What's New in Eth2 - 28 March 2021

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

Edition 65 at eth2.news

Top picks

Security minded types, check out the Beacon chain security and testing RFP.

Editorial

Danny Ryan is on a mission to deprecate the use of the terms eth1 and eth2. As the creator and curator of "What's New in Eth2", this is a bit of a blow!

But, on the whole, I agree that these terms are beginning to outlive their usefulness, and are in danger of becoming quite misleading. It's not really about fiddling with names; it's about capturing fundamental changes in direction, and communicating them clearly. Changes in the roadmap are confusing for the community. Clarity in naming aids clarity of understanding.

Let's put this in context with a brief history of naming things in Ethereum's roadmap.

As far as I can tell, the first use of the name "Ethereum 2" was by Vitalik in Hacker News, way back in April 2014. He positioned Ethereum 2 as being a future step-change in the capability of Ethereum through the use of "fundamental cryptographic upgrades".

Ethereum 2 then drops out of view, and the main narrative becomes the familiar Frontier → Homestead → Metropolis → Serenity progression, first described in March 2015. In this scheme, Serenity is the network upgrade to Proof of Stake; Serenity isn't Ethereum 2 per se. Interestingly, sharding/scaling doesn't have a clear place in that roadmap, although it was certainly discussed around that time.

Progress largely followed this plan for some years, with Homestead going live in March 2016, and Metropolis (in name at least) in a pair of upgrades in October 2017 and February 2019. (References).

Serenity was to have been delivered via a process of upgrading the existing Ethereum chain to bootstrap Proof of Stake. That upgrade was specified in EIP-1011 and there was even a testnet that went live on December 31st, 2017.

However, the road to Serenity was found to be a dead-end, not least because of the vast amount of Ether that stakers would require - 1500 was the number back then.

So, in mid 2018, we tore up that roadmap and started afresh with Ethereum 2.0. Rather, we reverted to Vitalik's original idea from 2014 of an Ethereum 2 that was to be a significant cryptographic step-change to the protocol. Calling this Eth 2.0 made sense: we were planning to change pretty much everything about the platform. Ethereum 1 would eventually be brought in as a kind of legacy system. This Ethereum 2.0 roadmap came with neat consecutive phases: Phase 0 for Proof of Stake; Phase 1 for data shards; Phase 2 for executable shards; and, finally, bring Eth1 over to Eth2 after all that.

We've now delivered Phase 0, but the rest of the plan has evolved. As per the latest Eth2 pages on the ethereum.org website, "Eth2 [now] refers to a set of interconnected upgrades that will make Ethereum more scalable, more secure, and more sustainable" (emphasis mine). This looks much more like the old Serenity plan! We're upgrading the Ethereum chain in-place rather than doing the old rip-and-replace.

So, you see, the roadmap narrative has vacillated over the years between effectively doing a replacement of Ethereum 1, referred to at various times as Eth2, and doing in-place upgrades to Eth1. Today, we're more or less back in upgrade territory.

In view of all this, applying the Ethereum 2.0 moniker to the entire project is now seen as confusing: to some it implies a different chain to today's Ethereum; to others, a different coin. It suggests incompatibilities and problems ahead.

The latest naming trend, then, is to limit Eth2 to refer only to the consensus layer upgrade of Ethereum to Proof of Stake, and to refer to all of today's Ethereum (fka Eth1) as the application layer.

While I am reasonably on board with all this, I won't be changing the name of this newsletter. It's too late for that. Sorry, Danny!

The Beacon Chain

PSA Stakers! Update your Eth1 nodes!!! The Ethereum network is set to execute the Berlin upgrade. Be sure to update your Mainnet Eth1 node before April 14th.

The Beacon Chain itself continues to chug along nicely. Really nothing to report. It's seven weeks since we last had a slashing event, so everyone seems to have got the hang of this now.

The two main things going on are the Altair upgrade, expected in June, and planning for the Merge that will put all of Ethereum onto Proof of Stake.

The actual big event that happened this week was the launch of the Prater testnet.

See Danny's Finalized no. 24 for some more brief context on the headlines.

Altair

Altair is the planned mid-year upgrade of the Beacon Chain. Two alpha releases of the spec are out: Stargazer and Half of 'em just look like dots. The celestial theme is strong with this one.

Some nice summary docs have been added to the specs repo:

• Beacon chain changes
• Altair fork - how the fork will be triggered, and the transition process.
• Light client sync protocol - how to build a beacon chain light client.

There are some interesting updates to the rewards for validator attestations that I'll write up another time. It's mostly about reforming the inclusion delay stuff as far as I can see. Currently it's sometimes better for a validator to delay making an attestation to make sure that it votes for the right head block - Altair fixes this unhelpful incentive, among other things.

For a very nice overview of the Altair changes, see the write up in the latest update from Viktor Bunin (Bison Trails).

The Merge

The hottest thing in Eth2-land right now is The Merge. This is the transfer of the current Ethereum chain from Proof of Work to Proof of Stake. As discussed above, we've had various different plans for delivering this over the years. The currently winning proposal is a quick merge that simply replaces the Proof of Work based consensus layer with the Proof of Stake Beacon Chain that's already running.

This week, the quick merge was merged! It's a step towards making it an integral part of the Ethereum 2.0 specification. It's a simple proposal, minimal and not difficult to deliver from a technical point of view. It's a route to getting Ethereum onto Proof of Stake as swiftly as possible, while deferring some work to future upgrades (such as validator withdrawals).

The big question is: when Merge? One option is to focus exclusively on delivering The Merge once EIP-1559 has been done in the London Ethereum upgrade due for July. Aftab's Twitter poll shows huge support for this approach.

I am on record with some reservations about rushing to Proof of Stake. In short, it probably delays everything else by years. But if that's the price of killing Proof of Work sooner rather than later, I suppose it's a cost we just have to bear. Jacek Sieka of Nimbus would like to see a trickle of upgrades to Eth1, while we continue to do plenty of crazy innovation in Eth2.

As per Tim Beiko's All Core Devs update, a decision on timing is likely to be made after the EthGlobal hackathon discussed below, once a Merge testnet has been demonstrated. A finger in the air suggests that Ethereum could be on Proof of Stake by the end of 2021, but more likely Q1 2022.

Where Teku leads, others follow. Sigma Prime announced a demo of a merge transaction: an Ethereum transaction on a Proof of Stake network. And Aditya Asgaonkar has made a Merge PoC in Python, based directly on the executable specs.

There is a plan to use the upcoming EthGlobal scaling Ethereum hackathon to get together as many Eth2 and Eth1 client teams as possible to build a Merge testnet, with a stretch goal of adding sharding, and perhaps even a rollup demo using sharded data.

Testnets

The Prater testnet launched on time on Tuesday 23rd of March. This is the most ambitious yet, with 210,000 validators at genesis, almost double the size of the Beacon Chain Mainnet. The validators are split six ways: 40k each run by the four client teams; 40k run by the Ethereum foundation, split across the four clients; and 10k allocated to various third parties.

Increasing the number of validators increases the state size, increases the amount of work done to process that state, and increases the number of messages being gossipped on the network. Blocks also become fuller, which explores a new kind of constraint as clients need to optimise better for attestation inclusion.

The launch was fairly routine - testnet launches used to be so tense! I didn't even attend this one. A small hiccup as Nimbus's doppelganger detection prevented their validators from participating in the first two epochs (this would not be a problem on Mainnet, since we won't be doing genesis again). A couple of teams had to upgrade their hosting early on: we did that for Teku as it was lagging a little initially. The extra size of Prater has already revealed some hotspots that are easily dealt with and bring the CPU down again: that's what it's for.

You too can join Prater! The Prater Launchpad is up, and recent client releases all support it.

Meanwhile, the Pyrmont testnet continues to run. We are encouraging people to begin using Prater for their testing, and to migrate away from using Pyrmont over the next few weeks. We are planning to use Pyrmont for testing the protocol upgrade process (fka "hard fork"), and then to run it for long periods without finalisation to further stress-test clients. So expect Pyrmont to get a bit unstable in a few weeks' time.

Tooling

Alex Stokes's fork monitor continues to grow in capability. Most recently with the addition of weak subjectivity data.

The Great Explainers

Lots of good stuff on installing and running Eth2 validator clients this week:

• Unvetica of EthStaker has produced fantastic video tutorials for setting up validators, based on Somer Esat's guides.
• Stakehouse, the "one-click" graphical installer project from EthStaker, looks to be coming along nicely. They are looking for help with a length todo list. Breaking news: here's a video demo of Stakehouse!
• Stereum.net has released docker compose, secure docker images, and automation for each of the clients. There's also a one-line Eth2 installer (apply the appropriate degree of caution when piping anything from the web to a shell as the root user).
• And here's a guide to installing Rocket Pool on a Raspberry Pi

Some Tweet-thread explainers from Ethereum Foundation folk. I rather like this bite-sized, short-form approach:

• Lakshman Sankar on state size management
• Lakshman again on light client support
• Alex Stokes on weak subjectivity. For more on this, Meredith Baxter's video demo is superb.

Media and Stuff

Elias Simos continues his wonderful series of profiles of Eth2 devs with Paul Hauner of Lighthouse. Previous editions covered Raul Jordan of Prysm, and yours truly of Teku.

Check out the latest EthStaker update - it's just 5 minutes, but packs in plenty of news. This one is already a little out of date, but keep an eye on the playlist for the freshest news.

Coindesk continues to put out decent Eth2 content:

• Latest Valid Points newsletters: March 17th, Ethereum’s Proof-of-Stake May Happen Sooner Than You Think; March 24th, How Ethereum’s Governance Process Alters ‘The Merge’.
• Latest Mapping Out Eth2 podcasts: March 18th with Superphiz of EthStaker - this is my favourite episode so far! March 25th, just Christine, Will, and myself.

Proof of Stake made it into Wired UK!

Research

Dankrad with High confidence single block confirmations in Casper FFG. Lots of insights into the beacon chain's fork choice handling here. But the practical upshot is that it is easy to detect when the beacon chain is under attack. When it isn't under attack (which ought to be most of the time, since an effective attack needs very large resources), we can achieve very fast finality for all practical purposes, without needing to wait for full economic finality.

Vitalik has taken another run at placing a cap on the number of active validators. The idea is not to limit the total number of validators, but to limit the number active at any one time to half a million or so, which ought to be plenty for security. The more validators active at any time, the more gossip messages need to be processed, and the more work needs to be done by clients to track the state. This increases the resources required to be a validator and thus limits decentralisation.

And CBC Casper is still in view! I don't expect to see this implemented for several years, though, if ever.

Implementers' Call

Call #60 took place on the 25th of March.

• Agenda
• Video
• My quick notes, and Alex Stokes's live Tweet.

Discussion on the future of testnets, research updates, and planning for the EthGlobal Scaling Ethereum hackathon.

Upcoming events

• April 20th, 2021 at 5pm PST: deadline for responses to the Beacon chain security and testing RFP

In other news

• Development update from Prysmatic Labs
• Tweet-form update from the Lodestar team.
• Blox Staking got an audit. Blox has an interesting model, being fully non-custodial: stakers retain control of their private keys.
• The Ethereum Due Diligence Committee has been gathering information from staking service providers. Useful background if you are looking for a host.

---

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

 We also have an RSS feed.

Advertising on this newsletter.

Give Feedback.