Welcome to Valid Points. In today's issue, Sam Kessler discusses the race between Polygon and Matter Labs in bringing their zkEVMs to market and the compromises both parties will need to make in the name of security. The race to bring the first zero-knowledge Ethereum Virtual Machine (zkEVM) to market culminated in a meme-laden Twitter spat between two crypto co-founders last week. Polygon’s Miahilo Bjelic and Matter Labs’ Alex Gulechov are competing to launch the “first” zkEVM chain to help scale Ethereum. When both co-founders happened to announce major updates to their timelines last week, each of them used the occasion to jab at the security practices of their competitor. After Gulechov announced that zkSync Era – the Matter Labs platform – was open-sourcing its code and opening the doors to developer testing on Ethereum’s mainnet, Bjelejic accused him of lying about the progress of zkSync’s third-party security audits. Gulechov, for his part, suggested that Bjelejic was avoiding key questions about how Polygon's zkEVM system will work when it launches in March, purportedly the first such platform to come to market. If anything’s clear from this exchange, it’s that Ethereum’s layer 2 landscape – zkEVMs and other scaling platforms that allow users to transact on Ethereum with greater speeds and lower fees – is still in its early days. While a series of companies are racing to go live, all of the products are likely to have major security caveats when they open up to users. Ethereum’s layer 2 rollups are separate blockchains that sit above the base, layer 1 Ethereum chain. These blockchains host apps just like the base Ethereum chain, but they allow users to transact for cheap by bundling up big groups of transactions and then passing them down to the base chain for all-at-once settlement. Ethereum’s core developers envision a future where rollups become the main point of entry for most users. The big selling point for rollups is that they use technology that allows them to “borrow” Ethereum’s security – meaning transacting on the rollup chains should be functionally equivalent to transacting on Ethereum itself. But that’s not really the case yet. “There are currently a large number of (Optimistic and ZK) rollup projects, at various stages of development,” Ethereum co-founder Vitalik Buterin explained in a developer forum. “One pattern that is common to almost all of them is the use of temporary training wheels.”
Optimistic rollups, the first rollups to come to market, already process a greater volume of transactions than the main Ethereum blockchain. But as CoinDesk has previously reported, the leading Optimistic roll-ups – Optimism and Arbitrum – have safeguards in place that hinge on the honesty of third parties.
One example of how third parties come into play is code upgradeability. Due to the relative immaturity of their technology, rollup protocols can be upgraded to fix bugs. While the ability to fix bugs may seem like an obvious necessity, many keystone crypto protocols, like Uniswap, opt to be non-upgradable. This is because upgrades have the potential to introduce additional bugs (see the $190 million Nomad bridge hack) or be exploited by bad actors who then sneak in outright malicious code.
There’s no evidence that Optimism or Arbitrum have ever suffered in a major way as a result of their upgradeability. However, this isn't the only place where they bake in trust assumptions as a matter of bootstrapping. Until these trust assumptions are pared back significantly, using rollup chains won’t actually be synonymous with using the main Ethereum blockchain.
ZkEVMs are supposed to be a more advanced breed of rollup platform than Optimistic rollups, since they use fancy zero-knowledge cryptography to ensure transaction integrity. But because of their more complex inner workings, they will also come to market with more caveats. “I presume that long-term all of us will have your Aaves, Uniswaps, and all the standard DeFi applications that have been on Ethereum,” Toghrul Maharramov, senior researcher at the zkEVM startup Scroll, told CoinDesk. “So the question is, how do you stand out? How do you build a unique ecosystem?” According to Maharramov, “Unless somebody makes a major technical breakthrough and there’s a massive gap between them and everyone else, it’s more about the values you present.” For Polygon, Matter Labs and other zkEVM builders, security will be a major – if not the major – selling point; hence all of the debate online. But beneath the marketing and social media posturing, says Maharramov, all zkEVMs will face similar security challenges. The “primary concern” in the short term will be bugs in the bridges that people use to pass funds between Ethereum and the zkEVM chains, says Maharramov. Smart contracts that power crypto bridges have long been a major target for hackers, and in the world of zkEVMs, Maharramov says that “bridge contracts are quite complex and avoiding bugs will be difficult.” Perhaps more worrisome than bridges, though, is the difficulty of testing zero-knowledge circuits – the cryptography that will power zkEVMs under the hood. “The technology has evolved drastically throughout the years, so it's not like an established and battle-tested thing,” said Maharramov. Compared to Optimistic rollups, zkEVM code is “much more complex,” and there “are fewer people who can audit that kind of stuff, so it's also going to be more difficult to spot bugs.” As a result of their security risks, zkEVMs are likely to rely on greater trust assumptions (ie. training wheels) as they mature in comparison to Optimistic rollups. Code will be upgradeable, for instance, and trusted third parties will be able to step in to safeguard against circuit bugs. As for when any rollup – Optimistic or zkEVM – will be able to take off these training wheels and launch in earnest, Maharramov says it will still be a while. “I think for Optimistic roll-ups, the path is shorter just because they've been around for a longer time,” he said. “I would expect by next year to have at least protocols that are almost true rollups.” As for zkEVMs, “I would say two years would probably be somewhere around the mark that I would be comfortable with saying that the tech is mature enough,” he said. | 