What's New in Eth2 - 30 July 2021

Ben Edgington (Eth2 at ConsenSys — all views expressed are my own)

Edition 74 at eth2.news

Top picks

Happy Ethereum's sixth birthday

, my shadowy super-coder frens

This week, Olympic gold

goes to EIP-3675, "Upgrade consensus to Proof-of-Stake". This formalises The Merge within Ethereum's governance process, and is a huge step towards making it happen.

The other thing that really caught my attention is Lido's Road to Trustless Ethereum Staking. More below.

Calling all stakers
Be sure to update your Eth1 nodes before the London fork on August the 4th/5th!
See below for help.

The Beacon Chain

Eight months in and the beacon chain has passed 200,000 active validators!

This represents 6.4m Ether staked, which is almost 5.5% of total ETH supply. Current staking returns are around 6.1%. I remember when we client devs regarded 100k validators as a big network, but now, even at 200k, my Teku node is barely idling: CPU under 10%, heap memory under 1.3GB, and flawless attestation performance. It's been quite a journey.

Another big milestone: we have not four, but five clients validating on the mainnet beacon chain. Welcome Lodestar!!!

The brilliant Lodestar team (part of ChainSafe) haven't primarily aimed to be a production staking client, but have focused more on the extremely important work of building light client infrastructure for Eth2. Given this, it's all the more impressive to see them actually staking and validating.

Analytics

There's been some good work around measuring diversity and decentralisation on the beacon chain.

First is Miga Labs' crawler dashboard. Some nice stats here, and a choice of crawlers to look at.

Measuring client diversity is very far from being an exact science. We can compare Miga's figures with Stereum's weekly client diverstity watch. For example, Miga has Nimbus at ~0.44% and Teku at ~4.7%, whereas Stereum has Nimbus at 16.8% and Teku at 0.8%. In addition, these count beacon nodes rather than validators, which understates validators that are predominantly used by large staking operations, such as Teku. And there are other issues. So take the numbers with a pinch of salt; the error bars are very large. Nonetheless, it is clear that Prysm continues to heavily dominate the node distribution, and putting the details to one side, the macro view is extremely useful.

Shyam Sridhar of the Ethereum Foundation's Robust Incentives Group has performed a detailed analysis of the decentralisation of the beacon chain, using standardised metrics such as the Gini coefficient, the Nakamoto coefficient and the Herfindahl–Hirschman index.

Terence gives a rather dispiriting tl;dr of the article:

• Based on Nakamoto coefficient
  • 31 entities to control 51% of network
  • 8 entities to control 33% of network
• New validators using 3rd party services is much higher than new solo stakers

Altair

Progress towards the Altair beacon chain upgrade continues steadily. A Beta 2 spec release is out. The only substantive change is around validating empty sync committee signatures. The full diff looks huge as The Merge has now been rebased onto Altair, and some sharding changes are included, neither of which affect the delivery of Altair. I'd have rather these changes be kept on a separate branch until after Altair is deployed, but it's no big deal.

Client teams and the Ethereum Foundation have already run two multiclient Altair devnets to test the fork transition process. While both went ok, neither was flawless. We're shooting for flawless, so kicked off a third devnet (devnet 2) on Thursday. While the fork itself went through fine, there were still some minor niggles between clients.

Despite this, on the implementers' call we decided that we would go ahead with upgrading the Pyrmont testnet in around three weeks time. If you are running validators on Pyrmont, keep a close eye on client releases in the next couple of weeks for when they are Altair compatible, and be sure to upgrade in time.

We may or may not do another devnet beforehand, still TBD.

The Merge

The huuuuge news of the week is the publication of EIP-3675, "Upgrade consensus to Proof-of-Stake" - in my view, the most significant core EIP in Ethereum's history to date, and certainly the most ambitious. This is fabulous work by my colleague Mikhail Kalinin and a cast of supporters.

The Merge now becomes subject to Ethereum's governance process. While the process is not always swift or flawless, there is a huge will to get The Merge done, so I expect a great deal of focus on this over the coming months.

Staking

Stakehouse

I had the absolute pleasure this week of spending some (virtual) time with Colfax from the EthStaker/StakeHouse project. StakeHouse is "focused on building tools that lower the technical bar to staking and promote the health of the beacon chain".

Stakehouse is working on two fronts. First is building the tools that make it simpler to stake and operate a node. For example, Wagyu is an under-development "one-click installer" that supports the four main production clients. Closer to release is Wagyu Key Gen which is a friendly environment for generating your validator keys without needing to use the command line.

The other side of the work involves gathering a community of devs around the project to collaborate together, and to decentralise the development of these tools. To that end there was recently a StakeHouse community call (the first public one, I think). These will come round every two weeks from now on - see the EthStaker discord for announcements. The call had demos of Dappnode, of StakeHouse's Wagyu, and ChainGuardian from NodeFactory (now ChainSafe), with an honourable mention for Stereum (whose v1.1 installer is out!). There's a huge amount of promise here, and I am looking forward with great interest to what will come out of this.

Ensuring that individuals can stake confidently and reliably is vital in a network where centralising tendencies are everywhere. SuperPhiz gets this completely, hence his challenge to the community to build a desktop UI that can manage all four main production clients by the beacon chain's first birthday.

Lido

You may have noticed that I rarely mention large staking pools here. I do take a keen professional interest—we built Teku with staking pools particularly in mind—but, as a community member, my passion is to facilitate the decentralisation of staking as much as possible.

To that end, I was delighted to see the excellent post from Lido this week, presenting their intention to move towards trustless staking, which means decentralising themselves as much as possible.

Lido is becoming a significant force in Eth2 staking, with around 3% 10% (corrected - apparently this is not correct) of all validators, largely as a result of providing liquidity via the stETH token. From the outset, they have made some attempt to decentralise operations, for example by using a variety of node hosting providers and managing them via a DAO. But it remains largely centralised and custodial.

So, I am very heartened by Lido's thoughtful and determined commitment to decentralising themselves and becoming fully trustless. And I love the transparency. Lakshman also has some comments.

Note that Lido is putting its money where its mouth is with a couple of projects:

1. Grants to both Blox Staking and Obol who are both working separately on implementations of SSV (secret shared validators).
2. An RFP for a dashboard showing validator performance across all their node operators.

Tooling

Jacek of Status has another weekend project: building a block explorer that you can run locally against a Nimbus node.

The Great Explainers

EthStaker did a Validator Workshop to help people upgrade their Geth nodes ahead of the imminent London fork on Ethereum mainnet. [The link goes to the start of the workshop, but I recommend rewinding a few minutes and chilling to the intro for a while.]

Vitalik and Georgios of Paradigm wrote some analysis on the susceptibility of the beacon chain to reorganisations (reorgs) after The Merge. This relates to the so-called Time Bandit attacks that caused a fuss a couple of weeks ago, and that Christine and I discussed in last week's podcast. Tl;dr: it's harder to orchestrate a chain reorg under proof of stake.

Justin Bons of Cyber Capital has a good set of reasons why Proof of Stake (PoS) is superior to Proof of Work (PoW) in every meaningful way.

Media and Other Stuff

The Beacon Book

The Beacon Book project is reaching its culmination!

• You can now read all about the book's carefully designed graphics and narratives.
• And you can join the crowdfund, potentially gaining yourself one of the 100 available physical copies. The crowdfund closes on August 3rd at 16:00 UTC. [Reminder - I am not personally benefiting from the crowd fund; my share goes back to Stateful Works for future projects.]

EthCC

Here's a selection of talks mostly from the Eth2 track at EthCC last week. See the agenda for more detailed descriptions.

• SSV: Distributed and Trustless Staking Infrastructure, Alon Muroch
• Managing Eth2 Validators at Scale, Laszlo Szabo
• MEV in Eth2: Introduction & Open Questions, Alex Obadia, Alejo Salles
• An Outlook on Block Explorers and the Beacon Chain, Stefan Starflinger
• Eth2 & Staking: Retrospective and Projections, Robert Drost: I can't find the video, but here are the slides.
• Liquid Staking in Ethereum Security, Vasiliy Shapovalov
• Eth2/Lighthouse Update, Mehdi Zerouali
• It's Time to do Something About Hardware Decentralization, Eduardo Antuña
• Decade After Eth2: Your Post-Eth2 Dreams for Scalability, Defi, NFTs, …, Joseph Chow

But don't limit yourselves to these! There was a ton of great talks, and I am very sad not to have been there

Other media

The Merge made Fortune magazine! Ethereum risks it all on going green. (Unfortunately mostly paywalled. Pun intended.)

On the Coindesk podcast with Christine, we talked about Time Bandit attacks on the 22nd, and had Alex Obadia from Flashbots as a guest on the 29th.

Research

Eth2 contributor Potuz—I believe as part of the Core Devs apprenticeship programme—has formulated a nice approach to speeding up the hashing of the beacon state to calculate state tree roots. Basically, in Merkle-like trees, we only ever hash small amounts of data: 64 bytes at a time. Thus, the padding that's used can be pre-computed for an excellent ~2x boost.

On ethresear.ch, there is a more sophisticated version of the previously known balancing attack on the Gasper beacon chain consensus. The new version claims to be more practical in a network with random delays. The post points out, however, that Eth2 as implemented uses a slightly modified version of Gasper that makes the attack harder to pull off.

Regular Calls

Implementers

Call #69 took place on the 29th of July.

Danny wasn't around for this one, so Alex Stokes chaired. The livestream was not set up, and Zoom recording failed, but a local recording was made that should be uploaded soon.

• Agenda
• Video - not yet uploaded
• My quick notes.

My PC locked up during the discussion towards the end about implementing checkpoint/weak-subjectivity sync. I could hear but could not take notes. Hopefully the recording will have the discussion intact.

In other news

• Danny Ryan's Finalized no. 27, with links to some good PEEPanEIP explainer sessions for Altair.
• Lighthouse update no. 37

---

Follow me on Twitter to hear when the next edition is out 🙌.

 We also have an RSS feed.

Advertising on this newsletter.

Give Feedback.