. Mercy provided an update on Gwendon’s progress, mentioning they had a working PR for FCF that had been validated end-to-end against Luster in a live DevNet. The discussion concluded with Will inviting Mikhail to provide updates on the spec work and other client teams’ progress.

The team discussed the status of the Lodestar specification, which is currently in the final round of reviews. Mikhail explained that they decided to merge the current PR first before implementing GLOS compatibility, as the GLOS implementation introduces backwards-incompatible changes to the fork choice abstractions that need to be addressed. Barnabas asked about the configuration parameter CONFIRMATION_BYZANTINE_THRESHOLD, questioning whether it should be a config or preset, but the discussion was cut off before Mikhail could fully address the question.

Mikhail explained that the adversarial threshold for the Byzantine threshold is configurable between 0-25% to balance safety and efficiency, with backtesting currently based on the conservative 25% threshold. He outlined several recent changes to the fast confirmation rule specification, including switching from a “greater than or equal to 1/3” to “strictly greater than 1/3” comparison, adjusting restart conditions to use block epochs instead of checkpoint epochs, and introducing a fast confirmation store to reduce dependencies. Mikhail also described the addition of optimistic sync functionality to prevent confirming blocks not yet validated by the execution layer client.

Mikhail announced the release of Nimbus v26.3.1, which includes the FOST Confirmation Rule implementation, and requested more code reviews to help get it merged. He provided an update on backtesting analysis work conducted by Sam from PanOps and Anton from Consensus, which showed similar results to the initial performance measurements. Luca reported that their technical paper on the FCR implementation is complete and receiving positive feedback, with plans to address the feedback and release the paper soon. The group confirmed that the FCR specification is finalized and not expected to change.

The team discussed pending approvals from client developers and Justin, with Mikhail noting that Justin’s feedback had been addressed. Mercy asked about available client images for testing, expressing a desire to expand beyond Lodestar to include other clients. The team also touched on reviewing client implementations and a potential stable release for Nimbus, though specific details about timing remained unclear.

The team discussed the implementation status of FCR and related projects. Mikhail explained that while Lighthouse and Lostar are ready to implement FCR, they are waiting for the specification merge as a signal, though there are concerns about potential conflicts with ongoing EPPS implementations. Barnabas raised concerns about merging FCR with EPPS, suggesting that once EPPS becomes more stable, other projects like FCR can proceed. Mikhail also mentioned plans to revisit abstraction changes in the main specification before making FCR compatible with GLOS, noting this would not happen soon.

Next Steps:

  • Client developers: Review and approve the Fast Confirmation PR to help get it merged
  • Mikhail: Get approval from Justin for the Fast Confirmation PR
  • Mikhail: Work on revisiting folk choice abstractions in the main spec before opening FCR PR to make it compatible with GLOS
  • Mercy: Test with additional client images beyond Lodestar to increase client coverage
  • Sam and Anton: Share detailed backtesting and analysis results on Fast Confirmation delay in the near future
  • Luca and Roberto: Address feedbacks received on the technical paper and release it soon

Recording Access:

3

YouTube recording available: https://youtu.be/3U_Vm78QKLM