Ethereum Rollup Evolution with Matt Finestone — Taiko Tuesday Special #1
A series of conversations with the bright minds at Taiko.
Welcome to the first edition of Taiko Tuesday Special! This is a series of conversations I’ll be having with the bright minds at Taiko.
In this first installment, I had the pleasure of chatting with Matt Finestone, co-founder of Taiko and a true OG in the Ethereum rollup space.
Matt played an integral role in the development of Loopring, the first usable ZK-rollup deployed on Ethereum in 2017 that’s still going strong. Matt also brought Loopring to the masses by helping to integrate it within the GameStop ecosystem. To my knowledge, it’s the first such marriage between an Ethereum rollup and a popular brand.
As you can see, Matt has had quite a few “firsts” when it comes to rollups. And he’s about to add another one — the first Ethereum-equivalent validity rollup in Taiko.
That’s not to say that being first is somehow superior or even preferable. But Matt’s rich experience in the rollup world has made him an incredibly interesting person to explore this ever-changing space with.
That’s exactly what we do in this conversation — go on a journey through the Ethereum rollup evolution.
Let’s dive in!
zkMatt, I want to start with a simple question: Why rollups? Why does Ethereum need rollups specifically? What makes rollups an especially attractive way of scaling Ethereum?
Thanks for having me on this illustrious publication, Umede. And thanks for being a great Taiko ambassador!
Ethereum needs rollups because it is unwilling to sacrifice decentralization. That is basically how I think of it. Ethereum’s base layer prioritizes decentralization, and everything else comes after that, including scalability. Decentralization comes in many forms, but here I mean mainly architectural decentralization, as opposed to, say, political decentralization.
So, when it came time to think about scalability, it had to have been done in a way that does not hurt the ability of many people to run full nodes on consumer hardware. Scalability must be measured when holding hardware requirements and the ability to be decentralized constant. As you know, there is always the temptation for blockchain protocols to simply require heftier hardware to achieve higher throughput, but this would limit the ability of willing participants to verify and participate in progressing the chain. So rollups were indeed a game-changing design for Ethereum to solve this fundamental tradeoff. The result is the ability to maintain its most valuable properties at the base layer (decentralization, censorship resistance, security), and provide scalable environments one layer atop, which can inherit Ethereum’s security.
In addition to the achievement of the technical goals of architectural decentralization above, I think it is quite fitting that Ethereum’s scaling layer is driven by a plurality of teams and approaches. A free market of rollups allows for the rapid experimentation and competition we have all seen, and adds to Ethereum’s technical and social resilience.
You were heavily involved with Loopring, an app-specific ZK-rollup that, as far as I know, is also the first ZK-rollup on Ethereum. I believe in one panel you said that Loopring kind of stumbled into the rollup world by accident. How has that rollup world evolved since Loopring — and you personally — got introduced to rollup technology?
I was indeed lucky enough to be part of the Loopring team and witness the first rollup being built and deployed to Ethereum. (Before there were several first rollups😀). It’s true that we did “stumble” into it — as a way to solve our own problems we were facing then, around 2018: high gas fees to execute trades on the Loopring DEX protocol. Thinking back to it now, it strikes me as really nice that we had a problem and were seeking a solution, not the other way around. Of course, many projects and users were facing this gas fee issue back then, but our financial use case particularly felt the pain of trades not being feasible in small amounts/for regular users on our orderbook-based DEX protocol.
So, when we heard about this line of research called “rollups” (it was at Devcon IV in Prague), we looked into it and realized we can implement it to scale Loopring DEX protocol. The research at this point was being done by Barry Whitehat, Harry R, Vitalik, and others, and then the Loopring team built the first implementation or at least the first implementation aimed at being used. This was led mainly by Brecht on our side.
It’s hard to describe just how quickly and meaningfully the rollup space has evolved since then. Anything I could say would not do it justice. It has become its own beast, with many different flavors and styles, design decisions and tradeoffs. Then one simply has to look at all the projects, research, and individuals devoted to each one of those, and you get the idea that this has become one of the hottest “corners” of Ethereum. That makes me feel happy and grateful that I have been able to contribute in a small way to something that has become so crucial for Ethereum.
I sometimes laugh when I think back to my initial rantings and ravings, trying to educate Loopring users or just Ethereum friends about how game-changing rollups are/were going to be. Now there’s an army of such educators, and it’s nearly trite to need to espouse the virtues of rollups — everyone already knows!
Besides this blossoming of the rollup industry itself, the most meaningful evolution since when Loopring was the only game in town is of course the achievement of general-purpose rollups, with the goal of being able to support arbitrary computation and applications, just like Ethereum.
Let’s talk about Taiko now. Taiko is a validity rollup that strives to be Ethereum-equivalent, which is a goal currently no other rollup team has. Can you talk about why the team has decided to target Ethereum-equivalence and not, say, just EVM-equivalence? What are the philosophical reasons?
At Taiko, we find Ethereum equivalence (type-1 ZK-EVMs) to be at an attractive position in the tradeoff space. Perhaps more importantly, as we try to look at the horizon (to the extent we can), we find this position to increase in attractiveness as a few things may come to pass.
We target Ethereum equivalence, firstly, because we love Ethereum and many of us have been obsessed with trying to make it scale securely since the stories mentioned above. That is to say, our goal is not to scale some obscure blockchain or build the next Ethereum (famous last words), it is to scale Ethereum. This means the (extended) EVM blockspace that is secured by Ethereum, but also its community, culture, values, existing projects and people, and all the latent projects and people that could be empowered by it.
Along with those philosophical/mushy reasons, it also means, very concretely, that if we achieve our aims, Taiko would in fact be able to scale Ethereum itself — i.e. prove the execution of Ethereum L1 blocks! The fact that this could be a proving ground and contribute to a SNARK-ed Ethereum L1 is extra exciting.
The above are broader statements, so I will list more specific reasons which are very practical in nature:
EVM “has won” or is in the best position to win the hearts and minds of developers.
Maintaining equivalence with the Ethereum specs beyond EVM equivalence allows us to reuse existing infrastructure in a simple and reliable manner (e.g., Geth). Minimal or zero differences allow us to make use of amazing battle-hardened components for the network.
Maintaining equivalence with the Ethereum specs beyond EVM allows us to be more sure of the security properties of the infrastructure we do reuse, and allows developers/users to be more sure of the security properties and behavior of the contracts/dapps they deploy or use. Even a small group of possible edge case contracts or behaviors performing uncertainly in a modified environment is enough to drive devs or users mad with cognitive overhead.
The “give-up” that Ethereum-equivalent (type-1) ZK-EVMs make is in terms of proof generation time and cost. We believe that:
In regards to time for proof generation, this matters less from a UX perspective than many people believe. It only matters for withdrawals/cross-rollup transactions. See: (i) optimistic rollups with week-long bridge delays with users being happy to use LP-bridges, (ii) clever solutions that could shorten the proof generation time for withdrawal or other transactions (sidecar proofs).
There is a strong tailwind for these proofs to be generated more quickly and cheaply with all the optimizations and progress being made across the industry. We believe that the most zk-unfriendly opcodes or data structures can be tamed and will continue to decrease in cost. Software and hardware optimizations are already yielding this result.
For all these reasons, we are extremely happy and excited to “hug the extreme side of the spectrum” and aim for 100% Ethereum equivalence.
Taiko co-founder Daniel said during the recent Community Call that the team wants to “operate” Taiko like the Ethereum Foundation “operates” Ethereum. Is imitating Ethereum organization-wise also part of being Ethereum-equivalent?
This is a very good question. I do think it is an aspirational goal of ours to even emulate Ethereum organization-wise. As you see above, a lot of the equivalence goal is philosophical, but with very practical underpinnings. I think the same applies here: yes, EF would be a great role model, imho; and also, “operating”/guiding the Taiko project as such stewards would have practical benefits in making Taiko more attractive as a credibly neutral, decentralized rollup.
That said, I think the conditions are a bit different for what is possible. It gets slightly more difficult to approach an “immaculate conception” type of creation and stewardship of a permissionless blockchain as the space grows and as there is more competition. Capital formation differences may also shift things: these days early projects are mostly funded by VCs vs the broader community (who often receive a stake later). I do think a maximally organic, grass-roots, community-led organization is a competitive advantage, and Ethereum is a role model for us in this regard as well.
Finally, it’s crazy to think that you’ve been involved in the rollup space since virtually its very beginning. You’ve seen it all — app-specific rollups, optimistic rollups, validity rollups, and all the other different rollup designs and implementations that have already died out and have yet to be created… How do you see the rollup space going forward?
The one thing I see for sure is that I am capable of wrangling less and less of it into my sphere of understanding 😀. I used to think I “knew” (read: kept abreast) of most of everything happening in the rollup space since it first started, but these days it is simply impossible given the pace, breadth, and depth. There are so many incredible minds contributing, and so one thing I know for sure is that I will continue to be surprised.
That answer is not satisfying, though, so here are some random thoughts, more so observations than predictions:
In the past, we had a need for scalability to afford some popular dapps cheap/fast blockspace. Now we have a glut of cheap EVM blockspace from rollups (with more coming online) and not enough demand from dapps. I hope we see a swarm of new use cases and devs and projects! Let’s solve real problems, and then be so surprised by the magnitude of adoption that we need to keep squeezing out scalability gains. This pattern has been a clear lesson for me of the app-infra cycle.
Similar to the above, the proliferation of “new” designs that are appearing from the modularized way of thinking may be getting ahead of itself. It’s great to have tons of disparate yet connected research come out, but not if it is not meant to seriously solve issues we face today — and especially not if it is being driven by out-of-whack financial incentives to build these solutions with no demand for them.
What’s old is new again. People used to yearn for general-purpose rollups, and now that we have them, app-specific rollups are all the rage. This is interesting to see, and again just another lesson worth noting. I’m unsure if these app-specific rollups will be most compelling as L2s or as L3s, and under what circumstances: the need for max cheapness app-specific L3s, app-specific privacy preserving rollups, etc.
Consolidation within the space. This can mean two things: M&A-style project combinations as many projects seek to increase their competitive position, but also technical consolidation or convergence with rollup projects beginning to use multi-proof systems, best practices from each other after battle-tested tech and strategies, etc.
The progression to decentralization/training-wheel removal may be a bumpy and slow one. It may be worth trying to start decentralized from the onset, without privileged network participants. This is the path Taiko is taking.
Matt, thank you so much for your time. Ethereum is super lucky to have you!
Thanks for reading the first installment of Taiko Tuesday Special!
If you liked this series, please share it on Twitter or Lens. If you don’t want to miss future editions, make sure you hit that Subscribe button and follow me on Twitter.
Got any feedback or would like to collaborate? Send me a message on Twitter!
Note: This newsletter is for educational purposes only and in no way constitutes financial, investment, or any other type of advice. Always do your own research before making financial decisions.