Three Things #36: September 25, 2022
Three reasons to be excited about the Ethereum Merge
Ten days ago, around 2am ET, the Ethereum “Merge” was activated. At that moment, the security of the Ethereum protocol was switched “in flight” from proof of work to proof of stake. Ethereans around the world watched on several livestream calls, the largest of which had over 70,000 simultaneous viewers. People in several places, including the EF Berlin office, the de facto Ethereum HQ, erupted in cheers when the Merge was triggered, and then again a few minutes later when the first post-Merge epoch was finalized, signaling that the Merge was successful. It represented the culmination of over seven years of research and over five years of development, and it represented a feat that was never done before: changing the consensus mechanism of a live network. And not just any network, but the largest, most important smart contract platform.
I have many reservations about proof of stake. I’ve written and spoken about it many times, including here in this newsletter. Those reservations haven’t gone away, but I nevertheless think it’s good that the Merge happened, and was successful, for the following reasons.
Thing #1: A Triumph for Open Source
Take it from someone who knows: Ethereum has not been built in the usual fashion. It wasn’t built by a company, or by a single team, or even by a community in the traditional sense of people who regularly see each other face to face. Ethereum was built by an extraordinarily large, diverse global community of researchers, developers, and many others, with loose coordination provided by the Ethereum Foundation. This was true of “Ethereum 1” and it’s doubly true of the new, post-Merge Ethereum. The process has been very messy. Dozens of teams in different timezones, funded by different sources and business models, operating under different management structures, etc. coordinate via tools like Discord, GitHub, biweekly video calls, and occasional in person gatherings at conferences and hackathons.
If you asked any seasoned, professional software engineer how to build complex, mission-critical software, this is not what they would suggest. And, indeed, many professionals have looked at Ethereum and opined that this could never work. Yet, it has, and year after year it’s proven the haters wrong. It continues to grow, develop, and thrive. In spite of a massive increase in competition in recent years, Ethereum is still head and shoulders above the rest by several metrics: market cap, most obviously, but also number and quality of applications, and number of active developers. There’s also that metric that’s very hard to quantify, but you know it when you see it (especially at in person events): ecosystem health. Simply put, Ethereum has for a long time and continues to attract the best people. The difference compared to events I’ve attended from other ecosystems, which are usually (not always, but usually) driven by profit-seekers, is astounding.
Ethereum is a phenomenon that grew out of open source and the idea of commons-based peer production. Linux was the first large-scale, complex project that proved the model, and ultimately won when it defeated Windows (the bazaar to Windows’s cathedral) to become the most popular operating system. Bitcoin created money in the same fashion, but it’s not nearly as technically complex as Ethereum, far fewer people work on it, and innovation has slowed to a snail’s pace in Bitcoin as compared to Ethereum. Put simply, Ethereum is a triumph of the open source movement and of the potential of wide scale, distributed coordination and development.
The Merge proves two things about blockchain and about open source. The first is that blockchains can continue to evolve in major ways even many years after launching, and after achieving scale and market dominance. If Ethereum were run by a large corporation, it would’ve shifted from innovation mode to monetization (extraction) mode long ago, and would be raising fees to milk as much profit as possible from the v1 product. Because it’s not, and because there’s no for-profit corporation behind Ethereum, it can continue to evolve in exciting ways. (Although some make the case that it is, in fact, evolving in ways that provide economic benefit to Eth holders at the expense of others, which is probably true and not surprising.)
The second is that decentralized governance can be quite effective. Ethereum governance is messy and far from perfect—I feel about it the way I feel about proof of stake, I have major reservations about it—but it has delivered a series of major, successful upgrades over the past few years, from the launch of the Beacon Chain to EIP-1559 to, now, the Merge. (Whether it will be so successful at dealing with issues of a less technical nature, when even more is at stake, remains to be seen.)
The Merge cements Ethereum’s place as far and away the most important blockchain, and open source project, in the world. It’s Ethereum’s race to lose (as it has been for some time).
For more: Read this piece from a few years ago on how to get started contributing to Ethereum.
Thing #2: Decentralized Block Production
I’ve been running nodes on Ethereum and other networks for years. But until this past week I had never actually mined a block. A few days after the Merge, my validator was randomly selected to propose a block. The fees from the block landed in my account. It sounds trivial, but it’s actually kind of a big deal. After being part of the Ethereum project for years, working as a core developer, running nodes and infrastructure, the simple act of mining a block myself wasn’t possible until last week.
The jury’s still out on whether proof of stake is better or worse for decentralization. It’s complex and there are factors pulling in both directions. Also, Ethereum proof of stake is quite different, and in theory more decentralized, than the delegated proof of stake that most networks run. As I wrote about here previously, the majority of the stake in Ethereum is controlled by a tiny number of organizations, which makes censorship a real concern. But the fact remains that I can now actually mine blocks—and if I can, I am not alone. Thousands of others running validators at home can too.
But, even with this centralization of stake, Ethereum-style PoS is probably not worse than Bitcoin mining, which is also highly centralized into a small number of pools whose centralized operators can in theory censor, collude, or engage in all sorts of other bad behavior.
It’s also interesting to note that the move towards increased decentralization of block production actually contradicts the Endgame thesis, which suggests that centralization of block production is inevitable. If you consider all the factors at work, including MEV, that conclusion may still be unavoidable over the long term, at least in Ethereum. I’m really curious to see if it becomes easier or harder for home miners to produce blocks in future.
For more: Read this tweet thread.
Thing #3: The First Serious Test of PoS
The first large scale, production proof of stake networks went live a few years ago, beginning with networks like Cosmos, Tezos, Eos, NEAR, BNB, Celo, Solana, etc.. Nearly every “Ethereum killer” network that’s launched in the past couple of years uses proof of stake, for good reason. Proof of work is obviously very energy intensive and brings with it the risk of 51% attacks of hashpower borrowed or moved from other networks. Rolling your own consensus mechanism is (as we’ve learned the hard way at Spacemesh) a huge, complex process that I don’t wish upon anyone. Proof of stake is relatively straightforward and works out of the box.
However, all of the large, production networks that I’m aware of are not, in fact, using “direct” proof of stake, they’re using a variant called delegated proof of stake (dPoS). This means that, rather than running their own validators, ordinary users instead delegate their stake to a professional validator. There are a relatively small number of these validator operators—just as there are a small number of PoW mining pools—which makes communicating and coordinating with them straightforward. It also makes the network more scalable (since fewer nodes need to participate in consensus) and, in theory, more reliable (since the people operating those validators are professionals, have redundant setups, etc.).
All of this comes at the cost of decentralization. In most cases, there aren’t more than a few dozen validators. The consensus mechanisms that underlie proof of stake, such as Tendermint, only support on the order of a hundred validators. This might be good for efficiency and scalability but it’s not good for censorship resistance, especially in a world of reckless OFAC sanctions.
Ethereum-flavor proof of stake is different. Anyone can run their own node relatively easily and cheaply. It does require 32 ETH (currently around $40,000), but that’s a far cry from the previous requirement of thousands of ETH, and from what it costs to run a validator in most dPoS systems, which tends to run into the hundreds of thousands or millions of dollars, especially once you factor in the infrastructure cost. And, unlike Tendermint and its related protocols, the protocol underlying Ethereum supports hundreds of thousands of validators. This is no longer theoretical: the Ethereum Beacon Chain has been running successfully for nearly two years with no downtime, and since the Merge it’s now providing security to the entire Ethereum network.
The Merge is thus the first real test of proof of stake in two important ways: one, it’s the first test of “real,” decentralized proof of stake. Two, it’s by far the largest network secured by proof of stake in terms of value. So far, so good. But it’s still early days.
For more: Take a look at the following sites for statistics and data on validators and stake: Ethereum, Solana, Tezos, Cosmos.