dYdX logo
dYdX logodYdX icon
English
中文
日本語
한국어
русский
Türkçe
Français
Português
Español

ZK vs. Optimistic Rollups

Product
Product
debate-recap-zk-vs-optimistic
Product
Product

On June 17, we hosted a debate with Antonio Juliano (dYdX) and Kain Warwick (Synthetix) that was moderated by Dan Robinson (Paradigm) on Twitter Spaces.

Kain and Antonio discuss:

  • Their background and the stories behind Synthetix and dYdX

  • Different philosophies and approaches to building decentralized products and systems

  • Advantage and tradeoffs of ZK Rollups vs. Optimistic Rollups

  • The future of scalability, Layer 2s, and composability

The audio is available here.

A revised transcript is available below:

Dan (Paradigm): Hey everyone, thanks so much for inviting me. And I'm really looking forward to seeing Antonio and Kain face off on this. Just to introduce myself first, I'm a research partner at Paradigm. And I have to disclose my biases in here, but I think they are biases are balanced out, because Paradigm is invested in not only both Synthetix and dYdX, but also in the two layer two solutions we'll be talking about, Optimism and StarkWare.

So I'm very excited to see a spirit of debate about these from people who are much closer than I am. I should disclose Paradigm is an investment firm, but nothing that I say is investment advice. And what I'm saying here is I represent my own opinions and not those of the fund. We're going to be talking about layer 2 scaling and the different approaches that Synthetix and dYdX are taking to it, specifically using Optimistic Rollup versus ZK-Rollup approaches.

I think we're going to talk for about 30 to 45 minutes with the guests and then we're going to open the floor to questions for the remainder of the hour. Yeah, let's get started. Kain and Antonio, they need no introduction. Both of them have been in DeFi for longer than I have, and longer than even I think it's been called DeFi. I guess my first question is, can you share some of your story from how Synthetix and dYdX got started and how you reached this point, maybe starting with Kain?

Kain (Synthetix): Yeah, it's actually funny. I sent Antonio an email. I was trying to track it down, I couldn't find it before we jumped on here. But I sent him an email. I think I saw an announcement or something like that, what they were working on. This had to have been like back in maybe 2017 or 2018, early 2018, and I was like, "Hey, I think what you guys are doing is super interesting. We should chat about it." We've kind of been connected, I think, for a while now. And back then obviously there weren't a ton of projects that were kind of live or doing interesting things on mainnet yet. This is like DeFi pre-history, I suppose. But yeah, so I've always been following dYdX pretty closely. And I think probably the most interesting thing is the kind of approach that they've taken, which has been very pragmatic in maybe a different way. I think Synthetix has tried to be pragmatic about what we've done, but there's certain things that we've sort of optimized for and certain things that they've optimized for, and I think we've ended up in obviously kind of similar places.

But I think the advantage that dYdX has probably had with their kind of more pragmatic approach to scaling is that they've actually got leveraged trading where we're still waiting to kind of get Optimism live to a point where we can launch features. They've had quite a lead on us, even though obviously we've gotten the other derivative products out there, but I think the leveraged trading is a pretty big advantage that they've managed to kind of build over Synthetix over the last year or so.

Dan (Paradigm): Great. And so for those who don't know, Kain, can you tell us a little bit about what Synthetix is and how it works today, and your plans for the near future?

Kain (Synthetix): Synthetix, obviously we have a bunch of tokens that give you price exposure to a range of different assets, so non-Ethereum assets like Bitcoin, Litecoin, et cetera, as well as TradFi stuff like equities and some indices, as well as things like oil, some commodities, gold, silver. We have a range of assets, but all of them are kind of non-leveraged types so they're all just sort of spot exposure. We don't have a futures market yet, but this is something that we're working on.

Unfortunately, we got to a point, I think, with L1 where we realized that running a futures market on L1 would be quite challenging due to scaling constraints. So we're working closely with the Optimism team to get to a point where we can actually launch futures in a kind of higher throughput environment.

Dan (Paradigm): Yeah, and we're definitely going to get to that and talk about the constraints of layer 1. Antonio, how about you, can you tell us about your history, including your history with Kain, it sounds like, and what dYdX is, for those who don't know?

Antonio (dYdX): Yeah, absolutely. Well, yeah. Mad respect to Synthetix as well. I think Kain is far too modest. Big supporters of Synthetix since the early days, and so I really appreciate the approach Synthetix overall has taken to kind of decentralization first and building really high-quality diverse products. But yeah, answering the question as to how I got into DeFi and kind of why a derivatives exchange on DeFi in the first place,.

I've really been working in crypto my whole career since 2015 when I started at Coinbase. So it's all I know, I guess. And ever since Coinbase, it's been pretty clear to me that blockchain in some way was going to be the future, at least something that was going to play a really big role in defining what the future was going to be. And at Coinbase, we had a really great front-row seat in terms of seeing a lot of the really exciting things that were happening in the blockchain space back in 2015.

We had Vitalik come and talk really early on in Ethereum, obviously Fred Ehrsam, Olaf, Brian himself were there. And really the interesting thing that happened while I was at Coinbase back in 2015 was Ethereum came out, did their ICO and launched into production. And I think it took everyone a while to wrap their heads around what the Ethereum was, in terms of like for the first time ever you can write these programs that just run into on the blockchain, run totally deterministically, totally autonomously.

And that was really transformational for me once I kind of wrapped my head around that. And it dawned on me that this was a real new paradigm of computing, and there must be something that's interesting to build whenever there is a brand new way you can build technology and kind of write programs. And I didn't really know what that was going to be back kind of early in 2016 or so, but it was clear to me something would be interesting.

I eventually left Coinbase and started my own thing back in kind of mid-2017. At that time, the DeFi space, and yeah, again, Dan, I think you're right, it wasn't even called DeFi back then, just like decentralized exchanges were very much starting to come out, the very first ones. 0x and Kyber were launched in early 2017, and again, kind of knew the founders there and thought that, that was a really interesting development. Those were really the first actually useful programs, actually useful smart contracts that I could really have seen built on top of Ethereum.

I took a look at that and really thought, "Well, this is really interesting, but what comes next after this?" And really think about finance as a stack where kind of the first thing you need to build is assets and decentralized assets. And we got that in Bitcoin and Ethereum, things people wanted to trade. Then you need decentralized exchanges. At that time, in 2017, that had been built at least in the first iteration with 0x and Kyber.

And then kind of the logical next thing to build in my mind is more advanced financial products, things like margin trading and synthetics and derivatives. It seemed like a pretty logical step in terms of building a decentralized derivatives exchange, and I guess just kind of to motivate the space and motivate the problem a little bit.

Fast forward to 2020 into 2021, one really interesting thing that's happened this past year is the derivatives trading volume on crypto has surpassed all the rest of the trading volume in crypto put together. So it's already the biggest market in crypto, and we think it's going to really continue to grow from here. So huge space already. I think obviously Synthetix and dYdX as well are really on the forefront of pushing forward that really huge market into DeFi, and that's what we're really excited to be working on.

Dan (Paradigm): Cool. And so for those who don't know, can you tell us a little about dYdX, and specifically, I think the perpetual contract and how that works?

Antonio (dYdX): Absolutely. dYdX is a DeFi derivatives exchange, as I was saying. We do have other products as well. We do offer margin trading and spot trading, but really focused on derivatives. And as you mentioned, specifically, this type of derivative contract that's called a perpetual contract. Financially, it's very similar to what's offered on dYdX to what was kind of popularized in the crypto space by BitMEX, and now it's going to proliferate it to a bunch of other centralized exchanges.

These perpetual contracts are traded on tons of different exchanges now, Binance, FTX, BitMEX of course. And like I said, they are the biggest product in crypto today. So a huge market already, perpetuals kind of by share of trading volume in crypto derivatives trades roughly like 10 to 100X more volume than I think all of the rest of the derivatives on crypto put together. So a huge market already, and that's really why dYdX is focused on it.

At a really high level what perpetuals are. They're a leveraged trading product that tracks the price of another asset. So you could have like a Bitcoin perpetual or an Ethereum perpetual, for example. And there's no actual Bitcoin or Ethereum involved when you're trading these assets, so they are synthetic, but you can trade them with really high leverage.

So if you've ever seen kind of the 100X leveraged products of the world, that's what these are, they're perpetuals. Kind of uses this mechanism called a funding rate, which is effectively this dynamic interest rate paid between those who are long and short the contract to tether the price to the price of the underlying, and that's kind of the core construction of the contract. But like I said, huge market already, and that's why dYdX is focused on it.

Dan (Paradigm): Cool. And so, Kain, I think you mentioned you're both derivatives products, but you do take a somewhat different approach. You're not directly competing right now and you don't have a leveraged trading product. I'm curious, how would you describe what differs in your philosophy or your approach to building applications, and I think with a focus maybe on the factors that may have caused you to pick different L2 solutions, which we'll get to?

Kain (Synthetix): Yeah. I think fundamentally the decision that we made, and this comes kind of out of the history of Synthetix a little bit as well, right? Like if I could go back and kind of replay that early history, maybe would have made some different decisions, especially knowing what we know now about how long it would take to actually get scaling sort of live and in production. But coming out of Havven, where we were the stablecoin and the entire state of the system was all in L1, and it wasn't necessarily as critical, I guess, to have high throughput, we basically made these design decisions to kind of keep the entire state of the system on check.

And I think that the approach that dYdX took and some of the other exchanges took, where they basically had like off-chain order matching, that's kind of the pragmatic approach that I was alluding to, I guess, at the start. And I think that it's kind of held them in pretty good stead. The interesting thing, I suppose, as we get into 2020 and 2021 is that kind of forced like this fork in the road in terms of which scaling solutions you could really use.

And so StarkWare, which I think is pretty highly optimized for scaling these off-chain kind of computations and off-chain order matching systems, was obviously very beneficial for dYdX, but wasn't really viable for Synthetix. Obviously, we talked to all the scaling solutions out there for a while before we made a decision to go with Optimism. But the Optimism philosophy, I guess, keeping all of the state on-chain and not using any kind of off-chain computation or order matching was what led us to go down that pathway.

It's probably harder, I would say, it's a more challenging scaling solution to go down, but I think eventually, we will kind of converge as all of these different scaling solutions improve and get better, where we will get to a world where all of the state and all the computation is happening on-chain even for a dYdX. I think I was reading a blog post from like maybe last week that Antonio wrote, where that seems to be the pathway that everyone's going, but obviously, we've just taken different ... That's the destination everyone's targeting, is full-on-chain computation, but we just we've taken different paths to get there.

Dan (Paradigm): That's fascinating. I think some people might actually disagree that the long run necessarily does involve more going on-chain rather than more being done off-chain. But we'll definitely be getting to that when we're talking about the future of L2.

Antonio (dYdX): Yeah, definitely. And just kind of responding to that point, I definitely agree, Kain. And for those who aren't aware of what kind of Kain was alluding to there is dYdX is kind of what's known as a hybrid exchange right now, which means we have some centralized components to the exchange and then some decentralized components. Decentralized components of course are run on smart contracts that are fully decentralized, but we do run a kind of the central order book and central order matching system, which just happens like on servers on AWS.

It doesn't compromise in any way, like the non-custodial trading of the exchange, but you do lose off out on kind of some of the more censorship-resistant properties of decentralization is, at least the way I think about it. And absolutely Kain was right, one of our biggest priorities is going to be for the next like 12 to 18 months to kind of, both from a research and implementation perspective, fully decentralized dYdX.

And, again, Kain was also rights in terms of right now StarkWare is scaling solution, and I'm sure we'll dive into this a lot more really soon, right now is really optimized for some of these projects that have kind of these central matchers or some sort of like central component to the system, because effectively they run, right now, an emphasis on right now, on a single prover system and StarkWare is running these provers.

On StarkWare themselves, one of the big things they're going to be working on over the next year or so is kind of also effectively decentralizing their prover system, which will allow them to open up to more things that are fully decentralized, kind of like Synthetix or other DEXs like that. So really just echoing a lot of the points Kain made there. I do think things like Synthetix and dYdX are kind of going the same way over time.

Of course it's a huge market and enough space for everybody, but for us it's like we ... Not that we 100% solved it, but it's like I think we've made a lot of strides on the performance trading and high leverage features that traders want, but really have to improve on kind of our access of decentralization and censorship resistance, and then I don't wanna put words in your mouth, but potentially like opposite path for Synthetix.

Kain (Synthetix): Yeah, I think that's very true. One other aspect, I guess I would say, and obviously everyone has to make compromises in different dimensions, right? And one of the aspects where I think we've compromised on censorship resistance, certainly in the early days, was having the ability, the contracts weren't fully immutable. So even though everything was on-chain, the contract could be updated at any time. There was a protocol DAO that had control of that, that was run by seven different people within the core team. This is back in the foundation days before we got rid of that and kind of pushed towards more decentralization. But I think for everyone, you have to kind of make pragmatic calls about the things you're willing to compromise on. And for us, even though all of the state was on-chain and the system was like sensibly more censorship resistant, we did compromise in some regards in terms of our ability to do kind of iterative upgrades and what have you.

But I think one of the outputs of that, of having everything on-chain is the fact that you have this kind of property of composability that we're able to tap into. And so Synthetix is used by a bunch of different protocols and things like Curve, obviously where we're integrated with 1inch and other protocols that kind of leverage that on-chain nature to kind of pull different functions together and compose them. Whereas I think once you have that off-chain order matching, you do kind of compromise a little bit on composability.

We've also got some compromises on composability that we have to make in terms of latency and throughput and putting some speed bumps in to ensure that the articles are not exploited. We're not fully composable either, unfortunately, but I think by the time we get on to Optimism, we'll be back to an environment where we're fully composable again, which will be a huge bonus for us in kind of the path that we've taken there.

Dan (Paradigm): Yeah, so that brings us to our next topic, which is the choices that you made for layer 2, the main topic of this panel space. Before I get there, I think I've observed that so far there's been altogether too much agreement between the two of you, for what I thought was supposed to be a very spicy space panel. But I think maybe we'll get there. Here's a big difference. As we've alluded to, you chose different paths, not only different platforms, but altogether different approaches to your first real foray into scaling systems on layer 2 scaling systems on Ethereum. So maybe, Antonio, if you want to talk a little about ... Well, first, if you want to talk a little about just what a rollup is, and then ZK-Rollup specifically, and then Kain could elaborate on Optimistic Rollups and how they differ.

Antonio (dYdX): Yeah, absolutely. Roll-ups, kind of as they're defined, effectively mean building a system on top of another blockchain, and then that kind of secondary system places, commits effectively back to the base level chain, committing to certain state updates over time. But the rollup part of it really means that you're putting the data for all of the transactions, or at least how to kind of prove that what the state of the system that you're committing is, you're putting that data back on-chain on kind of the layer 1 blockchain. In both our cases, our layer 1 blockchain is Ethereum that both StarkWare and Optimism are rolling back up to. So yeah, maybe you have more to add to it, but that's kind of how I would define a roll up there.

Kain (Synthetix): Yeah, agreed. And I guess the differences obviously being using zero-knowledge proofs to kind of validate any fraud or validate those state transitions, versus having these kind of broad prover kind of games that Optimistic Rollups use. But fundamentally, it's a fairly similar approach just with different, I guess, mechanisms for ensuring that the transactions that are rolled up onto L1 Ethereum are legitimate transactions and there's no fraud.

Dan (Paradigm): Kain, what would you say was the core feature of Optimistic Rollup or Optimism that made you pick that over trying to build it with the ZK-Rollup?

Kain (Synthetix): I mean, ZK-Rollups are just not ready to do what we wanted to do as kind of the fundamental thing, right? And this is where I guess I might get a little bit more spicy. We obviously had a view that Optimistic Rollups would be sort of production-ready much faster, and wouldn't require a compromise in terms of off-chain computation in any way. Obviously that hasn't played out as quickly as we'd sort of hoped. We're kind of hoping earlier this year we'd be fully alive and running on Optimism.

But I think these things sometimes take longer than you would like, and there's challenges that you didn't necessarily anticipate, but the end result I think will be better. I think the investment that we've made in Optimism will allow us to have a scaling solution that maintains all of those properties of censorship resistance and the various other things that we want without having to kind of compromise by keeping computation off chain.

Whereas StarkWare, I think, just it's going to be a while longer, and this is in my personal assessment, right, I could easily be wrong on this, but I think it's going to take much more time before zero-knowledge proofs are ready to kind of do all of the things that they're trying to do on-chain in the way that Optimistic Rollups will be able to do this year at some point.

Dan (Paradigm): And how do you respond to that?

Antonio (dYdX): Makes sense. Well, I guess in the name of being spicier and all of these things in respect to Kain and very much respect to Synthetix disclaimer. But yeah, I mean, I think that's the thing that people thought was going to happen is that Optimistic Rollups were like a bit of like a stepping stone on kind of the way to scaling with ZK-Rollups, and zero-knowledge proofs are very complicated so it would take a while for them to get to mainnets, or at least that was kind of the narrative that I saw out there.

But just look at empirically what's happened, right? StarkWare has been live with DeversiFi for like a year and a half now, which is DeversiFi is just a spot exchange. So not quite as technically complicated as potentially building a derivatives exchange, derivatives are fundamentally more complicated from an implementation perspective than trading spot. And I think to my knowledge, dYdX was the first, or one of the first, derivatives exchanges to build on layer 2.

And really the reason why we chose to start, but not the only one, but a really big reason was wanting to get this out into production as soon as possible. And empirically, we looked at StarkWare success with DeversiFi, kind of the quality of their team and what the technology that they were building. And yeah, like I said, building a derivatives exchange is not easy from a technical perspective, especially if you're trying to build it decentralized, and StarkWare is really able to support that already.

Certainly a lot of the drawbacks Kain's pointed out with StarkWare's current implementation, I definitely agree with. Like I said, I don't think today it's like it really makes sense for fully decentralized protocols, like a Synthetix or like a Uniswap or something like that. But I think it easily could over the next year to two years or so. I think there are pretty proven ways, maybe not proven, but at least a well-tested ways that you can scale or can kind of decentralize this prover network with potentially some aspects of like a proof of stake system. I'm not exactly sure that's what StarkWare is doing, but I assume they're doing something like that. It seems like that's a solvable problem, and we've really already solved a lot of the, "Can we build complicated systems on zero-knowledge rollups?" I think dYdX is a good example of like, "Yeah, you can. It's running right now. Go use it."

Dan (Paradigm): What do you think is the biggest disadvantage? And maybe I'll ask Antonio for the biggest disadvantage of Optimistic and Kain for the biggest advantage of ZK, maybe other than it not being ready yet for what you're trying.

Antonio (dYdX): I mean, yeah, first of all that, but sort of like Optimism is a great team and I have no doubt they'll have a really great system up and running in a few months from now, is kind of my understanding, and Kain probably knowms more than me on that. So let's assume that's going to happen. I think there's just kind of a fundamental scaling difference between the two systems as well in terms of ZK-Rollups are able to scale a lot more effectively, at least for our use case, which is obviously the one I'm most familiar with, so I can talk about that a little bit.

One of the fundamental things you have to do with Optimistic Rollups is you have to put every single transaction on-chain, like the data for every single transaction on-chain, and that's the way you're able to really reconstruct the state because people have to be able to look at a given transaction and potentially challenge it. But one really interesting difference with zero-knowledge rollups is even in the rollup mode, which dYdX is using right now, you don't have to put every transaction on-chain, you only have to put kind of the Delta state on-chain. For dYdX, that's just like all of the balances that changed. In the most recent Stark block, you have to put those on-chain.

But the way a trading systems work, or at least the ones dYdX is focused on, there's much more trading activity. A given account may trade like 10 to a hundred times in a given batch, so that's just like an extra 10 to 100X scalability on top of what you already had. I'd say just like zero-knowledge rollups are able to scale in a better way, and that's one big advantage of them for specific use cases like you dYdX.

Dan (Paradigm): Kain, what do you say to that?

Kain (Synthetix): I think that's broadly true in terms of the disadvantage with using ZK-Rollups. Right now, I would say it's fairly clear to everyone that Solidity has a significant lead over a lot of the other languages or approaches to decoding these decentralized systems. And we've got some amazing tooling, and obviously it's improving every month. And so the fact that with Optimistic Rollups there isn't a need, they're EVM compatible, there isn't a need to kind of rewrite these contracts and have people kind of learn a new language or a new style of coding I think is a pretty huge advantage. And yes, obviously, they're not production-ready yet, but I think once they are, we'll see that there'll be kind of a proliferation of people that will be building natively on these L2s almost immediately. There'll be able to take all of the learnings that they've had from building within a year of L1 for the last five years and quickly kind of take advantage of that scaling from Optimistic Rollups.

Antonio (dYdX): Yeah, I think that's a good point. And I do agree that, certainly at least for the current state and the next few years likely, probably Optimistic Rollups are going to be better for composability. Even from a market perspective, what Kain was saying, with kind of just being able to natively and easily port the Solidity app to layer 2, if you can use it mostly as is, is a really big deal, I think. I agree with that.

I think it really comes down to specific use cases and thinking about how important is composability to a given DApp. And I think for a lot of DApps it is fairly important. Certainly, and again, I don't put words in your mouth, but I would assume you think it's pretty important for Synthetix. I just brought this up because I feel like I personally have kind of a counterculture view on this in DeFi, so it may be interesting to dive a little bit more into.

But for dYdX at least, I really think that kind of relative, emphasis on relative, but relative to other DeFi DApps that are out there, scalability is somewhat, or not scalability, composability is somewhat less important for dYdX for two kinds of reasons. The first is kind of organizational, just like we at dYdX really take a full stack approach to building products. We build the smart contracts, we build the front ends, we build the servers, provide support on our products, all of this, and really just think that it's more likely than not going to be us that's going to build the most interesting things on top of dYdX. And we're not 100% on that boat, still really think that composability is somewhat important for us. And other people building other interesting and composable things on dYdX is interesting, but probably not necessarily required for us in dYdX to really win or get really big.

And then I guess kind of the second piece, which kind of I'm curious to get your thoughts on this too, but I feel like it's somewhat less important for derivatives, or maybe at least perpetual contracts, that I'm most familiar with that, but potentially like all or more types of derivative contracts, composability is less important than like a spot trading DEX, like a Uniswap. Because first of all, you don't actually need the tokens to trade on the exchange, and then second of all, kind of a lot of these derivative products are really bespoke. The dYdX Bitcoin perpetual is different than a Perpetual Protocol Bitcoin perpetual, it's different than the Binance Bitcoin perpetual. It makes a little bit less sense to kind of aggregate liquidity across multiple exchanges for a given type of derivative product, but that's just my opinion. Yeah, I'm curious to get your thoughts on that, but I thought I'd throw that out there.

Kain (Synthetix): Yeah. I think kind of philosophically, there's an interesting, if you go way back, right, there's an interesting kind of path dependency here. When Uniswap came out, everyone was, I think, fairly skeptical about this idea of, "Could you have a trading system that didn't have limit order book?" And at the time, you had 0x, you already EtherDelta, you had Kyber, you had a few things that were doing kind of some pooled stuff, some order matching.

But I think fundamentally a lot of the people who were building back in 2017, 2018, were really just trying to replicate centralized exchanges in a decentralized environment. And I think the big thing that Uniswap did was kind of flip that on its head and say, "Actually, let's lean into the fact that we have these different properties in a decentralized environment than we do running stuff on a database. Right? Let's not try and replicate the old world way of doing things, let's kind of reinvent how we do things."

And I think that the advantage that you get there is that you do have this kind of property that emerges out of that, which is composability, right, when you've got a pool system that anyone can kind of tap into. And I think composability is one of those things where it's highly complex. You don't know what you're actually going to get out of it, you may not even know. I think a lot of the things that happen with composability over the last two years, very few people would have been able to even predict that those things were going to happen or how that would play out or whatever. And so I think that it's just one of those things where it's an extremely valuable property to have, even if you don't necessarily know why it's valuable.

And so when we do have composable, perpetual contracts, which, and I know there's a few examples on L1, but, they don't have a ton of liquidity and they've got some limitations, but I think when we've got like a scaling approach that allows for perpetual contracts that are composable, it would be very surprising to me if some things didn't emerge out of composability that we really just can't even predict that people would find ways to leverage that and create new products that maybe don't exist yet.

Antonio (dYdX): Yeah, I think that's a really good point. And I certainly agree that it is super important. I think maybe we just think about this from different perspectives. I guess from my perspective, it's clear to me that trading perpetuals is an incredibly massive market right now, trading perpetuals on limit order books is a massive market. How do we move more of that to the DeFi space is kind of the lens, the really narrow lens that I think about at dYdX.

And all of our users come to us and they're, for our layer 1 products, they're like, "Our trades are reverting," and it's like, "The order book is slower. My trades don't settle until like a minute after they get mined," and all this stuff. So just I guess the question as a business leader is like, "Do you address those problems head-on and try to build solutions for them, or kind of go a different path?"

I don't even 100% know the answer. Empirically, Uniswap is right and they're ridiculously successful. So very clearly that model does work and there are really interesting new things that can be built on DeFi that just fundamentally were impossible before. And yeah, I don't have a conclusion to this thought, but I just think it's kind of an interesting point that you brought up and that I agree with. What is really going to be successful here, is it the things that kind of replicate, or at least like tackle head-on some of the big markets we already see out there, or new projects that just create new markets or things that never really existed before?

Kain (Synthetix): And I think, fundamentally, that approach is maybe much more kind of same approach, if that makes sense, right? If you believe that having things that are decentralized, that don't have the same sort of counterparty risks and all custodial risks and all that stuff, if you think that a scaled out dYdX is better than BitMEX, which I think it would be hard to argue it isn't, right, all things being equal, it's better to not have custodial risk and not have counterparty risk and not have all of these other risks that exist within a centralized system like BitMEX, then yeah, of course we know BitMEX works. We know how it works, we're just taking that and decentralizing it. But I think that there's another level that you can go if you're maybe a bit crazier.

And one thing I will say, as you've been pretty laser focused on like, "I'm going to go out and replicate these centralized systems in a decentralized way. And we know that market exist, and we're just going to go out and get it," and I think that's impressive, and you've never really kind of wavered from that approach, which, again, I think is a very sensible approach. But I do also think that there is a much wider design space in these decentralized systems, and maybe they're less kind of probabilistically successful taking something like Uniswap and trying to replicate it with perpetuals, but I think that it's certainly what trying to do that. But I think both of them are really valuable. And I think that ultimately we're on the same team in terms of, trying to decentralize these systems and move away from centralized custodians and centralized platforms.

Dan (Paradigm): I mean, I think the two of you are really representative of this particular divide, philosophically within DeFi. It was really fascinated to hear you talk about it. I want to talk a little about, for the final part of our space, the future and the long-term future of these protocols. You've talked about what's ready now. Which, for each of you, do you think that the solution you've chosen is also going to be the winner long-term in the asymptote of this technology, and what advantages are going to give it that?

Antonio (dYdX): Yeah, I guess I can start. I mean, I think it's certainly not pre-ordained. There are a lot of unknowns with any of these technologies, they're so new and teams have to execute and things like that. But I definitely think there's a decently good chance that, kind of let's call it zero-knowledge rollups, and hopefully StarkWare leading the way on that, is a pretty good end state technology that a lot of the space can kind of coalesce around. Yes, there are limitations that we've talked about with things like the programming language. Yes, there are challenges that have to be overcome in terms of being able to build out a distributed network of provers and things like that. Like I said, I think that's all really solvable. I think that they are sufficiently good from a scalability perspective, especially after we get things like Eth2 or equivalents for other projects that are building this, that just serve as a huge multiplying factor on what's possible.

Basically, the way I think about layer 2 is just like you take a given layer 1 and then you multiply it by something. And it's kind of in my understanding, in Optimism and StarkWare case, that's like you multiply it by 10 to 100 X roughly, and that's the scalability that you get. But if you think about an Eth2, and I'm not an expert on this, but my understanding is Eth2's data availability is going to be a really big priority and it's going to be like orders of magnitude better than layer 1.

So if you can multiply that by StarkWare's 100X, and let's give them the benefit of the doubt and say that's going to go up to like 1000X multiplier. Over a couple years, I really think that it can run the entire suite of things that need to run in DeFi, if not more blockchain applications after that. Definitely, I think it is kind of an end state technology.

Also just like a slight side note, but on kind of the points of whether Solidity will win out in the longterm or whether some other language, I actually think it's more likely than not to be some other language, whether that's Cairo in what StarkWare is building or something that's more mainstream like a Rust or C++, like some of the other layer 1 chains are using. I just think that good engineers shouldn't really care that much about what programming language they're using, and that in the long run they'll kind of just focus around what languages serve their needs the best. I completely agree with the short to medium term. Like, it's amazing that you can just take your contracts on Ethereum and just deploy them to something like an Optimism, and that will just be a really great way for them to bootstrap adoption. But yeah, I guess just if the question is like what wins longest term, I'm not sure that, that's like the biggest factor.

Kain (Synthetix): Yeah, agreed. I think my view is like ZKP's win out, some zero-knowledge based technology wins out in the longer term, I guess the question is really like, what is the longer term, right? How many times have we been burned in the last six years, right, in terms of scaling and various other things that we think we kind of have a read on how difficult there'll be to implement and then find that there's significantly harder than we were expecting?

For me, I guess it comes down to making sure that we've got something that can scale now, because as I've pointed out in other places at different times, they're now credible threats to kind of the Ethereum hegemony that didn't exist two years ago, like BSC and Solana. And I think that for the Ethereum ecosystem’s sake, we need to be able to kind of compete head to head with those other platforms in the immediate term. But yeah, long-term I've got no question that it's going to be some kind of zero-knowledge tech that will take us to more than 1000X scaling.

Dan (Paradigm): Let's play devil's advocate there, and I think this is a common view and one that I would probably describe to you, but play devil's advocate, there is sort of a fundamental scaling limit to ZKPs in that you have to not only do, perform the execution, but also do the proof in order to ... There has to be prover that can prove this entire thing, right? Do you think it's still ultimately the architecture is going to involve that kind of prover and that's worth it or can be solved?

Antonio (dYdX): Yeah, I think that's a great point. People don't really realize like how inefficient it is right now. It takes StarkWare on the order of like minutes to prove thousands of trades on dYdX, which is really good actually considering where zero-knowledge proofs were at a little while ago. But then now you're right, it's very much an unknown, at least from my understanding, research question about like how much can these things be optimized. And even if you put everything off-chain, even off-chain isn't like infinitely scalable, you still have to pay for that in some way. So if it's like you put it off-chain, but it's still like, I don't know what it is, but maybe like a million times more expensive than if you had just written a program to do the thing, that's still a problem. I totally agree, and that's something that needs to be solved and it's a bit of an unknown.

Kain (Synthetix): Yeah, I agree with that. And I think this is where, kind of to come back to an earlier point, around leaning into ... I think everyone would probably say that it's unlikely that we'll get to a point where a decentralized system can compete with a database, right? There's just fundamental overheads in decentralization that even if you can get close to it. And that's why I think if we're trying to replicate existing systems that have significant advantages in terms of centralization and just being run out of a single sort of instance of a database that's kind of tracking what's going on, it may be hard to compete with. And so leaning into those trade-offs and saying, "We're not going to try and compete directly head to head with an existing centralized system. We're going to have some tweaks to it that are going to take advantage of the decentralization," either Uniswap or whatever, I think in the longterm may still be the end result. But if scaling is able to get close enough, I think that the trade-off of not having to deal with custodians and counterparty risk will be sufficient that a lot of the trading will switch to decentralized systems, even for TradFi things like equities and other instruments.

Dan (Paradigm): One final question before we throw it to the audience. Kain you talked a little bit more about the competitive pressure on Ethereum from other layer 1s, and the impact that's had on the L2 scaling race? Long-term, what do you think the effect of layer 2 scaling on Ethereum will have on Ethereum zone dominance? Do you think it will ultimately be beneficial or parasitic?

Kain (Synthetix): I mean, if you think of L1 as kind of turning into this state storage system, right, rather than something that actually performs computations, then I think it will just end up being a different stack to what maybe we initially envisaged, and you'll have a bunch of different roll up solutions that are all using L1 for state storage and that's just the way that it will go. But ultimately, Ethereum L1 will still be the thing that's kind of holding those altogether. And so I think that, that will still allow for Ethereum, the ecosystem to be dominant, right? It's just you've got like multiple layers in the stack that didn't maybe exist or don't exist now.

Antonio (dYdX): Yeah, I think that's right. And like I said, I really just see L2s at the highest level as a multiplier on what the scalability of an L1 can offer. So if you just think about it given L1, and then you multiply all of them, let's say because you could build like an Optimism on like Solana for the sake of arguments or same thing with StarkWare could settle to any L1 really. But if you multiply all their scalabilities by 100, it just lets you kind of prioritize more the security and decentralization aspects of the chain. So kind of for that reason, I think it would be really beneficial for those chains that are more maximizing for kind of that security and that decentralization. Of course, the king there for smart contracts at least is Ethereum, so I think it will be really beneficial. That being said, if I could just like throw in a little plug here for I'm very much not 100% sold at all on staying on Ethereum forever.

I think a lot of this stuff that's going on with other chains is super interesting, and I think that Ethereum really needs to execute on Eth2 on some reasonable time horizon. It's been like five years away, at least in my mind, for like five years now, and it doesn't seem like that much closer. I really think it comes down to execution. A lot of the research, not all of it, but a lot of it has been done, but these things just need to get built. And if they're not built by Ethereum, they're going to be built by somebody else and people are going to use that other thing.

Dan (Paradigm): Kain, do you think Synthetix will always be on Ethereum?

Kain (Synthetix): I think it will always be on Ethereum, but Antonio's right, if Ethereum doesn't scale and there's something that unseats Ethereum's dominance and is the choice of where people are going, composability is a double-edged sword, right? You need to be where the liquidity is. You need to be where the action is, where everyone's, everyone's playing. And if Ethereum starts to lose that, which is one of the reasons why I've been pushing so hard for a particular rollup solution, right? Because I think that the composable aspects of all these different systems, of Aave and Compound and Uniswap, and what have you, is where the secret sauce of Ethereum's dominance has come from. And if we can get everyone to coordinate on a single rollup solution, particularly an Optimistic Rollup solution that allows that composability to remain, then we can kind of transition more smoothly from the existing low throughput L1 environment that we're in, to a much higher throughput L2 environment and maintain all of that. And I think that, that will put Ethereum in the best possible position to kind of remain dominant. But if we screw that up and we don't get it right, or some other chain starts to really undermine Ethereum dominance, then yeah, I can see that you are forced into a situation where you've got to migrate somewhere else because you can't just be sitting there by yourself if all the liquidity has migrated elsewhere.

Antonio (dydX): Yeah, makes sense. If I can abuse my power here to ask the first question. I'm interested, what do you think is going to happen with composability? Because it really seems like a lot of different projects are moving to a really disparate array of chains. Do you think there'll be a re-centralization towards one winner chain over time, or is this going to last for some significant amount of time?

Kain (Synthetix): I've been saying for a while that I think there's going to be kind of six to 12 to 18 months of kind of chaos, and I think we're seeing that. And I think it's actually worse than I was even hoping, because every day that we don't have some kind of social consensus building up around one solution is just more chaos is kind of emerging, unfortunately. And so I think that, ideally, there should be some convergence on a solution, but if we get to a point where, say, you've got Aave and Compound and Uniswap deployed on multiple different rollup solutions, it's hard to see how that reconverge is. Because we've seen that liquidity is pretty easy to move around between different chains, and there isn't as much of a liquidity moat as we might have initially assumed. And so I think if each of these different rollup solutions is kind of good enough in terms of composability, what you'll probably end up with is something like a 1inch type aggregator that's sitting on top of these different rollups and sucking liquidity out of them as needed and executing transactions across them.

And you'll have different sort of fast bridges, fast exits, which we didn't really talk about that. One of the challenges with Optimistic Rollups is you've got these long challenge periods of like seven days where withdrawals are challenged. I think that they're obviously technical solutions to reduce those challenge periods in the withdrawal periods, but in the short term, that's going to be an impediment as well to composability and to kind of aggregating the liquidity across different rollups. But yeah, my sense is now that there's going to be a bit of chaos for the next year.

Dan (Paradigm): Well, so that's it for our scheduled programming, I think now we throw it open to the floor to see if there's any questions from the audience. If you have a question, raise your hand and we'll pull you up on stage in order. While people are doing that, I got a couple of questions by DM. One of them was, how are you thinking about increasing user adoption through on layer 2, given that there aren't even right now that many users on layer 1 today?

Antonio (dYdX): For us, and this was kind of similar to what I was talking about with the difference in approach and going after a really known market, but for us, just the name of the game over the next year or two is getting more people that are trading derivatives on centralized exchanges to trade them on a decentralized exchange. So just really going hard against that. In terms of ways where we're doing that to scale growth and scale adoption. I mean, a lot of it is just pure like Silicon Valley style growth hacking, and that stuff just works. So continuing to put the pedal to the metal on that, continuing of course to build a great product, one of the things I'm excited about that's hopefully going to be coming out soon on more layer 2s is just ways to move across layer 2s really easily.

Imagine you could deposit to a StarkWare chain from even Optimism or like a Matic or Binance Smart Chain, I think that would be really impactful, and those are the things we're starting to think about at dYdX. And then also imagine if you could just deposit directly to a StarkWare or an Optimism or whatever from centralized exchanges, I think that'd be hugely impactful for adoption, and also I'm sure a ton of other people are working on that as well.

Kain (Synthetix): I guess from Synthetix's perspective, the biggest challenge that we've had is you don't want to jam people into a leaky funnel, right, from a sort of growth perspective, right? And I think that as we've realized the limitations of L1, and we've spent more than a year now working with Optimism to kind of get to a point where we can get everything fully deployed onto Optimistic Ethereum, we just don't believe fundamentally that L1 right now is the place to be kind of pushing growth and trying to drive adoption. But I think when we get to a point where the entire system is deployed on L2, it will finally unlock our ability to kind of really push things forward and drive adoption. I think we've demonstrated, and I think the ecosystem has demonstrated that there are growth hacking techniques that are kind of unique to a system that has tokens like yield farming and various other liquidity incentives that are extremely powerful.

And I think for the last six months, we've had this weird dynamic where the cost of transacting on L1 has been prohibitive for the vast majority of people, and so they've ended up going to places which are maybe less savory and kind of had a bunch of issues emerge out of that. But when we do have a genuinely scaled solution or multiple solutions for Ethereum, I think we'll see a lot of the top tier projects and top tier talent start building there, and we'll have really robust and secure protocols that are kind of dragging people out of L1 onto L2 in a way that's much more sustainable and doesn't create a bunch of issues and trade-offs like we've seen with BSC and some of the other chains.

Dan (Paradigm): Cool. For the final five minutes, I think we've got a couple of people raising their hands. If the host could just pull one of them up and they can ask their question.

Speaker: I'm not here to show my project. But I would just wanted to ask you guys ... I listened to the whole thing, sounds great. I really love the fact that you guys are trying to go fully decentralized, that should be like a main goal in all platforms in my eyes. I wanted to ask you guys about multi-chain. Do you guys have any plans to go on different blockchains? We're building DeFi, we run a version at Ethereum Virtual Machine. It's pretty easy to bring your project over without leaving. See, a lot of misconception is, when I ask people if they want to build on Qtum, they think I'm asking them to leave the the platform they're on, but I mean, you can build on both. Do you guys have solutions for that, and are you interested in that? That was my question.

Antonio (dYdX): Yeah, it's a great question. I think Kain alluded to this a little bit before with like the phenomenon of a Uniswap or a Compound or an Aave deploying onto multiple chains. I think there's two different classes of kind of DeFi or like DEXs, in my opinion, let's just focus on DEXs for now. Ones that kind of store collateral and ones that store state, and ones that are more logic-based.

A good example of the logic-based one is maybe more like a 0x, for example, whereas on dYdX you're really storing collateral on specific chains. And as much as you bifurcate that across multiple chains, it just makes the product worse overall, in my opinion. I'm not the first to say this, but it's you're like fragmenting liquidity across multiple chains for that reason, and it's hard to re-aggregate it back in into one system that's kind of pointing out all the different chains that you're on.

I guess that's a long-winded way to say, in my opinion, for a given DApp, and this is not to speak for everyone in this space, but for dYdX, it really makes sense for us to focus and throw all our efforts behind just one chain, so we can kind of centralize liquidity on that chain and just build the best possible product for everyone. Of course, making it as easy as possible with a lot of the bridges that I was talking about to other L2s and L1s, and even hopefully centralized exchanges for people to get on whatever chain that you're building on. But from my perspective, it makes the most sense just for a given project to focus on one, assuming that it's stateful, like a dYdX.

Kain (Synthetix): Yeah, I think the only thing I would add to that is, unless there's significant activity somewhere else that you don't believe can be pulled onto your kind of primary chain for some weird reason, it just doesn't really make sense to kind of fragment. And I think there's not that many reasons why there would be significant activity somewhere other than Ethereum, other than the fact that it's not scaled, so being cost prohibited, for example. But once you get to a point where transaction costs are significantly lower, again, on L2s, I just don't see a situation where the vast majority of liquidity doesn't migrate back to that L2 or multiple L2s.

Dan (Paradigm): Oh, so I got another question by DM for Kain specifically. I think my old accountant always wanted to ask, "How do you plan to deal with the withdrawal delay?" You mentioned this a little, but if you could elaborate on the withdrawal delay on Optimistic Rollup, and how that works with composability with other applications either on the main channel or on other rollups?

Kain (Synthetix): Yeah, I don't think there is a solution, at least that I'm aware of for composability across L2 onto some other chain. I think that there are ways that you can kind of speed things up and force withdrawal bridges and things like that. There's a number of different projects that are working on these solutions to bridge L2s and L1s. But I think that you will only have the same type of composability that we have on L1 today within that L2. I think that anything outside of that, you're just not going to be able to have atomic transactions. Maybe there's people working on that, but I haven't seen anything that gives me confidence that, that's going to be the case.

Dan (Paradigm): That's our time. Thanks everyone for coming. Kain and Antonio, where's the best place to follow your project's progress or the whatever call to action you want to give the audience?

Antonio (dYdX): Yeah, I think dYdX's Twitter is on this call, so just go on Twitter and follow us. And then we're at dydx.exchange if you want to check us out, and can come into our Discord and ask me anything else you're curious about.

Kain (Synthetix): Yeah, same. Synthetix's Twitter is Synthetix_io, or you can follow my Twitter as well, obviously it's on here. Either one of those is pretty good to keep up with the project, and our discord is a good place as well.

Dan (Paradigm): Thanks so much. I think this has been a really great debate between two very articulate and smart people. And I think we got into some of the contention that I was looking for, so all in all, thanks so much for coming on and for having me join.


About dYdX

dYdX is the developer of a leading decentralized exchange on a mission to build open, secure, and powerful financial products. dYdX runs on audited smart contracts on Ethereum, which eliminates the need to trust a central exchange while trading. We combine the security and transparency of a decentralized exchange, with the speed and usability of a centralized exchange.