speaker stringclasses 4
values | text stringlengths 4 3.06k |
|---|---|
A | Welcome to bankless, where we explore the frontier of Internet money and Internet finance. And today on the show, we are exploring the frontier of the unification of Ethereum. This is a continuation of our uniting Ethereum series that we've been doing with Justin Drake. Today we have, in addition to Justin, Brendan Far... |
B | Yeah. So, David, as you mentioned, this is our third episode in a series, all of which included Justin Drake. The first episode we talked about fixing fragmentation across Ethereum, and we had a zoomed out look at the landscape of Ethereum and what some of the solutions might be. Then in the second episode, we went dee... |
A | Banquet Nation we are exploring the frontier of Ethereum's recomposability, recomposing all of Ethereum's different layer twos into one seamless experience. Today on the show, we are continuing this series that we've been doing with Justin Drake. Justin stepping in as a technical co host today, not just a guest being i... |
C | Thank you. Really excited to be here. |
A | Justin. We've been doing this series, and maybe as you are now elevating yourself to technical co hosts rather than somebody that we are interviewing, maybe you could talk a little bit about. Just prep the listener for the conversation that they're about to hear today. What are some themes? What should they be preparin... |
D | Right, so we have this problem with ethereum, which is the fragmentation of the l two s and the roll ups. Each l two is its own individual silo, and to a very large extent, that's due to the fact that each l two has its own sequencer. Now, we can use ideas like chai sequencing to try and reunite Ethereum, but we need m... |
A | Justin, I'm a pretty visual person, and so I'd like to actually kind of try and ask you about how this ag layer proposal, this proposal for this ag layer fits into the topology of Ethereum. I often view Ethereum as kind of like this sphere, this globe, this planet. And then we have vertical layer twos coming out of it,... |
D | Right. I guess I would think of it as one of the horizontal layers of the stack. So we have settlement data availability, sequencing, and now there's this really thin, actually, aggregation layer, which is adding to this cake. And the reason I mentioned the word cake is because there's this meme going around called cha... |
A | So with this aggregation layer, is this like an intersection point for layer two s to meet before touching the ethereum layer one, the way that I think. |
D | About it, is as a small modification to the settlement layer. So right now, the way that settlement works is that each individual roll up has a different l one contract to do the settlement. But if we do a little tweak to the settlement contract, we can do several things. Like, one is that we can make sure that the set... |
A | Brendan, let's go ahead and get into your brain a little bit. Can you kind of walk us through the maze, the decision tree that you and the Polygon team came up with to produce the ag layer, how did the ag layer become inspired? |
C | Yeah, sure. So I think if we go back to a year ago, there was this question that kept being asked about polygon that was to me a very annoying question. And it was like, doesn't the polygon Pos chain and the polygon ZKVM, aren't they competitive with each other? And for us it was like, no, of course not. We don't envis... |
B | Can I ask a bunch of just dumb questions before we get like, continue this and go deeper? So one maybe dumb question, Brendan, is, is this just for polygon chains versus for all roll ups? |
C | No. So this is not like we've tried to be very intentional in like, not calling it like the polygon ag layer and like branding it as like a polygon zag layer. Because, like, we see this as, first of all, like neutral and public infrastructure for Ethereum. And we can get into, like, Jess and I had a really good convers... |
B | Okay, so this is like, you're trying to make this an Ethereum ag layer. And that requires, of course, we've got Justin here, who's a protocol researcher, very familiar with principles of credible neutrality that requires maximum neutrality here. Justin, I'm wondering if you could comment on this from your perspective. ... |
D | Right. So I think there's definitely an intent to build a public good, one which is well beyond the boundaries of polygon, one which is maximally inclusive. And as Brendan said, there is no opinionation on the specifics of the virtual machine, of the specifics of the governance of the token, of the branding and all of ... |
B | Okay, a couple other things. I understand this is just for ZK evms, or can optimistic roll ups also participate in Ag layer? |
C | So, in our mind, it's really important for the latency of the native bridge to be very short. And so in that sense, the long delay window that is required for a challenge period for fraud proofs is not compatible with this model. And so it would only be for chains that have a ZK prover and can come to finality in a ver... |
D | I guess what I would say is that, in theory, the AG layer could support optimistic roll ups. And qualitatively speaking, it's a very similar system. You have a proof mechanism which takes some amount of time to materialize on chain for settlement. Unfortunately, there is a quantitative difference, which is so large, we... |
B | Okay, another dumb question there. So is something that you opened up, Justin, that gives me some optimism, but also some pessimism, is that other ZK chains are exploring a similar kind of ag layer type of solution. Right. So the optimism there is, oh, we're coming to consensus on a technical solution that will help so... |
A | So now we're fragmented again. |
B | Yeah, now we're fragmented again. So how do we get outside of that coordination failure mode? I guess a few questions I have is, like, what does it take for a ZK chain to sort of participate in Ag layer, and what are kind of like, the incentives towards converging on a single solution like this ag layer, versus kind of... |
C | So I can give an answer for that. I don't know too much about the Zksync proposal, but I can say from the polygon perspective, we sort of anticipated that this might be the case. And so one thing that really drove our, our approach to designing it was to make it as minimal as possible. And so in this design, there's no... |
A | Are other people's ag layers, their ag layers, do they have higher requirement, like higher or less levels of credible neutrality, higher domain specific requirements? Is Ag layer the theoretical minimum level of compromises about it being some sort of proprietary ecosystem? |
C | Yeah, so I can't speak with any authority, unlike what other people are doing, but I do think that what we're building does represent the theoretical minimum for requirements. So you can use any VM safely in this ecosystem, you have complete control over your execution environment and like your governance process, your... |
A | Considered an angular, it's the most minimal version that's still not formally enshrined in Ethereum, though. And so that gap is the difference between something that's purely credibly neutral and what the AG layer currently is, correct? |
C | Yeah, I think so, although I do think that there are ways, and certainly we're thinking about ways to progress along that roadmap. So you could imagine the AG layer being powered by Ethereum, restaking or partially powered by Ethereum restaking or eventually it being enshrined into the protocol. But I do think that fro... |
D | I mean, one of the things that I think is important is that really this should be branded the Ethereum aggregation layer, not the polygon aggregation layer. I think it might also be good to have the repos be in a neutral territory, as it were. And I think it will also be good to incorporate ideas from the metalabs peop... |
B | Justin, so are you optimistic that we can find some social convergence here on what the right sort of ag layer is, at least in the ZK community? It sounds like you are maybe cautiously optimistic, although I haven't heard anyone saying definitively, yeah, Matterlabs and Zksync has signed on to this and they're very exc... |
D | Yeah, I mean, I think one possible way forward is that each ecosystem experiments with these ideas. So you might have shared sequencing, for example, within an ecosystem, you might have some sort of aggregation layer within an ecosystem, and I think people will start converging towards best practices, and then as a nex... |
C | I think that makes sense, and certainly from the polygon side. I think there's a lot of willingness to operate along that roadmap. If you look at what we've done with planky three, which is our next generation proving system, I think it represents the most performant ZK tech right now. And we have been very deliberate ... |
B | Okay, so here's my last dumb question for you, and then we can go deeper. |
C | None of these have been dumb questions. |
B | All right, well, I'm feeling on this subject. Okay, so I'll let David, do you like be the smart podcaster here. Okay, so obviously, Justin, we had you on for our first fixing fragmentation episode, and you brought up this term, universal synchronous composability. I think at the time you weren't fully aware of what egg... |
D | Right. So with shared sequencing and real time proving, you have the ability to do synchronous composability. But there is a catch. And the catch is that you're giving the ability for the shared sequencer to do synchronous composability. But what if they choose to do something different? Now, the solution that I had in... |
B | Okay, so is this better than kind of shared sequencing? Does this work alongside of it? Is it sort of a competing standard? |
D | It's completely complementary. It's like a match made in heaven in some sense. |
B | And it only right now will work for ZK, the ZK EVM world, and we won't be the optimistic rollups, won't be able to use AG layer. At this point in the future, they'll have to use the kind of espresso shared sequencing from episode two. Is that correct? We're not getting universal across AG layer, because we can't integr... |
D | I think all chains can share a shared sequencer if they so choose to, including the optimistic roll ups. And I also think that there is a pathway for the optimistic rollups to also join the aggregation layer. But ultimately, at the end of the day, it's going to be user choices, and maybe choices out of the chains of ho... |
C | So, I think that it would. So, the way that I see it is that chains need to basically navigate the trade offs between safety and liveness. And so, some might only accept sort of synchronous bundles or asynchronous messages if they're able to verify, or like, if a proof has been generated, or if they're able to verify, ... |
B | Okay, here's my dumb summary of things so far. Right. In order to get universal synchronous composability across all of the ethereum chains, we have a few solutions. We have shared sequencers, which can give us this composability across all chains, as long as the chains adopt shared sequencers. And then for ZK evms, we... |
C | Yeah, I think it's right. I think the way that we talk about it is there's a coordination layer. And one component of that, maybe the foundation, which is very, very thin, as Justin said earlier, is the AG layer, and it's providing cryptographic safety and the strong guarantee that chains can rely on shared deposits an... |
A | Okay, so the way I see a lot of these layers, like shared sequencing, closer to the top of this infra layer, ag layer below that, and below the ag layer, ethereum, all of these are doing their own processing, their own efficiency optimization, compression of transactions. Everything is going through this processing sup... |
C | Yeah, I wouldn't necessarily impose a hierarchy of levels, but I think the AG layer is a very thin layer, and then there are shared sequencers, and there's the settlement layer that has a shared deposit contract? I don't know, Justin. Is that how you see it? |
D | Yeah. I mean, one way to think about it is as essentially being the settlement layer with a small change to the small contracts that we already have. I mean, another way to think about it is to start with the sequencing and then adding safety to that. So you can kind of think of it as either being on top or below. But ... |
A | So from my understanding, all ZK chain development kits have some notion of an ag layer. And I think really the two players here are polygon and Zksync. But you could imagine scroll and Tycho in their future iterations. They would also have some sort of fractal scaling structure. Fractal scaling. And ZK has almost been... |
C | Well, so I think this brings up a really good point, and this is maybe a failure in how we talk about the AG layer, but I think a lot of people have sort of read who are building in the ZK community, read what we put out, and they're like, oh, we also do proof aggregation. So we are doing the same thing. And you know, ... |
A | Okay, just to make sure I understand what's going on, because this is getting at one of the points that I want to get to, which is talking about that one central bridge contract. But I think what you are saying is that the AG layers being maybe like application specific, where the applications are chains, it's supposed... |
C | Yeah, exactly. And it's generating very, very specific proofs that allow for chain aggregation. So they give these safety guarantees. |
A | And so this chain specific infra allows the sovereignty of these many, many chains to be preserved while the soundness and security of the aggregation is also preserved. |
C | Exactly. |
D | From a technical standpoint, I think the way it works is that you have this idea of chain level accounting. So you have this central smart contract which keeps track of every single deposit from every single roll up. So, for example, rollup A has deposited 100 ETh. If they have a bug, which allows an attacker to reques... |
C | Exactly. |
A | So there's something I want to get at with this, just at a very non technical, intuitive vibe level, which is if you tell me that Ethereum is going to have this one single deposit contract as the portal to get to the layer two level of ethereum, there's one single deposit contract. All of a sudden, I totally understand... |
D | Fear so we don't have to get to this endgame immediately? I think there's these intermediate steps where there's partial aggregation within ecosystems, and there's a lot of experimentation happening. I think another thing worth mentioning is that this small contract can be extremely simple, orders of magnitude simpler ... |
A | Yeah, all three of you? |
C | Yeah, I think it's a really valid concern to have, but I would follow Justin. I would say we're not proposing a world in which tomorrow we wake up, and we have $35 billion locked in a single deposit contract without formal verification or audits. But certainly from our perspective, we are very proactively taking steps ... |
A | Okay, so what I'm hearing is that, yes, there is a movement towards fragility in the single central contract that the AG layer proposes, and we think we can contain that fragility with formal verification and auditability. Not too long ago, Vitalik put out this tweet about how he's had some sort of change in priority i... |
D | Yeah, I think that's a fair point. I think what Vitalik specifically had in mind is this idea of having a ZKE EVM opcode within the EVM that the roll ups can just leverage so that they don't have to re implement the EVM with their own circuits. And instead you can leverage the client diversity of the execution layer as... |
B | I want to ask how this starts uniting the chains. We live in this world right now where we have a bunch of fragmented chains. And so this is like all of the states in the US being sort of fragmented, their own kind of jurisdictions. So we got Florida over here, California, Colorado, they're all kind of fragmented. We d... |
C | Yeah, so that's a good question. Right now, we're just at the beginning. There are two chains that are currently on this sort of shared bridge, and using the shared infrastructure of what will become the ag layer. Those two chains are the Polygon ZKVM, which is a ZK rollup, and then a star, which is not run by polygon ... |
B | Oh, so they're not. So they're not all like chain development kits? CDK polygon, CDK chains. These are outside of that ecosystem as well. |
C | Yeah, so I would say that we use CDK sort of very loosely. So right now the CDK is a client improver stack. That's basically just a reimplementation of the ZKVM client improver stack. But in the future there's going to be an incredible amount of flexibility in terms of what chains will be able to join the AG layer. I w... |
B | So, Brandon, let me repeat some of that back to you right now. As of now, is Aglaya on Mainnet and it's supporting two chains. Is it live? |
C | Maybe. So when we say the AG layer, there are many different phases, and the first phase is a unified bridge that has a unified deposit contract. And so that's the piece that's currently on Mainnet. The proof aggregation and the guarantee of safety and chain level accounting is still being built out, which is why the c... |
B | How long till we get to that first major milestone where we have polygon proof of stake, working with the Polygon ZK EVM chains all on AG layer, how long do you think that'll take? |
C | So I'm hesitant to give a date because that's a dangerous and risky thing. But I would say sooner than people think. Yeah, so I think six months. |
B | Sooner than that. |
C | Probably sooner than that, but we'll see. Let's be conservative. |
B | Okay. All right, so we'll measure that in months. So we get there in months, and then once that happens, we get kind of like this new set of chains, obviously the CDK chains, but even beyond that, you mentioned near and some others, and then you also mentioned this tantalizing opportunity. Now, chains don't have to boo... |
A | And so you need even less reasons to spawn a chain. |
B | You start to collect some of those new chains as well. And so, like, one way to measure this chain interoperability, I guess, is number of chains that are hooked into ag layer and also like assets, total number of assets that are hooked into ag layer. And I'm imagining once you get to kind of polygon on this, we have i... |
A | Avalanche on the AG layer? |
B | Yeah, like avalanche. I don't know. It has to be ZK, right? |
A | ZK avalanche. |
B | Well, let's start with Zk, like Evm. Let's say the fine folks over at matterlabs or scroll start to join this thing. |
C | Sure. |
B | That would be a next major phase. How long will that take and what has to happen for that? |
C | So I do think that there are like political and narrative challenges around that view, but I think that the really interesting thing about this idea is that you start to see network effects develop around Ethereum rather than around a single competitive ecosystem in the l two world. I think that what you're describing,... |
B | Justin, I want your weigh in on this, because this is as much a social coordination thing as it is a technical. And I think we got from Brendan some of the first major milestones. And to me here is like, we're at milestone one where we have a couple of chains and kind of like this beta version of ag layer. And then a b... |
D | Right. So I think I need to think about it more. But what makes the most sense to me today is to have a two phase approach. Phase one is where every single ecosystem, so the optimisms, the arbitrums, the polygons, the scrolls, the zksync, within their little clan of chains. They experiment and they try things out. And ... |
B | Justin, I think, just like, because we've called this the united chains of Ethereum, just by way of analogy, I think the big question for Ethereum is, can it hold the union together? Basically, is it going to devolve into fragmentation and different islands of state in different countries? Or can we unite all of these ... |
D | Yes, I resonate very much with this, and it gives my work a lot of meaning. In the last few months I've been on this mini crusade around shared sequencing, and I've learned a lot during the process, and I'm very much delighted with how much the ball has rolled forward. A lot of ecosystems, one by one, are starting to a... |
C | Yeah, I think that's a lot of reason to be hopeful. I was just going to say that it also parallels the development of the early Internet. There were a lot of geographically distributed networks that were all competitive and you couldn't easily exchange information between them. I think a lot of the things that we take ... |
B | I hope we get to one Internet of value as well. The way that TCP IP kind of like dominated everything and the way it won, I think, is through credible neutrality. And it became sort of the protocol that everything else was using. My early Internet knowledge is kind of hazy, but I recall people like Bill Gates thinking ... |
A | Why? |
B | Because it's irresistible. They have a fantastic execution environment, fantastic user base and tools and all of these things, and they want to tap into the liquidity, the union of all of the Ethereum chains. If we can get there, then we will have one Internet of chains. And I hope we do get there. |
A | I kind of want to suss out the AG layer. We've been poking at it from the bottom up. I'll say I want to poke at it from the top down. Two chains on top of the AG layer right now, polygon ZK. And then in the future, Polygon. Proof of stake. Let's imagine both of these things have a Uniswap implementation. Now there's tw... |
C | Yeah. So I think this is a really interesting question, and I'm not sure what the end state looks like in terms of the topology of how liquidity is arranged in a multi chain environment. I think that you can imagine it in maybe two forms. The first form, I would say, is a hub and spoke model where you have a single cha... |
A | Okay, just to make sure I understand that there's extra cost if I am on the super fast cheap chain and I'm not on the very liquid more expensive chain, which makes sense, like liquidity and security and expensiveness kind of go together. And so I'm a user on the cheap chain, but I want to make a trade and I want to acc... |
C | No. So there's no extra cost. But we assume that the place where the liquidity is, like you said, is expensive because you have whales and market makers and users that are transacting a lot of size that are competing for priority and for block space on that chain. I think that's one model. The hub and spoke model is on... |
D | Right? I guess the first model is basically the app chain. And I think this makes a lot of sense when the app has to be in one single place. And I think maybe the best example is Ens can't be in every single chain. At least the root of trust needs to be in one place. And when I was in London, someone from ENS was telli... |
B | Justin, I want to make sure I understand that ENS example. So they could either launch, so they want an app chain understood. They want a layer two. They can either launch with their own shared sequencer, which is a pain, or they can use, do you say a base roll up shared sequencer, or use ethereum as the shared sequenc... |
D | Exactly. They can just reuse Ethereum as the shared sequencer. So reusing the layer one proposes, that is basically what it means to become a base roll up, to adopt the base sequencer. And in doing so, there's no degradation of security and liveness. There's no degradation in credible neutrality. And also in some sense... |
B | And then how would ens do this? Do they just adopt espresso or something like this? Because espresso is kind of like tapping into based Ethereum sequencing. Is that a way they could do this? |
D | Right, so it turns out that there isn't like one necessarily base sequencer that could be multiple base sequencers. I guess the most simple and maybe the most credibly neutral base sequencer is just to reuse the vanilla sequencer that Ethereum has. And I have a write up about this, but there might be good reasons to ha... |
B | Sequencer while we're on espresso and shared sequencing as well. Brendan, did I hear that there was a partnership in play between AG layer and espresso? Talk about that. So what does that even look like? What's the nature of the partnership and what value does it create? |
C | Yeah, so I think just at a high level it's like my understanding of shared sequencing and sort of like interchange coordination has really, really benefited from speaking with Ben and obviously speaking with Justin. And so I think this is an effort to continue that collaboration and to really work together, because lik... |
B | Justin Brennan, this has been absolutely fantastic. I'm wondering, in conclusion, Justin, because you've been along for the ride on this whole fixing fragmentation series, and here we are in March of 2024. We've talked about shared sequencing, we've talked about based roll ups, universal synchronous composability for a... |
D | Where will we be in March 2025? I hope that we'll be in a position where all the large ecosystem will be experimenting with their own shared sequencer, as well as with their own aggregation layer. And that there will be the emergence of rough consensus around what is a good design, and that there will be initial discus... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.