Thank you to everyone who has shared their perspective on this thread so far!
Putting it all together, there are three overarching themes that emerge:
- Increasing the frequency of network upgrades (and limiting scope to achieve this)
- Agreeing on clear technical standards for EIPs at various stages of the process, from their initial introduction to AllCoreDevs to their inclusion in an upgrade and eventual deployment
- Higher-order process reforms for ACD
I think we can and should make progress on all three tracks in parallel!
Starting with (1), our guiding question should be “what would it take to ship Fusaka ~6 months after Pectra”? My personal answer here is that the scope for Fusaka should be limited to the already SFI’d EIPs: PeerDAS and EOF.
While this may not be the “perfect fork”, it undoubtedly advances Ethereum’s roadmap by scaling L2 blob capacity. Additionally, PeerDAS and EOF would both meet any technical bar we’d want to impose on EIPs for consideration, as they already run on multi-client devnets.
This brings me to (2). Assuming we agree to freeze the Fusaka scope ~now, then the first step before debating the Glamsterdam’s scope should be to agree on what the technical requirements for proposed EIPs should be. Hopefully, we can agree on that relatively rapidly, leaving several months for us to discuss various proposals (and for their champions to get them up to our standards).
In parallel to this, we should have a deeper discussion about (3). I’m excited about the different proposals for more radical rethinkings of AllCoreDevs but want to make sure we don’t rush to implement a whole new system without fully thinking thought through its second order effects. Realistically, it will take many months to get people onboard with a new way of working and I think we should accept that rather than naively expect to overhaul ACD overnight.
Putting all of this together, my tentative proposal would be to:
- Try and ship smaller, faster forks, starting with Fusaka
- Agree on a set of minor reforms (like those mentioned above) for the Glamsterdam planning cycle in the next few weeks
- Kickstart an effort around a broader ACD reform, with the expectation that more substantial process changes would be proposed for the next fork planning cycle, hopefully in ~6 months.