DAPPCON 2025: Why another Rollup? - James Prestwitch

Transcript

[Applause] Found it. All right. Hi everybody. Uh I'm glad people showed up after lunch. About five minutes ago, this room was looking very sad and depressing. Um so my talk is why another rollup. And I think a lot of you probably have intuited what this is going to be about already. Uh so uh my name is James. I work on something called Signit. Signit is another roll up and I'm going to tell you why another roll up. Um I've been involved in crypto for going on 12 years now and so I've seen a lot of trends in the space. I've seen a lot of things

come and go and rollups are one of the things that has come and stuck around. So we're going to talk about it a little more. Um, so first we're going to need to go 45 minutes in-depth history of rollups, where they came from, how they started, why they're called rollups, and uh I only have a 20 minutes uh left in my talk, so we're actually not going to make it to the real material at all. Uh here's your quick reading list. You want to read roll_up by Barry Whitehat. This is Ethereum Foundation research from 2018 2019. This is the origin of the name rollup the first time it was used on a system that someone actually built.

Read Troubbit. Troubbit's the ones that invented the fault game uh that is used by optimistic roll-ups these days. Uh everyone kind of uh built on their work even though Trit somewhat exploded and went nowhere and then launched a token three years later and did nothing. Uh and read plasma. Plasma is the single most influential thing for modern rollups. If you don't understand plasmas and how they were put together, you won't understand why anyone made any design decision in any modern rollup. And to give you more, here's the spicy reading list. Uh about two three years ago, there were a bunch of debates on Twitter, a bunch of blog posts written back and forth from Kelvin at Optimism,

from John Charb, uh which were basically all definitional arguments about what a rollup is. Uh, and you're going to want to read Mastercoin and Counterparty. These were Bitcoin sovereign roll-ups from 2014 2015, way before we called anything a rollup. Way before we made up the term. But by any reasonable definition of a rollup, these are the first rollups. These are the first things that are embedding extra state attached to a blockchain using that blockchain for DA.

Uh, read fork genen.te. This actually doesn't exist anymore. This was a website where you could uh just fill out a little form and get a Bitcoin fork. Uh so this is the reason there were like 300 Bitcoin clones back in 2015 2016 is because you could actually just go to a website uh pay 0.1 BTC which was not very much money at the time. and that was like a hundred bucks or less. Uh, and you could just have a blockchain, binary, everything, mining pool, whatever, set up for you in about 10 minutes. Uh, and as a result, we had all of these like coins that nobody remembers or thinks about anymore. Uh,

this entire family of Bitcoin clones like Gristalcoin, Ravencoin, Auroracoin. Uh, that one was dropped to the residents of Iceland, by the way. one of the first significant airdrops back in 2014. Um, and if you want to understand roll-ups, you have to read my blog. Uh, my blog is essentially trying to take all those definitional debates and get someone to actually engage with what is a rollup by swearing a lot and trying to be very wrong, very loudly. It unfortunately didn't work, but it does still have what I think is the best definition of a rollup in it. Uh, so you know, like take a quick photo of the reading list. Go like Google all these

things afterwards. I'm actually not going to explain them. Um, okay. So, how are all of these things relevant today? Like, why am I bringing up all of this reading? Why am I trying to get you to learn history? No one wants to learn history, I think. Uh so let's talk about where we are because the reason to understand the past is to understand why we are where we are. If you understand where we've been, where we are, you're going to get a good sense of where we're going.

So uh I saw Bartekch in the back. Uh I actually cited L2B for quite a bit of you know like this section of the talk. Uh we have a bunch of rollups. We got like a bunch of them. Uh, I think that says 58. 58 roll-ups, 23 under review, and 73 things that L2B doesn't think are rollups. Uh, so we've got great great stuff like snacks chain and roar, uh, dairy L3 on arbitum one. Has anyone ever heard of or used any of these things? Any hands?

No. No. Okay, so we've got about 50 rollups that nobody uses or thinks about or cares about, right? This is the state of thing. 33 of these 50 rollups are based on the OP stack or optimisms work somehow. Uh 20 of those secure more than my lunch costs every year. The other 13 are I mean they have TVL measured in the hundreds or thousands of dollars. Like it's we have these chains that do nothing and we have a dozen of them. So rollups are kind of like shovelware at this point. Most of these exist just to

VC fund a company to get a token out the door and to say you're doing something. Uh this is all kind of crap. Like most rollups are junk. um they're just being shoveled out. Hopefully you use it. Maybe you buy their token. Uh but you know, we have 58 of them, right? We're doing a lot of scaling of the L1. We've scaled the L1 58 times. It's great. Well, unfortunately, we have like three and a half. Uh like I said, most of these are built on the OP stack. Uh some of them are built on arbitrum's codebase, some of them are built on zk sync. Some of them are bespoke like the starkware stuff or

the tao based rollup stuff. Um let's take like the three main stacks and run a quick comparison. Here's your main points of roll-up design. How is it sequenced? How do you bridge into the rollup? What's the internal VM? How do you bridge out? And you know, starting from the top, uh, ZKSync uses this run-head central serer. So, some guy signs the sequence and then submits it to Ethereum after it's sequenced. Uh, Arbitum uses a run-head central signer where some guy signs the sequence, submits it to Ethereum after it's sequenced. Obstacer where some guy signs the sequence, submits it to Ethereum after it's sequenced.

So when bridging in uh you make a transaction on Ethereum, the sequencer is forced to include that uh and can include it before they are forced to. Uh in arbitrum you make a transaction on Ethereum, the sequencer must include that, can include it before they're forced to. And on Upstack like you probably get where this is going, right? Like uh the only differentiation between the three major stacks is that zk sync uses a zk proof and the others use fault proofs. Uh optimism arbitrum used to be differentiated on how their fault proof works. There was this thing called the OVM for a while. They never actually got

the fault proof working on it and it was a bad design from the get-go. So, optimism eventually copied arbitum's homework and went to the recursive true bit style fault game. And so, you have these three major rollup stacks that are securing I don't know how much money or have I don't know how much TVL. Uh, and they're all basically from a design perspective exactly the same thing. Uh, with the one exception of zk sync has zk proofs. Uh the design space is just stagnant. Nobody does anything new here.

Uh rollups are stagnant. Like so why another rollup? Because they're stagnant. Uh because they're shovelware. Like you can take the design space, you can do new things with it. We don't have to just be copying like the big stacks. uh they're all the same. We've explored like this tiny little portion of the design space. So, I'm going to go through real quick and talk about just four of the four big design points. What else could you do? Um this is how sequencing works. This is from the optimism spec. If it looks as clear as mud, it's because it is. Uh there's

really like a super complex thing that they do to enable this run ahead sequencing. um you have to like have your transactions in blocks that get put in data batches and then those blocks could span different batches and that's why you have this like a 1 b zero thing and so you can just make it a lot simpler than that with a base roll up like Tao does. You could also look at federated sequencing where it's not just optimism signing blocks where you have multiple people doing it. You could have open sequencing where anyone could sign a block and submit data to Ethereum. Uh there's base sequencing. You can also do things here with fork choice rule

changes. Um we wrote a whole blog post on this. The a rollup can actually like change its sequence retroactively uh based on what the the host chain does, based on what Ethereum does. Nobody has really explored this, but You could have something that is sequenced by an outside sequencer and then changed by Ethereum after the fact. And you can actually re-evaluate the whole rollup based on that. And so if this sounds like a little abstract, it's because no one has built anything concrete here. We've built exactly one sequencer. You know, we've built a

central signer that says what the order of transactions is. We have never experimented in this space for bridging in like they all work exactly the same way. The sequencer must include your transaction from Ethereum. Uh and they do it after a time lag. If you're willing to change how the sequencer works, you could remove that time lag. The time lag is there because the sequencer runs ahead of Ethereum. If the sequencer is not running ahead, you could remove the time lag. Uh so there's already like one big improvement to be made here. if you are willing to change the sequencing model, which no one is because 50 rollups are just kind of crapware that people shoveled out to launch a token. Uh when uh this one like

I don't want to spend a whole lot of time on because I actually agree with using the EVM. Uh you should be EVM compatible because it has the most tooling available. But there are still things that we should be exploring that we aren't. uh NIST compatible curves for transactions are done in pre-ompiles on base and optimism and other places. They should be first class citizens. Uh arbitum is actually experimenting with Wom embedded in the EVM so that you could switch execution context from EVM to Wom and back. It's really interesting and cool, but there's not as much experimentation as there should be.

Everyone mostly just makes an EVM with a couple extra pre-ompiles. And for bridging out, this is really one where I think it's uh we're kind of like paying for plasma sins here. We went into rollups coming out of plasma group coming out of uh like all these plasma designs which had this thing called the mass exit problem where if there's an issue uh everyone wants to exit the plasma at once, everyone has to submit these proofs and go back to Ethereum and etc etc. Uh and so roll-ups were designed from the ground up with proof systems as seen as critical to their functioning because

they came out of plasmas. Proof systems are not critical to a rollup's functioning because a rollup is not a plasma. And so there's this unexplored space of what do what if we just don't have a proof that we call that a sovereign rollup. uh what if we use like something other than you know optimism has control over the bridge you could federate it you could use restaking or something I don't care um signate which I work on uses conditional transactions where you're not proving the state of the rollup you're making a conditional movement on the rollup if certain conditions are met on them on the hosts

this allows you to do things like trade across chains with zero block latency. So rather than using a proof to bridge funds out, you're instead trading with someone who already has funds outside the system. Uh there is a huge design space of re-examining whether a rollup needs proofs at all and why we built proofs the way we have. uh that you know is unexplored because we have exactly two systems that were built already and nobody launching roll-ups wants to innovate.

So the big takeaway of this talk is that roll-ups as a design space is way bigger more complex than you think. There's a huge amount of unexplored rollup choices things we could be building that we aren't. Some of those things are going to be useful. Some of them are going to be useful in ways that are impossible for the current systems to achieve. Current systems cannot achieve zeroblock latency on bridging. They can't achieve zeroblock trades between chains uh without drastically altering the sequencing model, the VM, the bridging models without making completely different fundamental design choices and rebuilding from the ground

up. So, we've talked about like roll-ups are shovelware. Most of the people launching roll-ups don't understand how they work, why they were designed the way they do, or why they're even launching one in the first place. They're just trying to get something out the door so they can justify their funding round and maybe get liquidity on, you know, greater fools. Roll-ups are stagnant. We have three designs which are really one design with a slight variation on proving system. and roll-ups are just kind of sad. If I go look at L2B, like it doesn't make me happy. Uh I don't think any of those are particularly interesting. We're not launching good work here.

Like, and roll-ups could be creative, right? We should be exploring the design space. We should be building things that are new and interesting. And roll-ups can be unique. We don't need 30 optimisms. I don't think there's any use for 30 optimisms. I think there's probably not much of a use for three of them. Uh, and roll-ups can be joyful. Like, we can have fun doing this. We can go build things like uh Fuel V1. We can go build things uh like all of the farming or the party stuff from 2021. We can like do something besides launching a centralized corporate chain.

Uh, and that's my my talk. Uh, I'm James. Uh, I got a lot of opinions. I have about five minutes left for questions. Um, I see in the back over there. Oh, it looks like there's a microphone coming towards you. Thank you, James. Um, I mean, the most obvious question from me, by the way, I'm Bartk. I'm a founder of L2 beat. what L2B can make you can do better to make you happy. I want you to be happy when you go to L2B. But more seriously, uh most of these rollups that you showed that no one seems to be caring about are essentially launched by uh rollup as a

service providers that actually told everyone, hey, with one click of a button, you can launch a rollup. Uh do I read all this as a critique of, you know, their fundamental existence? What about Celestia's promise of you know build whatever you want you know we're going to have 10 thousands of rollups you know each doing this little thing you know what's your take on uh generally RAS providers and the ability to launch a rollup with uh just one click of a button. Yeah. So, I'm I'm old enough to remember 2017, 2018, and 2013, 2014.

And one big thing that happened back then was, you know, ICOs and token sales and uh an entire like industry formed around those. There wasn't just like token sales. You didn't just go launch a token and get $30 million. You paid service providers to do it for you. the service providers would write the contracts, audit the contracts, run the front end, run KYC if you wanted that. There was an entire ecosystem of token sale as a service providers that grew up because there was money to be made. And so for rollup as a service providers, uh that's a natural consequence of money

being made by launching roll-ups. People want to launch a roll-up to make money. someone provides that as a service. Uh I don't think it's a business that I would want to be in personally, but it is a natural consequence of the amount of money that flows towards these things. Someone else. Any other questions? Maybe I can ask a question. Please. I like the idea of uh no bridging out.

Like what do what does it mean? Um so I brought up Counterparty a couple times in the talk and Mastercoin. Um for both of those they were essentially like you would burn Bitcoin to get tokens on the roll up. So Counterparty you could actually like purchase XCP with Bitcoin and your Bitcoin would be burned. Um, side note, Mastercoin you might know better as Omni, which was the home of Tether. So like Tether was homeed on this Bitcoin rollup for a decade uh before they finally went multi like fully multi-chain. Um, I don't know if burning Ether is like a good business strategy or burning your assets on the the host to mint them on the rollup is a

great idea, but it is like a part of the design space that nobody is exploring. Um you can also do things where uh you can get rid of the proof but still have some fallback me fallback mechanism on the host for getting those assets. Right? So for years and years optimism was essentially a two of two multisig for moving assets on Ethereum. Right? The company had one key and the company also had a key that could veto changes. Uh, this was considered production ready and good enough for billions of dollars for a long time. I think you could have a bit more of a secure setup uh than that.

But yeah, the the point being like the proof is a bolted on system after the fact. If you can achieve bridging back to the host, if you can achieve bridging back to Ethereum without the proof, you don't need the proof at all. Users want to bridge back, they don't care about the proof. Um, the way this is done right now is that the companies have multi-IGs which control upgrading the proof system. So those multi-igs are also a root of trust for the system uh and in control of funds on mainnet.

I don't think that's actually uh I think that that's actually a fine long-term setup and we see that in most major assets USDC, WBTC uh all work that way. There is a multisig as root of trust and that's probably okay. Yeah. Um, I have a question as a person with some background in one of these boring rollups. So, I just copied his stack.

Um, and the question is when you think about uh users using your rollup and you especially when you think about developers developing on your rollup, what's your opinion? How far can you stray away from let's say standard tech? Um, how far can you go? So, the users can actually come and you know use the metam masks and unis swaps that they used to and the developers can come and use the things that they also used to to develop. Yeah. Um I think that's a very good question. When you're designing an ecosystem like this to maintaining tooling compatibility is so important.

Uh you see this with like the starkware with Cairo. It's so much harder to build on Cairo than it is to build on optimism or something similar or on arbitum. Uh what I like to see is things that extend the EVM rather than modify the EVM. Rollups also have always have to make certain EVM modifications just because they run in a different environment. Uh so for example, what does the prevando op code do in various different rollup environments? it can't work the way it works on mainet. So what does it do? Uh but maintaining compatibility with wallets with tooling and strictly

extending that functionality is the most important thing. This means that most of your extension should be available over common standards like EIP7 12 not 721. I hate that those are named so similar. 712 the structured data signature standard. If you're not exposing new functionality over existing standards, uh then probably people aren't going to use your new functionality at all. Uh I'm not sure if that answered your question, but it is my ramble for right now. All right, more questions.