HackMDWhat's New in Eth2 - 8 April 2022
- The image file may be corrupted
- The server hosting the image is unavailable
- The image path is incorrect
- The image format is not supported
Edition 91 at eth2.news
Top picks
Two gems for you this time, both from Tim Beiko:
- This one just missed my last edition, and is a couple of weeks old now, but nonetheless remains a must-read: AllCoreDevs Update 010.
- The Ethereum Roadmap FAQ which Tim created in response to this Twitter thread.
The Beacon Chain
All quiet on the beacon chain :eyes:
Some are excited that Prysm's supermajority (>67% share) on the network seems to be history now. That's good, and pulls the sting of the worst potential outcomes. But we're still a very long way from seeing a healthy distribution.
The Merge
It's getting a little hard to keep up with the details of all the Merge testing, testnets, shadow forks and wotnot. Here's a summary (Discord) from a week ago on Goerli shadow fork stuff, but things have already moved on from there. If you want to drink from the fire hose then the Eth R&D Discord is the place to be. See the "Merge" channels.
In any case, the lay of the land as I see it is like this:
- Do a couple of Mainnet shadow forks, one next week, one starting at the end of Devconnect. The shadow forks are continuing to flush out some issues in clients, though nothing very serious yet.
- Go/no-go decision on doing the Merge or postponing the difficulty bomb to be discussed on the 29th of April ACD call.
- If it's "go" then start merging the existing testnets at 2 week intervals with a view to doing the real thing in July.
Beyond the Merge
Alex Stokes recorded a PeepAnEIP on stake withdrawals (EIP-4895, updated). I don't think the video is up yet, but Alex's slides are available. And he made some Tweet-thread notes to go alongside.
I told you last time that I didn't know of a plan to migrate 0x00 withdrawal credentials to 0x01 creds (which will be required for automatic rewards harvesting). Turns out that it was here the whole time.
Away from withdrawals, one interesting issue we currently have is that the Eth1 client a beacon node is attached to must keep all its historical data back to the deployment of the deposit contract in Q4 2020. This is because, in the current design, to sync up a new beacon node via a checkpoint sync still requires all the deposits to be re-processed in order (though I believe Nimbus has their own workaround for this). The problem is that Eth1 nodes would rather like to be able to drop old history (see EIP-4444). Enter EIP-4881 by Mark Mackey of Lighthouse which seeks to define a standard Deposit Contract Snapshot Interface :raised_hands:
Staking
Somer Esat's incredibly valuable staking guides have had an update.
Evan Jones has written a paper on Accelerating Validator Returns Using Rocket Pool. Spoiler: he likes it.
Another thing to like about Rocket Pool is their new terminal UI. Superphiz and JCRTP gave it a whirl in this video walkthrough. "It's an extremely simple and intuitive way to run Rocket Pool and stake with a graphical interface!" (Phiz on Twitter).
Adrian Sutton took the opportunity of adding 10,000 more Teku validators to the Prater testnet to do same analysis on how the load on a beacon node varies with the number of validators it hosts: the cost of adding validators. The tl;dr is that, up to about 70 validators on a node, the load increases in proportion to the number; above that number the load flattens out. The reasons are interesting, and you should read the article. The obvious question is, doesn't this create centralising pressure? I mean, we are running 5000 validators on that node for the same cost as 64 validators, so our cost per validator is much much lower. It is true that this is a factor, but the cost of running a node is so small compared with the cost of the stake and the rewards available that it's only a very minor factor in the grand scheme of things.
And How to Stake on Ethereum, April 2022 Edition; Superphiz updated his high-level overview.
The Great Explainers
Jim McDonald is Attestant's great explainer, and his latest offering is an article on Understanding Post-Merge Rewards, including a section on plans for stake withdrawals.
Guillaume Ballet continues to pioneer the implementation of Verkle trees for managing Ethereum's state as a prelude to having stateless clients. Here's his short explainer on the required changes to the state tree structure. He is posting regular updates on the progress of his Verkle tree testnet.
Media and stuff
My old friend and podcast co-host, Christine Kim, continues to produce insightful and well-researched reports in her current role at Galaxy Digital Research. The latest is What to Expect from Ethereum's Merge Update, published today. It's really too good to put under "media and stuff", but I don't really have a category for "must read research reports", so here it is.
We're dealing with misconceptions this week! Fortune dispenses with 3 misconceptions about the ‘merge,’, while David Hoffman of Bankless goes one better in dealing with 4 Misconceptions about PoS vs PoW.[1]
Crypto Briefing interviewed Trent Van Epps about the Merge.
Finally, Vitalik's article on The roads not taken is well worth a read. He asks in one section, Should we have gone with a much simpler version of proof of stake? It's an interesting set of reflections.
Research
Vitalik published a method for simplified single secret leader election. It's been an interesting journey over the last four years or so from "we have no idea how to do this", through "here's a complex and expensive way it might be done", to "we now have some viable options". Progress only moves one way.
Regular Calls
Implementers
Call #85 took place on the 7th of April, ably chaired by Mikhail Kalinin in Danny's absence.
We covered the Goerli shadow forks and future mainnet shadow fork plans, regular client team updates, a recap on Merge timing decision making (go/no-go on delaying the difficulty bomb by the end of April), and got into a good discussion on the Builder API (formerly known as MEV-Boost). Plus some other bits and pieces.
- Agenda
- Video
- My quick notes, and from Christine.
All Core Devs
ACD call #135 took place on the first of April.
- Agenda
- Video
- Excellent notes from both Tim Beiko and Christine Kim.
Upcoming events
- 18-25 April: Devconnect.
In other news
- Stereum update number 5 covering, "the finalization of Stereum 2.0 and our preparation for the devconnect".
- Nimbus is serving light client data on the Prater testnet.
And finally…
Looking forward to Devconnect! Aside from the Amphora dev team gathering in Greece back in October this will be my first trip out for over two years. For a couple of decades I used to get on a plane three or four times a month. Tbh, I can't say I miss that, but two years is too long.
Anyway, I can't wait to meet up with friends old and new. Almost the entire Teku team will be there; I will be meeting some of them in person for the very first time.
Do come and say hi, or even buy me a beer - I'll be floating around all week :beers:
Learn More →
Learn More →
Advertising on this newsletter.
The "equity" vs "commodity" money distinction discussed in that article doesn't seem very interesting to me. I felt that the recent PoS vs PoW debate between Justin Drake and Lyn Alden was a bit spoiled as this was the only thing the Bankless guys wanted to talk about, while neither of the debaters seemed at all interested by it. ↩︎
