Epicenter - Learn about Crypto, Blockchain, Ethereum, Bitcoin and Distributed Technologies - Sam Bankman-Fried: FTX & Project Serum – The World's First Decentralized Derivatives Exchange

Episode Date: October 27, 2020

Alameda Research is a quant trading fund founded in 2017. Today it manages over $100 million in digital assets and trades $600 million to $1.5 billion per day across thousands of products: all major coins and altcoins, as well as their derivatives. Whilst running this fund, the team thought there was space for a robust crypto derivatives exchange build for traders which solves some of the issues they saw in derivatives trading. FTX is the crypto derivatives trading platform which came out of Alameda. In just under two years of existence, it has grown to become one of the top trading platforms for crypto, trading over $1B per day in derivatives. The FTX team are also working on Project Serum. It’s an ambitious project to create a fully decentralized and permissionless DEX and DeFi ecosystem with trustless cross-chain trading. Serum is being built on Solana to allow a centralized orderbook. This offers a much higher speed and throughput than Ethereum. The goal is to create a robust DEX ecosystem which can compete with centralized exchanges on speed, all while being fully interoperable with Ethereum.Today our guest is Sam Bankman-Fried, he is the CEO of FTX and Alameda Research, and co-founder of Project Serum. We hear all about his journey from Alameda, to FTX, and to his latest venture, Serum.Topics covered in this episode:Sam’s background and how he became involved in crypto tradingWhy and how FTX was createdHow FTX works and why the big focus on derivativesThe cross-over between FTX and AlamedaWhat Serum is and how it fits in the long term vision of FTXSam’s view on where products on DeFi are falling short and how it can be fixedWhat are the trade-offs on Serum?The off chain Serum order bookHow cross-chain swaps workThe boundaries of the Serum ecosystem in relation to SolanaWhat is the governance mechanism on SerumSam’s view on the AMM argumentWhere Sam thinks the DeFi ecosystem is heading and what is needed for it to gain legitimacy in the traditional finance worldEpisode links: FTXAlameda ResearchProject SerumEpicenter episode 312 with Anatoly Yakovenko, Co-founder and CEO of SolanaSushiSwap Twitter threadFTX on TwitterSam on TwitterSponsors: Algorand: Learn how to start building on Algorand – Free webinar on November 17th - https://algorand.com/epicenterThis episode is hosted by Sebastien Couture & Sunny Aggarwal. Show notes and listening options: epicenter.tv/363

This is Epicenter, episode 363 with guest Sam Bankman-Fried. Hi, I'm Sebastinkuujo and you're listening to Epicenter, the podcast where we interview crypto founders, builders, and thought leaders. On this show, we dive deep to learn how things work at a technical level and we fly high to understand visionary concepts and long-term trends. If you like the podcast, the best way to support us is to leave a review on Apple Podcasts. If you're on a Mac or iOS device, the easiest way to do that is to go to epicenter.rocks slash Apple. Today our guest is Sam Bankman-Fried. He's the CEO of FTX and Alameda Research. FtX is a crypto derivatives trading platform, and in just under two years of existence, it's grown to

become one of the top trading platforms for crypto trading over a billion dollars a day in derivatives. FTX came out of Alameda Research, Sam's other company. Alameda is a quant trading fund, and today manages over $100 million in digital assets. While running this fund, the team thought that there was space for a robust crypto derivatives exchange built for traders. which solved some of the problems and issues that they saw in other derivatives trading platforms in the ecosystem. The FTX team is also working on Project Serum. This is an ambitious project to create a fully decentralized and permissionless Dex and

DeFi ecosystem. Of course, with most dexes, there is usually a critical part of the infrastructure, which is centralized, usually it's the order book. And this is because it would be far too costly and slow to run a fully decentralized order book on Ethereum. This is one of the reasons which less than you. the team to build on Solana, which offers much higher transaction speeds and throughput than Ethereum. And the goal is to create a robust Dex ecosystem, which can compete with centralized exchanges on speed,

while being fully interoperable with Ethereum. So by the way, we did an episode with Solana CEO Anatoly Yekovenko last year. It's episode 312, if you're interested in learning more about Solana. One thing to note about this interview, we did not talk about Sushi Swap and Sam's involvement in that matter. First, because there's just so much to discuss regarding FTX and serum, and we wanted to give the most time to these topics, but also because it's already been discussed and debated in my new detail on other podcasts, newsletters and on Twitter, and we didn't feel that we would have much to bring to that story in addition to what's already been said. So my apologies, if that's why you're here, but there's no crypto drama in this one. A little bit of housekeeping. We're doing a meetup this week.

It'll be on Thursday. It's at 8 p.m. Central European time, noon Pacific, 3 p.m. Eastern. I know those are different times from the ones I gave last week, but we just changed times in Europe. And so those are the new times. If you want to register, it's free and it's happening on Zoom. You can register at epicenter. Dot rocks slash meetup. And we'd be happy to see you there. Since today's episode is all about trading platforms and derivatives, I'd like to tell you about the types of things that you can build on Algarand. And in fact, the Algarand team is hosting a free webinar on November 17th to teach developers how they can use the platform to build sophisticated applications for things like crowdfunding,

asset tokenization, supply chain management, and even gaming applications. I'll tell you a little bit more about that later in the interview. But for now, here's our conversation with Sam Bankman-Fried. We're here with Sam Bankman-Fried. Sam, thanks for joining us today. Yeah, thanks for having me. So yeah, we've been wanting to get you on the show for quite a while. Obviously, you are quite well known in the space, and particularly, I think a lot of our listeners will know you from more recent times with everything that's happened around Sushi Swap. But we're going to talk about a whole bunch of things today, including Alameda Research, FDX, and the Serum Project. But before we get started, tell us a bit about your background and how you became involved in crypto and specifically. how you became so heavily involved in crypto trading.

Yeah. So I guess before any of this, I was, you know, I was an ETF trader, Jane Street Capital in New York, trading international ETFs, which are basically U.S. listed funds that hold foreign stocks. And, you know, did that for about three years. It was a really good place. Had a lot of fun there and had a really good time. They were good to me. I think sort of wanted to try starting things on my own,

wanted to try seeing what would happen. And there's sort of just a lot of pretty diverse things I wanted to, you know, think about doing with my life and sort of felt like the only way to really, to really get a sense of them would be to try. And so, you know, I left in 2017. Broadly speaking, I was, you know, trying to think about what I could do that would have the most impact. And, you know, at FTX, I've been thinking, you know, about how could I be able to donate as much as I could?

And that was sort of the primary, you know, the primary means I think about impact. And, you know, there are a lot of ways that you can have an impact on the world. That's certainly one of the clearest ones, which is, you know, you could think about, like, you know, working for some charity. Or you could think about, instead of working for it, you could think about, well, what if I worked somewhere else and donated to it, sort of which would they prefer? and, you know, for a lot of these places, I think, like, sort of not a close call.

Like, they definitely would prefer my money than my time. And so anyway, you know, left Jane Street. And I, you know, tried a few things, including, you know, including working at a charity for a bit, but also looked into crypto. And, you know, sort of like pulled up coin market cap, looked at Bitcoin prices across exchanges. and they're like, oh, those aren't the same number. And in fact, like, very much not the same number. It's like, oh, yeah, a few basis points off. Like, probably one of these data feeds is lagging.

It's like 5% off. And so that means either the data's bad or something's weird or there's an arbitrage to do. But it wasn't totally clear which of those it was going to be, but sort of one of those. And so anyway, I, you know, kind of opened up a Polonex in a Beverex account. So, like, maybe I can do this. And I tried doing it. And the answer is, no, I couldn't. but sort of because of like things that could be fixed.

Like things like, you know, oh, okay, like I could have done this except that I didn't have the withdrawal limits or something like that. But sort of got the sense of like, well, probably there's something here. Like probably I can fight through a lot of these things and end up being able to do some pretty good arbitrages. And that's sort of how it started. And so, anyway, started Alameda research, say crypto quantitative trading firm. you know, started growing out the team and really just started diving in and doing everything we could to scale up sort of each part of the operation to the point where we'd be able to kind of put together enough pieces and connectors that we could do some of these trades. And how long ago was this approximately? This was basically late 2017, so three years ago at this point.

and did you have like any further goals at that time or did you envision like creating your own exchange or was that like sort of the you know what were you just mostly focused on trading at the time and then this stuff came later it's somewhere in between and really what happened was actually pretty early on we did think about like everything from listing structured products to starting in exchange to a lot of other stuff and sort of felt like okay clearly clearly there's a lot of value there if we can do it. But it's hard. And in particular, we sort of kept running against like two different problems. The first was just regulatory stuff and not so much like this is illegal as like, is it clearly a complicated regulatory topic? And boy, do we

not have the know how and the corporate infrastructure to like be able to figure out how to structure these things. And the second was, we don't have any fucking customers. And we have no way to get them. Like, like, just no idea. You know, like, sure, we could build some products, could release them. And just like, hope people start using them. You know, sort of standing there like, all right, guys, you know, new exchange, time to trade. Do you like built a prototype exchange at one point? And it's like, cool, we have a matching engine. Like, what now? You know, where are the customers? Of course, there are no customers. We haven't told anyone about it. Who do we tell about it? I have no idea. And that's sort of like the death of,

of our like early thoughts at, you know, building consumer facing products. And so from this quantitative trading firm, when did you decide that you should build a product or when did you decide to build FDX? Yeah. So, you know, how did we get from sort of there to actually no wait, we have an exchange now? And basically the answer is like, well, a bunch of things happen. over the course of 2018, particularly late 2018, one of which was just like it became more and more clear that existing derivatives exchanges had big problems. And, you know, you can sort of see this in a bunch of ways, but like huge clawbacks and, and just, you know, generally like, like, sort of like really poorly managed risk and margin, huge pain to use. Not very many of them,

but they're huge. And so it sort of like, you know, felt more and more like, okay, like, we should really try and find a way to do this. It was like, what, $350 million lost to clawbacks or something in 2018? And which is basically like derivatives exchange is failing to liquidate people. And so that's one piece of it. But that's solved and solving the problems. There was just like even more reason to want to do this if we could. And so the other thing that happened was I moved to Hong Kong.

It was sort of an accident. I wasn't intending to move to Hong Kong, but I did. I sort of came here for a conference and then just sort of never. left. And all of a sudden, like, I mean, first of all, just, you know, my network just started like expanding pretty quickly as I talked to people here. And so that's sort of one big change. And I sort of feel like, oh, wow, like, actually have some people that could reach out to about this and coordinate with, and, you know, maybe customers or partners. And the other thing was that Alameda building up an OTC desk, which had become over the course of 2018, one of the

larger OTC desks in the industry. And that also meant that sort of there's a bit of a growing customer base, almost all institutional. It wasn't really a retail-facing product. But anyway, you know, the combination of those meant that there's sort of like something to grab onto, you know, like some sort of a customer base and a network. And putting those together, you know, we, and there's maybe one other factor actually sort of worth mentioning, which was the opportunity to talk about maybe white labeling it. And, you know, one thing that we're thinking about was, okay, like, even if we can't find customers, if we can build a good product, you know, maybe there are some existing exchanges that don't have futures but want them. And sort of don't really know how to build that. And maybe we could do that. And so that's sort of the last piece of this. And putting all those together, it felt like, all right, we should go for this. We should try it and see what happens. And so we did. You know, we sort of started building it out late 2018, early 2019. Cool. And so how big is FDX today?

It's, you know, probably the fifth biggest exchange. It's depending on exactly how you count. It's sort of like, you know, volume depends on the environment. Sort of all exchanges sort of like, you know, grow and fall together. But, you know, trade something a billion dollars a day. A volume primarily futures, although also some spot. And, you know, has a small customer base relative to the volume. I think it's probably, you know, whereas its volume is, you know, a factor of eight below the largest exchanges. I think its customer base is probably more like a factor of like 50.

But it's been growing. It definitely started out very institutional heavy. And over the course of 2018, it's becoming, or 2020, sorry, it sort of like has been getting an increasing retail presence. And it's just like that's a network that takes a lot of time and grinding to sort of build out. And, you know, it is a bunch of different tokens on it, a bunch of products. You know, derivatives, spot markets, Fiat, OTC, structured products, and, you know, more coming. Why the big focus on derivatives? Yeah.

So I think there may be a few reasons. The first was, like, it was our differential advantage. Like, we understood derivatives quite well. And, you know, I think that, like, we weren't the only ones who did. but most people didn't. At least most people didn't understand them well enough to really know how to build a derivatives exchange and build it well

and think well about things like risk parameters and liquidations. Second thing was that most of the volume goes up in derivatives. There's more volume in derivatives than in spot. And so it seemed like that was where a lot of the value was there were only two big players in it when we started at OKX and Bitmax. So it's a relatively open space despite having like more than half of all the volume and it was a space that was having a lot of trouble scaling and a lot of trouble supporting

the increasing volume and customer base and flow. And so it just looked like both a sort of high value opportunity and also like a relatively soft space to enter. Algarand's running a free webinar to teach developers how they can use the platform to build sophisticated applications for use cases like crowdfunding, asset tokenization, supply chain management, and even gaming applications. You'll learn how to get started with the command line tools and use the SDK and rest APIs. You'll also learn about the Algarand Foundation's grant program and additional funding

opportunities that the Algarand ecosystem has to offer. So if you're building on a blockchain protocol that has unfeasibly high transaction fees and doesn't provide the speed you need, or if you work for a large enterprise or financial institution and are interested in learning how to build applications that could integrate in your current technology stack, or if you have no blockchain experience at all and just are looking to take that for a step into something new, well, this webinar could be for you. Visit Algarand.com slash epicenter to sign up. Once again, it's free and it's happening on November 17th.

But if you're listening to this after that date, no worries, you can still go to that page and watch the replay. We'd like to thank Algarand for their support of the podcast. And one of like, you know, the things I've heard about FTCS is that, you know, it's a platform that's like very well designed for traders. And, you know, your product design comes a lot from your experiences running an active trading firm. What does that mean like tangibly? Like, what, what are the sort of things that make FTCS different than other exchanges? There's a bunch of things

there. And I'll just give a few examples, although a lot of this is in the details. You know, a lot of this is like a lot of just small tweaks and decisions that we made. You know, what are some of the bigger things? So one thing is, and maybe one of the bigger and sort of like most, you know, sort of most visible things is how margin is handled. You know, I don't know if you've like how much you've used derivatives exchanges before, but sort of the like the norm is that every single margin account is separate. And what I mean by that is like, you know, all right,

so you want to trade, you know, you want to trade Bitcoin derivatives, sure, put Bitcoin in one margin account. You want to trade EOS derivatives. right, put US in another margin account. You want to trade ether, put Ethan a third. You want to trade spot, put that in a fourth account. You want to trade spot margin while there's a fifth account. And many of these places have like 100 accounts or some of them have way more than that.

And it just quickly becomes like, you know, if you basically haven't spent like 500 hours building out like a dedicated automated system to march it to like I'm managing this, it's just like really hard to. handle, you know? Like, you know, there's just sort of like, oh, God, like, I really want to trade, you know, BCH futures right now. What do I need to do? Okay, well, first of all, I need to remove some funds from one of our account there and now. I have to go use those and buy spot BCH. First of all, what the fuck? Like, the whole point of training features not have, you know, that you don't have to have BCH. But now the first step of trading BCH features is buying BCH BCH tokens. And then you move them into the right account. And then you can start trading. And each time you want to scale up the position, you have to like rejigger things.

And then you start thinking about your risk profile. Wait, sorry, really quick. Why do I have to buy BCH in order to buy futures? Yeah, it's a really good question because on a lot of these venues, the margin for BCH futures is BCH. The only thing you're allowed to post collateral in is spot Bitcoin Cash. And so, you know, if you want to trade Bitcoin Cash features, you need physical Bitcoin Cash. And if you want to trade EOS features, you need physical EOS. And so you end up having to manage a spot book and all of these transfers, all within the same exchange, just to be able to trade the futures within it, which like serve really shouldn't be necessary.

And then it gets worse when you start thinking about your risk profile, because you think about liquidations. And these are all independent margin accounts, even though they're within the same exchange and they're all within your account. And it's what that means is that there's separate liquidations for every single one. And so you have to manage, like, make sure that none of these 50 accounts you have within this one exchange are close to getting liquidated. And if they are, you have to like go in and like move funds around. And so this is sort of like the extreme version of isolated margin as opposed to cross-margining. And it's just like a huge pain to manage. And, you know, I mean, it's a huge pain for a sophisticated trading firm with bots that move capital around automatically.

like, you know, it's so much bigger of a pain for, I like manual trader, you know, who like, what they have to like keep constantly checking all these things, like logging on to like, you know, move around their capital and do some trades to. So on FTX, it's just there's one account. You can put anything in it. You can put dollars in and tether Bitcoin, Bitcoin, Bitcoin, cash, eth, whatever you want. It all counts towards margin and it all powers all of the futures on it. And so you never need to reshuffle things. And so it's sort of like, yeah, it gets rid of that whole problem, unless you want to. And if you want to, if you want to isolate margin so that you have like clean trading records split up or so that you have, you know, margin on when it's protected from a liquidation on the other, then, you know, you can open up a different sub account and do that.

So you can customize that how you want, but like by default, you don't have to. So it's sort of like one example of a thing where it's just a pain to use a lot of these platforms and we tried to make it as flexible as possible. And, you know, another piece of that, which is different, is that, you know, FTCS will take a lot of different things as collateral. And so it's not like, you know, you have to store your collateral in dollars or you have to store it in Bitcoin. You can do either. And they'll both power all the future. So that's sort of like one example of a thing that FTCS does differently. They think just makes it a way cleaner and more intuitive platform to use.

FDX is like one of the much larger derivatives exchanges. But its spot volume is like, you know, not. that high relative to like many other exchanges. What do you guys like sort of use for like, you know, your oracles for your like, you know, liquidations and stuff? Is it your own spot markets or do you have like something more sophisticated than that? Yeah, it's a really good question. And basically it's, it's an average of a bunch of markets. And so, you know, our own markets are one of the things in some of these indexes, but, but not the only thing. And so I, you know, Typically, we'll take four or five exchanges and an average those together.

And so, you know, we might have, you know, the FTCS market, finance spot market, Coinbase spot market, and, you know, Whoopi, OKX and a few others, all sort of in there, all average together and, you know, throwing out outliers. And, you know, just to try and make it sort of like, at least like reasonably robust and, you know, not too exposed to like weird prints on one of them. one of the main feat products is you know Big Mac sort of invented this whole like perpetual stuff and you know you guys have

very heavily like sort of you know improved upon the like algorithm what are some of the improvements that you guys had to make there and how does it differ to like sort of the perpetuals offered by other platforms like Bitpacks and Binance and stuff So, and, you know, perpetrals are, you know, the bulk of the volume on, on FTCS, and increasingly everywhere. I mean, even places where it used to be, where it used to be the quarterlies or used to be the spot markets, perpetuals are sort of taking over. So what are some differences that we make?

There are some sort of small things. One thing, our funding periods are different. So we have funding periods once an hour instead of once every eight hours. And it's like not a huge change. the advantage is that is basically that you know you don't have these sort of cliffs three times a day where like funding is paid out and prices move but it's just not a huge difference either way the bigger differences are around margin and so you know one thing which I talked about was sort of the flexible margining that you can put margin in whatever you want and it's all cross-margined between all of the products on other things that it does so they're linear and what what that means is that and this is related to how P&L and margin work, is that if you buy, for instance, you know, seven ETH perpetual futures on FTX and then ETH goes up $20, you've made $140. And then you sort of close them out, and now you have $140 extra dollars in your account. And, you know, it's sort of the same for all

of them. Like what's your P&L? It's like movement in the price of the perpetual your trading times number of tokens worth you bought, which I think is probably like the intuitive way to think about this, a lot of places use what's called inverted futures, sometimes even worse. Sometimes it's quantos futures. And what this means basically is when you see, for instance, the Bitcoin USD order book on Bitmax or the perpetual order book, what they're showing you is not actually really the native version of the order book. And the reason is they don't have dollars.

They don't have anything like dollars on BitMex. Bitcoin's the only collateral. So they can't settle anything. in dollars. And so what they do instead is really it's a USD against Bitcoin order book. So it's a USD future paid for in Bitcoins. And what that means is that like really the price is like 0.000 and the sizes you're trading are like a million if you want to trade 100 Bitcoins. And so they're showing you sort of the inverse of the true underlying matching engines order book. And what that means is that like, you know, it's not like you buy 10 Bitcoin perpetuals and then they go up

of dollar each and you've made $10. It's that you buy like $100 worth of this Bitcoin perpetual order book and then USD goes up a little bit against Bitcoin and then you close out for your USD gain and you get paid in Bitcoin. They're basically equivalent. Like you can start to convert between the two. It's not like a huge fundamental difference. It's just like a little bit more clunky and awkward. If you want to measure your P&L, you sort of like can't do the intuitive thing. It does weird things if Bitcoin crashes a lot because all the collateral is crashing too. And also the P&L you're getting paid is in a thing that's crashing. And so again, it's not a huge difference, but it's just sort of like, you know, it's just a little bit more awkward and clunky and less intuitive. And the reason for this in the end is that like historically these platforms haven't had dollars.

They've only had coins, and so they couldn't settle anything in USD. And so because of that, they didn't try to. They just settled in coins and inverted everything. It also, by the way, does quite weird things if something might go to zero. Because if something might go to zero, then, well, you can sort of think about what might happen. You're settling something, P&L and a currency going to zero. You have to settle infinity of it.

The contract prints up to a price of infinity. It's like it just breaks, sort of like part of the story. behind the Bitcoin Cash fiasco, a fork fiasco, on OK, a couple years ago. And so that's sort of another difference. And again, it's not the biggest difference in the world. It just sort of makes it cleaner and more intuitive. So one of the things we wanted to ask you is with regards to Almeida research and whatever activity is still going on there for maybe like institutional customers, OTC trading,

how do you position yourself there, having sort of privileged information coming from FTCX, going into Alameda? So Alameda itself, it has an account on FTCS. It's one of the market makers there. Its account is not really different from other people's accounts. Like it's just consuming the API feed like everyone else is. And so it doesn't have any more direct information, you know, than that. It's placing its orders using the API feed or using the API.

It doesn't have access to customer account info or anything like that. So, you know, kind of structurally, Alameda is account is just like anyone else is on FTX. And the thing that makes it special is just that unlike everyone else on FTX, Alameda really wants FTCS to do well. I mean, some other people also want to do well, but Alameda really wants it to do well. And so what this means is that, for instance, you know, at the beginning, like more than half of all the, in fact, way more than half of all the liquidity on FTCS was just Alameda. It's just all Alameda providing. And that's changed over time. Now Alameda is, you know, most of the liquidity is, you know,

most of the liquidity is not coming from Alameda, but there's still some places where you can really see it. And I think the clearest example of this is like, you know, we launched some new token and, you know, we launched like Flamingo or whatever, futures on that.

And, you know, we need quarterly futures. You know, we're listening quarterly futures on this thing that like, no one's really thought about it. It's only existed for like an hour or something. And the thing falls apart if there's no liquidity in the order book, like the liquidations don't work.

And so a lot of, off what this comes down to is like you can't list a product until you're going to have liquidity in it. And yeah, it's one thing if this is the new Bitcoin quarterly future that a lot of market makers are going to be looking at. It's another thing. This is some shit coin no one cares about. And like no one is going to want to bother trading this or frying liquidity on it, really. And that's where Alameda is different sort of like Alameda is, it's always happy to, you know, provide liquidity on FTCs, even for things that are not going to like, see much volume because it's important for the platform. And so one of the big things there

is it lets us list products quickly because we just know there's going to be liquidity right there and ready, even if there's not that much incentive for anyone to provide there. What's the relationship between Alameda and FTCX? Is Alameda, like, an investor in FDX? What is like the personnel over cross overlap other than you? they're separate companies. Neither owns the other. In fact, neither has an equity in the other. Alameda does have a bunch of FTT tokens. And so through that, it does have a lot of, you know, exposure to the success of the product. And, you know, outside of my management, there's very little personnel overlap between the companies and Alameda is primarily traders and settlement. And FTX is primarily, you know, operations, business development, customer support, compliance.

and, you know, most of the work on FTX is sort of customer facing, and most of the work on Alameda is, like, you know, strategic and trading related and liquidity related. And so, you know, they're basically separate workforces. And, you know, partially, again, that's because, like, they're fairly different jobs, but partially that's because, you know, it is important that they be separate because you can have a moderately successful exchange, which is just an arm of a trading firm. it's hard to grow to get really big if you don't have that separation, partially because it's hard to get a lot of players on board,

partially because it's just the kind of thing where, like, if you're thinking of the liquidity as always only ever coming from one party, that that's not going to scale as much as it needs to. And so, you know, at the beginning, Alameda was a huge part of FTX. In addition to where a lot of the employees came from,

it was basically all the liquidity. But over the last year and a half, that's sort of been been decreasing quite a bit. So moving on to serum, can you describe at a high level what it is and how it fits in your broader vision for FTX if it does fit in the product vision for FDX? Right. So, you know, I think basically like, you know, we as FTCX, we're thinking as DFI started getting more and more popular, what's our plan here? And part of the plan obviously is list a lot of defy tokens, and, you know, FtX did that. But that sort of like not a very ambitious thing to do in defy. And beyond that, like one thing which I think became clear to us pretty quickly, as we were thinking about this, was like it's not a great fit for defy. Beyond things like listing defy tokens is not decentralized.

Like, you know, FtX it is and it's a centralized exchange. And we could try and have like a, you know, FtX decks, but it's sort of forever going to be tied to FTCS. and it's forever going to be, you know, limited in how decentralized it could get because of that. And so, you know, when we sort of decide not just to do something in D5, but to try and do something big and ambitious, it sort of seemed like the right way to do it was to do something separate. Like the right way to do it, the way that could really scale was to build a new product.

And there are some mirrors with FTX's launch there, where rather than just like scaling up Alameda's OTC trading desk, The really ambitious sort of scaling thing to do would be to start a totally new independent exchange, which obviously came from a lot of the Alameda people, but was its own product not tied to Alameda. And that that was going to be sort of necessary for it to be able to really get big. And we felt a similar thing about CERM, that it needed to be its own thing and independent and decentralized to really have that big of future. And so, you know, as we sort of started diving more into Defi and thinking about what we could do there, and particularly starting to think more, you know, started to think about bigger things

and sort of like really what is the biggest upside here. That's sort of what formed the initial parameters of serum. And one of the biggest things was like, is the goal to launch a defy token or is the goal to build the best defy project we think we can? And if the goal is to launch a defy token, serum was a terrible way to do that or is really overkill and went quite well, but it's massive overkill for that. Like, just like launch a token, right? Like, it's not hard to create a NERC20 token, some project associated with it, whatever. You know, the food industry has done quite a good job of this. That, those are of one path,

but that was obviously not the ambitious path and not the path that would create a great long-term product. That was sort of the path that would create a product and a token. But if we want to do the long-term thing, you know, we started asking questions, like, not like what would be a way to improve on DFI, but like, what if we want to have a DFI ecosystem that could support a billion users? What does that look like? If the answer is that's impossible, then you're sort of like already conceding from the outset, that it's never going to be huge. And we didn't want to do that. We really want to think about the upside here. And so if you're doing that, like, you know, what would it mean to try and, you know, start a DFI ecosystem that would have a chance

to scale to a billion users. Well, first of all, it meant ignore what products are out there now, try and build the best products. It also meant wherever you're doing this has to be able to scale to a billion people, or at least have a shot at it. And it meant, you know, with all these decisions and, you know, air towards a thing that has the biggest potential. And so, you know, it meant certainly can't be on Ethereum.

And in fact, it's not just like can't be on Ethereum. It can't be on most chains. Like it would have to be on, you know, throughput is just going to be a limiting factor. eventually. And so the more, the better. And, and yeah, it meant like what are sort of the fundamental financial components? Order books are sort of probably the biggest. So that's sort of where it started, was launching it fully on-chain or a book, I think probably the fastest, most performant on-chain order book anyone's created. And, and then moving on to, you know, everything from structured products, pools, AMMs, borrow lending, margin, derivatives, and sort of building out the

core pieces of the financial ecosystem. And then also thinking beyond the financial ecosystem about like what else could be on chain. Well, a lot of things could, you know, instant messaging, sure. You know, someone's built a prototype instant messaging, encrypted instant messaging protocol where it's all on chain. You know, P2P stuff can be on chain. A lot of what credit cards do could, you know, just like a lot of things in the world could be. So basically, you know, that was sort of where they started and and what generated a lot

of the sort of like initial ideas behind it. So where, in your view, does the, do the existing tools and the existing products in DFI fall short, you know, of fulfilling this vision of building, scaling to a billion users? Yeah. Well, first of all, just like chain throughput. I mean, Ethereum is, is overloaded right now, and that's at, like, you know, what, like 10,000 users or something. And so, you know, you need, like, many, many orders of magnitude. And even with the scaling solutions for it are not meant to scale that big. You know, most of the scaling solutions are meant to get another couple orders of magnitude. Second of all, if you look at the products being built, they're really tuned to that

environment, right? You have like very simplistic AMMs because doing anything more complicated on Ethereum would be too hard. The borrow lending pro calls, which are fairly dumbed down, but honestly work decently well. I think they're like one of the best pieces of Defi right now. But then you ask like, where's the rest of the financial ecosystem? There isn't a lot. Like, there really aren't, and their derivatives are like really nasty compared to or off-chain. Like there's sort of two types of defy derivatives. There's off-chain ones. Or there's ones which are massively over-collateralized, extremely reliant on not just

oracles, but on like weird collateral bases, weirdly pure to pure without market makers. Like it's sort of like is designed in a pretty constrained way as opposed to how finance usually views things. And so I think for most of what you're looking at, beyond a partial exception of our lending, you know, most of the real financial ecosystem is not built yet on Defi. It's sort of like these toy versions of it built to suit an environment where your protocol is going to get like, you know, five actions a minute and that's it. Would you say that like, you know, if we could get a lot of the existing defy applications to move on to a faster block. chain like Solana or something. I mean, do you think that Solana can achieve those like many, many magnitudes of like scalability over Ethereum? I think it can. And so first of all, it already has many.

I mean, it already, so Ethereum's like 10 transactions per second. And Salon is somewhere around 50,000 right now. So it already has, you know, three orders of magnitude above Ethereum. But the other thing is that, so first of all, it scales with Moore's law. So every few years, it's, its capacity is doubling. And then there's another few orders of magnitude it can get just by optimizing the current protocol without sort of, without massive changes to the, to the blockchain.

There's just a lot of things that aren't optimized right now. And the teams are aware of that and is working on these. And it's one of the priorities. And that can, you know, get, you know, again,

another couple orders of magnitude. And so, you know, I think already you could probably get close to a million transactions a second with, without better hardware. And then, you know, yeah, again, in another decade, you know, you add another factor of 10 to that or something.

And so is that going to be everything that you could possibly want? No. Like more is always better. And at some point, you might have to think about charting it. You might have to think about running multiple copies in parallel or something like that. But at the very least, it gets a very large number of, you know, more than what we have right now. And enough to scale up to, you know, way more than, for instance, what a current leading cryptocurrency exchange supports. And so at the very least, it's sort of getting to the point where it can host a bunch of real, sort of like global scale applications on it. But like, it comes at a tradeoff, right? So Salana today, it's, you know, this ability to like scale with Moore's Law, it basically uses very high tech hardware.

It's not like you're a normal consumer grade hardware. I've ran validators for Salana before. And, you know, you need to have like at least two top end GPUs running just to even run a node, which is not like, you know, most people don't have that laying around, and they don't have light client proofs and stuff. And so, you know, they choose a certain point on the tradeoff curve. And, you know, I completely accept that my entire view of blockchain is that we should explore different tradeoff scheme. But why do you think that that's the right tradeoff point for zero? And, you know, there is a tradeoff here and sort of like, I think a lot of what it's thinking about is like, you know, well, so the first thing I'll say

some of this is not a trade-off. Some of it, there are some of it which are not, I wouldn't say, free wins, but just like a lot of optimizations you can make to be able to increase throughput without sacrificing much. Beyond that, though, you know, I think a lot of this is thinking about, realistically speaking, what are the benefits and costs to various levels of this? And sort of like, instead of thinking, like, that the whole blockchain has to fit on, like, 2018 cell phone or something like that.

and that, you know, that has to be able to serve as a node. It's instead thinking, like, for anyone who's seriously interested in running a node, this should be possible. I think that is a change, and there is a cost associated with that. But I don't think it's a super massive one, because realistically speaking, it's not sort of like people's cell phones running full clients, which is powering basically any blockchain. And probably that's going to remain true. that it's, you know, people who are, are seriously interested in it. But at that level, you know, it's not super expensive. It's, you know, for, you know, certainly less than $1,000.

You can get a setup that can, you know, be a top end salon of validator. Well, what about like clients, though? Like, there's no like client support. I would be okay with that if there was some like middle ground where on my phone I could run a light client. The problem is for me that doesn't exist yet. Yep. And, you know, some versions of that are things that you can build, although definitely the more you scale of the blockchain, the more it's going to have to be losing.

And, you know, you might start running into restrictions of how far back it can look. And, you know, I think you might be in a place where, like, you could verify that the transactions are valid given a blockchain history as of yesterday or something like, you know, starting from the state of the blockchain yesterday, taking that as a given or something like that, as opposed to being able to replay the entire thing. you know, and verify it. And so again, I think that there is a cost to that, but that if you look at where, where people are right now and where this is, you know, probably going to end up, you know, I think that, realistically speaking, you can still have a ton of validators, you can still have anyone who's really interested in it being one, you can still have some amount of verification done by users. And that, while that does sacrifice something, I think you still keep most of the security and decentralization of blockchains. I think that, like,

you know, most of it comes when you go from like some sort of like small tight net whitelisted group of people verifying this to, you know, a third of the world could be doing it if they wanted to. And so, yeah, basically I think there are sacrifices, but they're not that big in terms of the ultimate security and decentralization. And what it lets you do is scale up by many, many orders of magnitude. It lets you go from like a thing which can support a few simplistic applications to something that can drive a lot of what the world's doing and that that's a pretty big difference. So let's get into the technical weeds here a little bit and explore the different components of serum. So if you go to the website says that there's seven components to serum.

So that's a good, I think like a good starting point. Let's let's break those down and we can maybe talk about a little bit about each of them. Yeah. And, you know, one thing to know, obviously is that sort of like, you know, that's from the white paper and things have grown a fair bit since then. And so there's a lot of other components which are being built out as well. And, you know, you'll notice that that things like AMMs, yield pools, things like that are not listed there, but are looking like they're probably going to be a pretty central

part of serum. But, you know, if you do sort of look at sort of, you know, the original pieces and one good place to start is the order book, that was sort of like. one of the core visions for Ceremony is can you have a fully on chain order book. And the answer is yes. You can have one. It can be performant. You can be fast.

It can be cheap. It's really finicky to build, but you can do it. And we did do it. And it's live. And you sort of like go to it right now and trade on it has, you know, one second settlement, hundredth of penny cost, gas cost. And, you know, fully on chain matching engine order book and everything else. So and that sort of like really one of the core pieces of it.

and also sort of proof that you can build real scale things on it. So it's sort of one of the first things that we focused on and one of the first really big thing that went live. And so the order book is like the first thing that went live, like you said. What's unique about this particular order book and what kind of things the Solana allow for that you don't see in other order book dexes? Yeah, so the big difference is that it's all on-chain.

And if you look at other orderbook dexes, usually one of the key components is not on-chain. And they vary in how they do this, but the matching engine often is not on-chain. Often you'll submit orders to just a centralized server, which will match. And then the funds will be on-chain, and there'll be settlement on-chain, but the orders and matching engine won't be. Whereas with serum, they are. And so this means a few things. First of all, it means that you can see everything that's happening and prove that everything is happening as it should. It also means that anyone can do this.

Like anyone can run an order book on serum. It's not permissioned. It's not centralized. It's not like, oh, yeah, the server is only supporting these. You know, you can, and in fact, there's been code released in that open source code, just like click a button and add new order book there. So when you say anyone can create an order book, like anyone can create a pair. Yep. Anyone could say, you know, exactly, cure two SPL tokens, click this button and I'll run the code to instantiate a new order book on serum.

Interesting. Right. I wanted to ask about this. Like on the docs website, you know, it lists like a, it's like, oh, here's a bunch of dexes on serum. What does that mean? Yeah, what does that mean? Yeah, it's a good question. And this really starts to dig into what is in Defi, what is a thing? And it's a tough question, right?

Like if it's all on chain code, what is a Dex exactly? I think someone had a post for that that was like a kind of cool post, which is like, you know, it's not clear. Like you could, you know, two parties could exchange things on chain. What does it mean for there to be a Dex if there's no centralized thing at all? So what that means is people who are hosting GUIs that allow you to trade on. order books on serum. And so if you click through each of those, it's different parties which are hosting,

hosting their own GUI, which can access either sort of like markets that other people have created an effort looking at, or, I mean, anyone can list their own markets as you'll see some dexes will, you know,

create their own markets on serum and then have their decks point to those. And so it's sort of this interesting in-between thing where it's like, you know, if you're saying, I want to host an FTX GUI. Like, what's that even mean?

It's sort of like, is it just like a front end? Whereas if you're to like build your own exchange, that's your own thing, what's it mean to be doing one of these things? It's a little bit in between, right? Like you list whatever markets you want. You can have this pointed at your markets

or you can have it pointed at the same market as other dexes are running on. So all the order to cross with each other. And, you know, there is like some example code uploaded, but you could, you know, write your own code. And some people have customized it. So I can have multiple markets. for the same pair, though?

Yep, if you wanted. And obviously, you know, I'd expect people to congregate because it sort of likes a little ridiculous to have, you know, everyone pointing at a different Bitcoin USDT order book and like no one agreeing on it. And so like all the liquidity swifts. But yeah, I'm sure they're like 20 Bitcoin USDT order books at this point from various people who have tried out creating them. Do I have any incentive to make my own order book?

Like, as the GUI, can I like take my own fee if it's on like some, order book that I controlled? Yep, it's a good question. So it's basically it doesn't have so much do with which matching engine or order books you're pointing at. And instead what it is, is, well, by default, without doing anything, the GUIs do get a fraction of all the fees on Ceremony. So I think 20% of all taker fees, which is more than 100% of all fees because they're maker rebates, go straight to whoever is hosting the GUI. And what that actually means if you drill into it whenever you send an order, there's just a field. You can append to it, which is like which SPL address should get 20% of the fees.

And do you write a GUI, probably you're going to put your address in there for, you know, what the orders that the GUI generates. And so, you know, the people hosting GUIs are all getting, getting fees from this, basically, from the activity that flows through their GUI. So that's the basic thing. Obviously, you can do whatever you want with a GUI, and you could write a GUI, which takes standard GUI code in, like, ads on, and also you pay an extra percent to this address.

And like, those are the orders that it generates. And people could choose to use it or not. So you could widen things out more if you wanted. There's no strict restrictions on that. But the sort of default thing that's actually built into standard on-chain code here is putting 20% of the fees to the GUI owner's address. And sort of the cool thing because it's all on-chain and permissionless is that anyone can do this. It's not like you work out a deal with someone. It's like there's code.

you can run it, you can build your own code, you can point it whatever you want, and you just like start running it. I want to ask you also about cross-chain swaps and how they work and which chains are compatible. Is it dependent on, you know, chains having compatibility with Solana or, you know, do we have bridges, for example, for Bitcoin or this sort of thing? It's a good question. And the basic answer is, I mean, sort of the stupid answer is, I don't know, it's whatever

people, you know, people can build whatever they want. But, you know, the real answer is sort of like, okay, what had people built? The first bridge live is the solid bridge. So something's built into one of the Solana wallets. And this has bridges basically for ERC 20 tokens, Ethereum, and Bitcoin. And you can sort of plug in any ERC 20 token. It'll create an associated SPL token. And you sort of convert back and forth between them. So it's sort of the first one that went live. And it's like fast and cheap and generalizable, but not the most decentralized. And so the next one, which I know about, which is coming on, is the wormhole bridge. That's slated to come out in a couple of weeks, I think.

I don't know exactly what come out means. Like there's going to be an auditing period and stuff like that for it. So it depends on exactly when you define it's being live. That's going to be an Ethereum to Solana Bridge, which is going to be, it's going to be more decentralized. And basically, I think the way that it's going to work is basically it's sort of like a multi-sig bridge, but the signers of the multi-sig are going to be, I think, a large subset of the Salana validators. And so because of this, it's like you sort of have similar-ish security guarantees to what you have to the Salana blockchain in some ways because it's sort of the same people who are signing both of those all. although not as much and obviously not the same as Ethereum. And so I think that's going to be sort of the next thing coming online.

And that'll be, I think, for arbitrary ERC20 tokens to SPL tokens, SPL, SPL being the Salana token standard. And then the eventual product here, which is, and you know, there are other people who I think are working on bridges as well, some Bitcoin bridges, some Ethereum bridges, and some bridges with other blockchains. And of course, people can use whatever they want. Like in the end, these aren't sort of, like officially sanctioned, which is just like, you know, they're all sort of like different

SPL tokens that you can use. And we can create an SBL token and can sort of deem these to be what they want. And then the other thing is that coming out at some point, all this is going to be a much larger production is going to be full light clients with fraud proofs, at least between Salon and Ethereum. And this is a sort of thing which is like, in some sense, it's the holy grail of cross-chain swabs. And it's really finicky.

The core process is not that difficult, all those, but there's just a shit ton of edge cases you have to deal with for it. And so, you know, that's, it's going to take time, I think, for that to come online. But that is something that people are working on and that I would guess within the next year will be online. And that will be something where it'll be more or less fully decentralized and that anyone could challenge it. Anyone can sort of like call out bullshit if there's an incorrect swap done. And anyone can sort of like get a swap through if it's, you know, if it is correct. And so that will have basically no centralized or even multi-di-sig reliances. But sort of as a part of that, it's going to be on-shank code, which means that you have to get it right.

You have to get it fully right. And it can't be like, there can't be like an and in complicated cases clause. and so it has to understand forks and split chains and, you know, everything else. So one of the things I've noticed, you know, I think I mentioned it on Twitter once, was, you know, you use the term like serum ecosystem quite interchangeably almost with Solana ecosystem. And you put out this like call to like, oh, like start building on the serum. Essentially you were asking people to build on Solana, but you called it the serum. ecosystem. If I go build something on Solana, how

meaningfully is it part of the serum? What do you define as the boundaries of serum? Yeah, that's a really good question. And of course, in the end there's answers, I don't know if they're just words, but but really how do I think about it? I sort of think of like most things on Solana probably being part of the serum ecosystem. And if you want to sort of like dig into it. Maybe it's something like anything which composes with products on serum is part of the serum ecosystem. And so if you want composability with a serum order book or with a serum AMM or with a serum pool or something like that, then you're sort of like participating in some

way in the serum ecosystem. And so if you want, you could sort of like start with the serum token or a serum decks or anything that has fees which flow through to a buy and burn or yield for serum tokens or something like that and then grow out with things composing on that and things composing with those things and anything that's sort of participating in that. And I think that that's sort of how we think about the serum ecosystem. I think that means that it's like not that different from the Solana ecosystem. And like practically I generally think of like Solana as the blockchain and and serum is a lot of the ecosystem on it, you know, if someone's sort of building a standalone application on

Solana, then it's like not clear it has anything to do with serum. But most things that are interfacing with Solana are interfacing with serum as well. Why not build your own blockchain? Why not just fork Solana and like have a sovereign chain for the serum ecosystem and, you know, bring more value to the SRM token. By the way, if you just talk about what that token does and everything, but yeah, why not do that? It's a good question. And, you know, part of the answer is that is basically that like it's hard to write a good blockchain. It's even harder to write a great blockchain. You know, I think it's one of these things where like if we wanted to dedicate all of our time to just the blockchain, then I think that would be very tempting to do.

But that would sort of take away from everything else. And to some extent, we want to be able to build with the product that's already there because that makes our job a lot easier. And that sort of like naturally substantially increases the set of people who are, you know, the people who are building on this and the people who are part of the ecosystem. And I think like one thing that you see with, you know, other, you know, you can take like the various finance chains and things like that. You know, one problem that you often see in cases like that. And there's, there's a lot of advantages to it. And by some metrics they've done quite well. But, you know, one problem with those is that it's often the blockchains themselves don't really ever get a whole lot beyond like a fork of Ethereum or a fork of cosmos or something like, you know, they usually get like, or really maybe more of the fork of Tron. I mean, it's sort of like, if you're thinking of like OG, you know, basically like proof of stake version of Ethereum, you know, I think that that's sort of like one place that you'd look.

And, you know, I think what that means is that, like, you end up generally not having super impressive blockchains themselves. They're generally, like, perfectly okay blockchains. But that's not the kind of thing that's going to scale as much as we'd want. Let's talk about the token and what its role is in all this. There's two tokens, I think. There's this SRM token and then the MSRM. Oh, yeah, the mega serum. The mega, yeah.

Yeah. So serum serves the core token of, you know, well, of serum. Basically, what's that mean? Fees from a lot of things get paid to serum or, you know, go to a buy and burn of SRM or yield for it or something. It's, you know, they're sort of staking rewards and staking yields relate to that. And it's also a governance token of the serum ecosystem. And so it has, you know, governance rates over a bunch of things in it, including, you know, the structure of of fees in it and, you know, token naming services and sort of like other things like that. And it's sort of organized through nodes, which are basically collections of stakers staking together.

And there's a lot of other things which are not sort of necessarily originally defined, but which have sort of started to grow up and probably more will. And sort of one good example is EcoSyrum, which is a serum node, which is sort of dedicated to helping projects build on serum. And so basically the, you know, sort of stake is there. It's a combination of, you know, coordinating with people, putting them in touch with right, people giving help and feedback, and also giving grants to people. And so there's a bunch of SRUM and fees set aside for grants for people building on serum.

You know, that's sort of helping a lot to grow out the, you know, serum ecosystem. And, you know, sort of more things like that forming. What is mega serum then? you get a fee discounts for holding serum and stuff like that. Mega serum is a million serum put together. So you can take a mega serum. You can crack it open into a million serum. And it can take a million serum and you can bind them together to create a megac serum.

And it's basically like a million serum, but a little bit better. Like it's a million serum, but like you get a little bit higher rewards, you know, higher fee discounts. And sort of everything else, you know, wherever you see serum, you know, probably get a little bit more for a mega serum than a million serum. and so why isn't everyone just turned their serum into megacyserm? The answer is they're scarce. So there's only a thousand megac serum. It's enough for a 10% of all serum to form it. And so it's sort of like if you want to be a sort of like big, you know, core player in the ecosystem and use it a lot, you know, trade a lot there, you know, govern, vote, stake, and everything else, then you sort of get a bit more of a boost if you do it through mega serum.

And so it doesn't sort of like really increase the total number of serum except for like lock up, you know, serum in order to form them. But yeah, it gives you a bit of a boost and is, you know, trading at a premium to a million serum because of the scarcity. Interesting. So this token is sort of like a golden ticket sort of token that that gives you some additional privileges and benefits on top of already on top of the ones that serum provides. That's right. Okay. And so you mentioned governance. I would like to dig in here a little more and ask you what is the governance mechanism and what is it that token holders govern upon? And is the governance simply like a like a steak baked governance like 100% stake based or is there some other variable here that that kind of weighs on the governance? Yeah. I mean, it's basically it's staking based governance. And, you know, it's sort of like organized around, you know, it's. It's organized around nodes, although anyone can form a node.

And, you know, basically nodes can have their own sort of standards for how voting works within them, whether you want to do sort of proportional things or just like a majority vote of the node to determine how the node votes. And then nodes are, you know, they're basically ways to organize. And then it has, you know, governance rates over everything is a little bit strong. And in particular, serum is sort of like some parts governance and some parts immutable. And this is to get around issues where you might be worried that governance would like do something crazy and brick your tokens or something. And so, you know, governance doesn't have the power, for instance, to like change in existing market on serum so that all funds go to someone. You know, like, like it doesn't have the ability to steal funds like that.

Instead, what is the ability to do is like, you know, vote on what should the standard. be for new markets on serum, how should fees work? And what should be done with those funds, you know, including, for instance, voting in Ecoserum, and, you know, how should nodes work? How should voting in nodes work and things like that? On the scalability aspect of like serum slash Solana, one of the things to take into account is how many like total transactions per second they could do. But also Solana has really fast block times, right, like, you know, less than like half second block times. So to what extent is this like more of an anti-feature where we're like recreating more

HFT-esque? Obviously, this is nowhere on the scale of like HFT. Just creating that kind of environment again. Are there social benefits to actually having things run slower, maybe using like batch execution, things like this? There's a few things there. And the first I'll say is like, are there benefits to it? And there's a few layers of this. But one thing that I'll say is that like the human time scale is an important one. Right. Like having things that can run at human's time scale is worth a fair bit. And this actually gets it around there. Right. Like how quickly do humans operate? Humans operate on, you know, about a one second time scale. You know, half second time scale. Human reaction time, it's like it, you know, something like a tenth of a second, you know, maybe a little less. And so, you know, you know, So what this means is that you can run things that aren't frustrating to use on it. And so one example of that is, you know, trade settlement within a second, which is sort of like about the amount of time that like a human user would care about.

It also means, you know, as an example, right, if you want to have messaging applications or news or something else that people are sort of consuming in real time, it has the speed that, you know, it has sort of like the block time that is necessary for that. Now, if you got neither two orders of magnitude off it, it wouldn't do a whole lot for that. Because already it's sort of like, you know, about as fast as we expect. But, you know, you look at anything from, yeah, from communication to gaming to settlement, and it gets it down to the scale where it feels natural and smooth to humans. I guess I'm talking more specifically about order books. Yeah. And so, right.

So if you're looking more at order books, there's some volume which you need like microsecond scale for. I think that volume tends to be non-perbook. worth a ton. I think it is good, but I think it's just very, very, very, very slightly good. And that like the amount that you sort of get, even though it's maybe most of all volume, I don't think it's most of all important economic information. But one way to think of this is like how much more economically efficient does a system get if it gets faster or does it even? The answer is it does. And how much, I think the right proxy to use for this is basically how much do markets move during a settlement cycle. However much they move during a settlement cycle is about

how inefficient your market's going to be. And I can sort of go more into that, but I think basically retail just ends up losing that amount whenever they trade. And so if you think about, you know, how much you markets move during a microsecond? The answer is about nothing. How much do they move during five minutes? The answer is sometimes a lot. You know, the answer can be a percent. you know, how much do they lose during a second, I don't know, a basis point or two? You know, I think that's sort of like the scale that you're looking at. And so I think what that means is that if you get a lot longer than the sort of salon scale block times, you do start adding in, you know, basis points and tens of basis points of uncertainty into the pricing

and of economic deadweight loss and, you know, failure to do transactions that would make sense and increase cost for retail associated with that. And you have some at the salon of scale, but I think that it's less. I think you're talking about a BIP or two, which is not the end of the world. And I think that if you could get it down by another factor of 10 to 100, you would get back most of that BIP or two. And so I think that's sort of what you're fighting over. And anyway, I do think that it actually matters,

that it actually does matter for users. And first, it's super frustrating to try, you try and do a trade and you're sitting there for like a minute, not knowing if it happened. But second of all, I do think that just in terms of rocket economic efficiency, it is valuable. Also, you know, last week on Twitter, you got into quite a large, like, you know, multi-thread debate about AMMs versus order books. And, you know, you and I have discussed this before. And, you know, I think we were generally in agreement that order book, you know, just to summarize, I think, like, you know, we both sort of agree that order books are the end state of any decently large pair, AMMs are artifact of, like, slow execution as well as they are useful for, like, tail end.

of assets. You know, one of the interesting arguments I saw from, like, the Twitter debate last week was about, like, this notion that, like, obviously one of the benefits that AMM's have is the ability for, like, just completely dumb retail to provide liquidity, which is something that they couldn't do before. And the users want to use the AMM because they know that they're not, like, being adversely selected against. Really, it's like, the LPs are the ones being adversely selected against while in order books it's usually the retail users who are and so the users are want to keep using the AMMs and they're willing to pay the larger fees for like the peace of mind of not being adversely selected against and that makes it worth it for

the LP providers and it's like this mutually beneficial cycle what do you think about this argument I think it's wrong um I think in particular that like who ends up winning from well first all I think there's just a deadweight loss to AMMs I think a lot of trades that should happen don't because of this loss. But to the extent there's a winner, who is it? I think it's the opposite of what people sort of think. I think the winner is the HFTs. Like, who wins from the existence of AMMs is people picking off the liquidity

providers, right? The people running bots, they're just trading against LPs. And there's a lot there. Like, there's significant loss each day from that. You know, we're talking tens to hundreds of millions of dollars a year that are lost. But the liquidity providers are the ones losing. That's right. Right.

Not the users. Well, who are the liquidity providers, right? It's sort of unclear and it depends on whether you're viewing AMMs as they are now or maybe as they're originally envisioned. Where as they are now, it actually is sort of the quote unquote users often who are the liquidity fighters and they're providing liquidity to get staking rewards. If you turn back all of that, if you imagine a world with no staking rewards, right? And it's sort of an unincentivized system. And then you say, who are the liquidity providers? There's were none.

I mean, not literally none, but like the liquidity was really, really bad in these things six months ago. And so I think part of the answer is like, I don't know who the liquidity is going to be long term. Like, certainly it's not going to be HFT firms. It's a terrible place for pride. It's retail they're just going to lose on every move. Right now it's been okay. It's been compensated for by all this yield. And again, you can drop enough yield into any structure such that it's what people

use. But I don't think that the current system, you know, survives in the form it is if you get rid of the yield. But the problem is like, you know, if all the users refuse to move to, like the traders refuse to move to a system that's better for the liquidity providers, what eventually the liquidity providers have to give in and provide liquidity on the AMM? Oh, at some point, that's definitely right. If takers, non-HFT takers, so people with natural demand are willing to pay 5%, then, yeah, to trade on an AMM instead of an orderbook, then yeah, they will. That's what's going to happen. It'd be a pretty weird state of the world. I don't know why they would decide that.

And when you zoom out a bit, they haven't. I mean, when you zoom out a bit, like almost all flows going through order books. Like the flow on AMMs is a small fraction of the orderable flow right now. Now, it's not true in defy. In defy, it's mostly going through AMMs. But a lot of that's because it's all on Ethereum and Ethereum can't have order books. As a market maker, like, so, you know, one of the things with AMMs is most of the liquidity on there is, like, unused liquidity, right? Because, like, you know, you're really only trading at, like, a certain tiny slice of the curve.

How much more efficient capital-wise is, like, Al-Mita, for example, than, like, a typical U-Swap curve? Yeah, it's a really good question. And it depends a lot on exactly what setup you have and on where you're trading. But like most places you're doing margin trading as well. So if you add that in, right, like margin trading versus AMMing, you know, I think the margin train gets you maybe a factor of two or three boost. Three boost maybe. And then also you're not using like 80 or 90 percent of the AMM liquidity. And so I think the answer is like, you know, I think you're talking like 10 times as capital efficient or something ballpark.

So I'd like to, as we come towards an end here, bring it back to the defy ecosystem and ask you, where do you think the defy ecosystem is heading? And what is needed in order for defy to gain more legitimacy in sort of a broader financial world? Yeah, it's a really good question. And I think basic answer is the first answer and the biggest answer is scaling. And just literally can't get more acceptance right now. Like you can't do any more than it is because it's out of throughput. So it has to start scaling. I have to scale a ton.

Outside of that, I think it has to be able to be faster and it has to get rid of failed transactions or at least have them be pretty rare. And sort of more generally, it has to get to the point where it feels like you're not making a big compromise by using it. Right. Like, yeah, sure, it's one thing if you're like a big defy person or even a big crypto person to get into defy. If you're talking to Visa and you're telling them they should do defy, right? First, they're going to laugh at you. And like, then if you somehow have managed to get through to them, they're not going to be like, great.

Is it no more than like three orders of magnitude worse than what we're doing right now? That's not the standard. The standard they're going to be is, is this a better experience, right? And maybe you'll be like, no, but it's just as good and you get these other benefits. and then they might take you seriously. But you have to get to the point where user experience on Defi, where businesses can build on it without making real sacrifices. And that's a pretty big shift.

And what that means, really, what would it take to get there? Boy, like, first of all, again, like way faster, way higher throughput. If you're talking finance, it has to be able to support the infrastructure they're used to, like market making, order books, OTC quotes, RFQ systems. If you're talking, you know, other protocols, it has to kind of feel like they're just using a database and they just have to feel not super constrained. And so, yeah, I think that's like a ton of tooling, a ton of adoption, a ton of things to compose on that are valuable, a lot of useful primitives.

And the biggest thing is just speed, throughput, and cost. It seems like at the moment, a lot of the enthusiasm, for Defi has to do with regulatory arbitrage. And I wonder if that regulatory arbitrage disappears because of more strict regulations, that the enthusiasm for Defi will kind of dwindle. But it seems like what you're saying is that the enthusiasm and the adoption of Defi needs to come from scaling from larger financial institutions adopting it rather than say like enthusiasts and hobbyists and sort of the crypto space using Defi. It's a really good question.

I don't know. I mean, it's an important question. It's not just I don't know. I suspect no one knows. I suspect regulators don't know. Like, I suspect that there are people starting to ask this question right now. And a lot of related questions. And I sort of like, oh, that's interesting.

Hmm. Yeah, we should think about that. You know, like this is, crypto is new. Defi is newer. And it's so far afield from what regulars are used to contending with that, you know, I think that there's going to be a lot of hard questions that they're going be asking themselves about what exactly they mean defy means and what exactly their goals are and how do they feel about the fact that it's all on chain is the fact that it's like does it matter if it's

like privacy private versus not private are they particularly concerned about private defy like are there limitations on this does it matter what does a centralized party mean here who would be responsible for it. I think there's questions of like, you know, do they just need to have access to the Fiat ramps that plug into it? I think there's a lot of uncertainties here. I don't think it's obvious how it's going to shake out. Yeah. If you ask the EU regulator, at least in what it's currently put out, the answer to that is like defy needs to be basically just the same as the existing financial system with blockchain. And that means that none of the innovation that happens that currently happens on DFI should exist or like they fully just sort of

negate the existence of all that stuff and they're like okay D5 should be for like funding products or you know funding companies and this sort of thing like none of this other stuff even is in their purview at least in in the current draft regulation that that they've put out so yeah I don't know I mean I think that if regulators as we know kind of suspect they are being influenced by large financial institutions. That's probably the direction in which things will go. If regulators are more interested in sustaining innovation and building something new, then that would be good for the ecosystem or maybe somewhere, some in the middle.

And I think a lot of regulators aren't exactly either in either of those camps. I think a lot of them, they sort of have a role in a job, and they're trying to do the best they can at it, but it's complicated. and a huge surface area, and they don't know what the right thing to do is. And they just keep getting all of these more and more complicated situations thrown at them and have to try and do something they think is reasonable given that, balancing all of these things.

And sort of like trying to prioritize. And yeah, they're just trying to do the best they can. But it's hard. Like all the finances is a big amount of finance to be regulating. And the regulators are not as numerous or large or well, funded as all of finances. So I have like, just two last questions for you.

So one is going back to like the effective altruism stuff you mentioned early on at the beginning of the episode. What do you do to like practically do this? Like do you donate a percentage of your income annually to particular causes or is this like, yeah, how do you kind of do this? Yeah, it's a good question. And I think basic answer is like eventually it'll be pretty much. much everything. Now, if you do that right now, then you don't have the ability to keep growing.

And so I think that there's this tradeoff between, you know, business capital and donations. And so, you know, for now, donating some amount each year, trying to make it enough that it sort of proves not just to others, but also to myself that this is still the ultimate goal. And enough to make a difference in balancing that against making sure not to, you know, not to hinder our ability to keep growing. And I think that there's, there's a balance. there and there's a trade-off. Do you think there's a cool way to use like smart contracts to like, you know, make this more effective? I, you know, one thing that I'll say is like, I'm always a little bit nervous about, you know, giving up too much optionality. But I think that some of this could be

super cool. And I think that, you know, whether it's a smart contract that auto-donates some fraction of some revenue stream or whether it's just making it super easy to donate is worth a lot. Like for a lot of people like, oh, yeah, I want to donate. How do it that? I don't know. How do it that? I don't know. do that to a really good place. Boy, do I not know. All right, let's think about it next year. And so I think, like, just really clarifying and simplifying that process could be worth a ton. And then finally, what is like something? What's what in the crypto space are you most excited about outside of serum or anything that you're related to in some way? I'm really excited to see the future of institutional adoption of Bitcoin, whether that happens, how it happens. but even more so integration of crypto with the rest of the world.

You know, whether it's integration of crypto with Fiat, integration in terms of crypto but also non-crypto apps, whether it's, you know, adoption of crypto within financial ecosystems, again, whether it's using it as infrastructure or as a product that you can invest in and, you know, involvement in remittances. And yeah, I think just generally like super excited to see to see anything that you know, really helps tie the crypto ecosystem to the rest of the world. Cool. That's a great note to end on. Sam, thanks for joining us today. Of course. Thanks for having me.

Thank you for joining us on this week's episode. We release new episodes every week. You can find and subscribe to the show on iTunes, Spotify, YouTube, SoundCloud, or wherever you listen to podcasts. And if you have a Google Home or Alexa device, you can tell it to listen to the latest episode of the Epicenter podcast. Go to Epicenter.tv slash subscribe for a full list of places where you can watch and listen. And while you're there, be sure to sign up for the newsletter, so you get new episodes in your inbox as they're released. If you want to interact with us, guests, or other podcast listeners, you can follow us on Twitter. And please leave us a review on iTunes. It helps people find the show, and we're always happy to read them.

So thanks so much, and we look forward to being back next week.