# Curing the Rollup Ecosystem's DA Addiction: High-Speed EVM vs Statelessness

- Channel: [ETHTokyo](https://streameth.org/ethtokyo)
- Date: 2025-09-17
- Duration: 27:08
- Watch: https://streameth.org/watch/68cadc5919a669634e7ebaf3

## Description

This session will compare these two distinct directions—high-speed EVM and stateless rollups—highlighting their respective strengths and limitations, while jointly exploring where “true scalability” may lie.

## Transcript

Next session is Curing the Robot Ecosystem via Addiction, High-Speed EVM vs. Statelessness. Please welcome Leona Hioki, Jack Serato, and Patrick McCurry. Hello everybody. Yeah, this session is about DA, stands for Data Availability. It's been a while, data availability, under the discussion, even in the crypto timeline. The data availability was quite technical, Tom, back to 2018 or 19. Actually, maybe we didn't expect that. It became so famous in the crypto timeline, or whatsoever we can jump in. So data availability is one of the most important parts of security, especially in the layered ecosystem. Actually, it's very important for everyone as well. It's well-discussed in the layered context. Today we have two people who are super related to this topic. Patrick McCurry from Arbitrum, and Jack Serato from Plasma 3. Would you mind introducing yourself a bit? Sorry guys. Hi, my name is Patrick. So I guess as an introduction, I work at the Arbitrum Foundation right now, and I've been working there for about two years. My job is mostly focused on governance recently, like DAOs and everything else. I'm sure everyone here has an opinion on the DAO, like the most hated thing in crypto at this point. But then historically, I'm a technical academic, so I did a lot of research in cryptocurrencies for about 10 years, I guess. Probably a bit too long. Hi everyone. My name is Jack Serato. I'm from Buenos Aires, Argentina. I've been in the crypto space for about eight, nine years, and building Plasma 3 to enable payments worldwide. Perfect. So yeah, let's talk about data availability first. Okay, so basically data availability, yeah, what our technicals have done before. What do you think about this importance in the future? So data availability will be a more important topic in the future, or people will ignore that in the future. What do you think about that? I can take a stab at it first. So to me, data availability is like a fundamental property of blockchains in general, and a really good way to illustrate this is the following. So I just want to get everyone's attention because I want to ask you a question. Who here has a cryptocurrency balance, like ETH or Bitcoin or any other token? Raise your hand if you have some tokens. So the question is, how do you know your current balance? Like, how do you know you have those tokens? Well, you need the data for that, don't you? So every new block publishes, you know, the latest state of the system. You can, you know, fetch all the data. You can execute on it. You can compute a copy of the database, and you can look up your balance and say, oh, I do have five tokens, for example. So that's why data availability is like fundamentally important. You need the data just to verify what your balance is. But when it comes to L2s and roll-ups, it goes a bit further. So you can imagine there's a smart contract that sits on Ethereum, and the smart contract is holding billions of dollars. Now, Alice comes along, and Alice wants to, you know, she's deposited money in this smart contract, and she wants to get her money back out. So the smart contract sitting on billions of dollars has to work out, you know, is Alice entitled to withdraw her funds? And, well, the only way the smart contract can work it out is based on the ledger for the L2 system. You know, so like any L2 is a list of account balances. You know, it needs to work out, well, according to the ledger, does Alice have this balance and can she withdraw her funds? In order for the smart contract to be convinced that, you know, Alice can withdraw her funds, well, it needs the data for that. So again, data availability is an important property, because even the smart contract has to verify the data before Alice can withdraw her funds. So we're never going to get away from, you know, the DA, basically. So that's my stuff at the end. Yeah, it's basically, you know, inevitable like that. Basically, their prosperity is how you need a solution to make it kind of less costly.  So what do you think about the DA and their importance in the future? I think the DA importance is vital, because at the end of the day, it's a back-end system that every single L2 is using. When module DA started appearing through different L2s, I think that's when the DA conversation came about on the cost structure of things and the scalability of usage, either in L1s or L2s. And I think the importance of DA got like a narrative within the space, mostly based on cost, mostly based on data storage and how that can be handled. In the growth of the different L2s and the growth of the different blockchains for users, I think that in two, three years, DA will still remain as a fundamental part of the stack you need to have. But the conversation around the DA is going to, in my way, standardized in a way, because the costs are going to be reducing and the usability for the end user at the same time is going to be massive. So it's like every single Web2 component that today you use on the back-end, users at the end of the day, they don't know that they're using it, but they know it's there on the back-end to be able to provide the value for the users. And within Plasma Free, we use Eigen DA exactly to manage the data availability of the users and to be able to maintain the cheapest transactions possible, because having in mind Plasma Free, the payments are into blockchains for mainly stablecoin usage and microtransactions to come into place. Do you think you have a kind of trade-off of the data availability, high cost, high security, or low cost, low security? Can we achieve kind of low cost, high security, for example? Yeah, by having the DA, we have a stateless chain where we can have high security and low fees and be massively adopted by users and we don't have the restriction of cost on the data side, which was something that we've really seen and figured out on the tech part of things. In order to enable massive adoption for the users and usability at the end of the day, we've seen every single L2 grow from cheap to expensive as the usability increases, and that's something that we want to mitigate directly in Plasma Free, that if we get one million users onboarding, we won't see a scale in gas fees on the transactional level, which eventually makes it unusable for most of the people worldwide. Sometimes first world countries see two or three dollar transactions and think that's doable. In the rest of the world, two or three dollars goes a much longer way, so we have to take it to cents in order for that to happen, or micro cents in order to make it possible. That's fair. Patrick, you're from Abiturum, and Abiturum basically aligns with the context of Plasma. It may be happening in 2020, right? Abiturum and optimistic rollouts, how do you feel about the difference, current situation of Abiturum, and what you intended when we see the optimistic rollout emerged back to 2020? Yeah, I guess I'm in contact on this. So, the history of Abiturum is quite funny. So, Abiturum, the white paper was published in 2017 at a conference called USENIX, and I wasn't working at Abiturum at the time. I joined in 2023, so I observed all of this as an external observer. What was quite funny is that the Plasma paper came out, and then the Abiturum paper came out, and they're both trying to solve the same thing, but one was written in an academic way that was really hard to understand, and Plasma was written in a very cryptocurrency way, where it's again very hard to understand. You go through halfway through a paragraph, and then it just drops off. Very, very crypto. But then, since 2017 until at least 2021, when Abiturum launched, and obviously OP launched, and other rollouts attempted to launch, if you compare Abiturum in 2021 to this white paper, it's very different. So, one example is, and I always laugh at this, is the idea of a sequencer. So, today we have this idea of a base rollout, where there's no native sequencer. You can use the L2 stickers for this, and you just rely on the smart contract. That was actually the original Abiturum paper back in 2017. Then they decided that was a bad idea, and they added a sequencer into the system. So, from 2017 to 2021, you just saw the Abiturum, because they're all academics from Princeton, so they weren't really part of the Ethereum community, and yet the Ethereum community were working on ideas, and yet this mismatched for four years. And then roughly at the end of it, we all had roughly the idea of, oh, this is how we should actually build the system and deploy it. So, it was quite fun actually watching that. I mean, I wasn't involved, I watched it as an external observer. I was more doing research than building. So, it's so different from the original version of Abiturum. So, how do you expect Abiturum will change after this? For example, there may be Blob introduced by Ethereum core developers, and basically I believe that Abiturum is using both L1 core data and also Blob as well. So, if you are aligned with the roadmap of Ethereum, what do you think is the biggest change after this, for Abiturum? The biggest change, let me see. So, I think there's two parts to this. There's what can Abiturum detect sector, then there's Ethereum and how can Ethereum support roll-offs. So, on the Ethereum side, maybe one thing to highlight is Blobs, for example. Why do we care about Blobs? I haven't heard of Blobs, by the way. The reason I haven't heard of Ethereum Blobs. Okay, cool. I just want to make sure I know what I'm working with here. So, this is why it's important. So, back in the day, before Blobs were invented, in a transaction, you have something called call data. And call data is really saying, let's give the data to the transaction. The transaction can parse the data, manipulate the data, interact with the data, and use the data in any way that it wants. So, one might be like, let's say, verify digital signature or something. The problem there is call data is really expensive because, well, it can potentially be manipulated in any way by the transaction. So, it has to be expensive up front. Where Blobs, Blobs were introduced in almost 2023, I'm trying to remember now, it was like two or three years ago. Blobs is basically saying, let's get this data onto the Ethereum blockchain, but the transaction can't actually interact with it. You know, it's just a Blob, it gets posted, and that's it. And then the question is, why are Blobs useful? Why should we care about Blobs if you can't even interact with it? And the reason why is because it goes back to that DA problem, that availability problem. What Blobs are really providing is proof of publication that this data was published, and anyone in this room can get access to it. And for Rollups, that's really useful. If I can prove that this data was published to anyone in this room, then I can guarantee that anyone in this room can take the data and then understand what the latest state of the system is for Arbitrum. So, to me, Blobs are really useful because they brought down the gas cost completely, and they really just provided that basic property that we need, you know, proof of publication. Where on the other hand, I guess for the Arbitrum tech stack, this is probably the biggest change since I started. So when I started, Arbitrum had a very restrictive license. It was BSL. You basically had to ask the DAO for a license, and no one was ever going to do that. So we changed the license so that as long as you share a portion of your revenue back to the ecosystem and back to the DAO, you can modify the Arbitrum tech stack as much as you want. And that's really cool because what makes Rollups really exciting is that you can have these crazy virtual experimental blockchains, but they still retain some security from Ethereum. So an example in Arbitrum is that we now have this new VM. Well, we have Wasm. So you can now write a Rust smart contract in Stylus, and then deploy your Rust smart contract, and it can interact with Solidity's smart contracts. And that's really cool. You know, you can do this now, and it's live. So to me, that's like the Arbitrum tech stack, and hopefully Rollups in general should be very customizable. Then people can just, like, add some code and try it out, and hopefully retain the security of Ethereum. Yeah, sounds very interesting. I want to see that. I know it happens in many places, like, you know, OPStack, I know it's well, yeah, commonly used in this space, and the Arbitrum as well. I want to see how it slides. So, Jack, so partner feed basically goes without blocks, and it's not only because using eigenlayers. Basically, it's a unique, optimistic architecture with decay to remove the necessity of the block or data availability in some context. Would you mind, you know, explaining the major difference between Plasma 3 and the other Rollups or the other layer tools? Yeah, absolutely. So Plasma 3 is based on a Plasma technology, which, as Crash mentioned, was one of the ways to scale Ethereum. So the users, we enable basically the user to, at a very low price from their end, to be able to do the transactions and to be able to have the least cost possible on the data side. And that's what enables for us the more part of the scalability. Something that I also want to mention is we have a rollback function where users don't have to depend on the trust of the system, where there's any transactions that weren't done. There's a forced back transaction system where, after a period of time, the chain is rollbackable to a certain state where it pushes to L1 and the funds are 100% secured for the user side. I think that's one of the key differences with the different L2s, where we provide the most amount of security for the end user, for their funds to be there. Because at the end of the day, an L2 is running a different system from L1, has a different, it's completely in a way parallel, connected to L1, but whenever you push to L1 is when you really verify that the funds are there and they stay there. So I think that's a very important change and it's a very different change that we've been doing on the tech side to enable security or prove security to our end users. Yeah, so Plasma 3, it was proposed by Barry Whitehead. By Barry, yeah. And it was quite interesting that in 2023 he published a new plasma and the plasma thought to be ended in the 2090s, and people said plasma is dead. But in 2023, Barry said that it's kind of a plasma bug and this is like a complete version. He didn't say that it's an L2 because it's like rollback instead of rollup, but it's quite a unique idea to facilitate the new way of securing layer 2 with extremely scalable way. So what do you feel? Maybe you are in the plasma community back to the 2080s or something like that, I guess. So how do you feel about this, talking about plasma today? I know it makes me laugh sometimes. No, it's just because 2017 was just so chaotic. There's this really famous diagram for plasma where there's like 200 different versions of plasma, like plasma EVM, plasma something else, plasma XYZ. And we spent like two years designing so many different plasmas and they didn't really go anywhere back then. It was so chaotic. It was so ridiculous on hindsight. But the reason why it didn't go anywhere, it wasn't because the idea was wrong. It was because we didn't understand the design space. And if you don't understand the design space, well, you just come up with weird designs, don't you? But now we understand it quite well. And there's like three parts of the design you care about. One is inavailability, as we just spoke about. State transition integrity. How do you guarantee that a transaction is valid? So you can go from one stage to the next. And the last one is censorship resistance. Like how do you guarantee I can send a transaction and it's eventually going to get executed even if the sequencer goes offline? So now that we understand that design space much better, the question is what is plasma and what is a roll-up and how does it fit into that? And to me, the difference is a roll-up, as the name implies, you roll up the data and you post it somewhere. That's the data availability problem. Where plasma is how do you just keep the data off-chain altogether so you don't have to really post it anywhere. You can just make it available when needed. Or at least in the case of some of the plasma designs, the user holds the data. It's not like it's widely broadcast. As long as the user has their data, then they can always withdraw their funds. So it's really just like playing with the design on how to do DA. And that, to me, is the real difference between plasma and roll-ups, which then hopefully implies that, well, yeah, plasma is a cool idea. And now hopefully we can actually build stuff around plasma that works. Because, yeah, it was chaotic back in 2017. I think that idea of having the entire user having the data instead of a different way of storing the data is what makes it more scalable in a way because centralizing the entire data is 100% super costly compared to the users interacting with their own data. Obviously, you need the user to be interacting with the chain more often. And it's like part of the burden that the user will have. But that's what makes it scalable at the end of the day and the payments throughout the system or the micropayments rule that I think that some of the other ones are lagging on that side, which in my view, it's part of the entire market perspective on the payment side is micropayments are a huge, big thing within that. We all understand macropayments and how those work. But having the user interact with the data, it's a pretty interesting way different to what Arbitron is doing with ROLAs. Yeah, I remember that plasma is said to be dead not only because of EVM compatibility. People said that it's because of the online assumption. People need it to be online. But actually, people are very often online. So, yeah, it might be possible that we can throw away that kind of attack-based or other weakness. Let me add a bit of context just so I can make the online assumption for plasma. So I think the problem back in 2017, it was assumed that the end user had to be online periodically to keep their funds saved because there's a fraud-proof system there. So ROLA sort of solved that because the fraud-proof system was global and not local to the user. But obviously, if you have zero-knowledge proof, then you could go back to the plasma and use a ZK proof and then hopefully the user should have no fraud proof. If the user proves that this is correct, then I hope my data is a user. So fraud proofs are just as tricky when they become local to the user. I think that also something that enabled this to happen was the CK cost and the CK growth. During the last few years, initially CK was kind of an idea and the cost was pretty comparable to certain L2s. And I think that the last two years, CK has been there in order to make this scalable too. Transcribed by https://otter.ai Transcribed by https://otter.ai Transcribed by https://otter.ai Transcribed by https://otter.ai Transcribed by https://otter.ai Transcribed by https://otter.ai Transcribed by https://otter.ai Transcribed by https://otter.ai Transcribed by https://otter.ai Transcribed by https://otter.ai