175
There's obviously a lot of interest in crypto and DeFi these days. And while it's growing rapidly, it's still not cutting much into traditional finance business lines. For the most part, trading on blockchains is slow and costly. But some projects don't accept the premise that blockchains have to be slow and inefficient. Solana is an Ethereum competitor whose native token has been soaring. And unlike Ethereum, its transactions are cheap and ultrafast. So what tradeoffs does it make? And what projects are being built on top of it? On this episode, we speak with Solana founder Anatoly Yakovenko and Kanav Kariya of Jump Trading, who is involved with an oracle project called Pyth. Transcripts have been lightly edited for clarity.
Joe Weisenthal:
Hello and welcome to another episode of the Odd Lots podcast. I'm Joe Weisenthal.
Tracy Alloway:
And I'm Tracy Alloway.
Joe:
So Tracy, obviously we've done a handful of like crypto DeFi episodes in the last few months, obviously, an area of growing interest. I would still say, however, that for all the enthusiasm, like by and large, it doesn't seem like anything crypto DeFi is like, seriously cutting into traditional finance in a big way yet. Like, it still feels like a pretty separate universe.
Tracy:
Yeah. Well, this is a point that you've brought up on a number of those episodes, this idea that we have DeFi, but so far, it basically seems to be sort of self-dealing in crypto and in various tokens, there hasn't really been an extension outside of the crypto space.
Joe:
Right. It feels very recursive — kind of like a snake eating its tail, some interesting proof of concepts about how market-making works or the idea of like an automated market maker or sort of like collateralized lending. But yes, by and large, you know, if you were to sort of look and say what banks are doing, or what trading firms are doing, at this point, the two don't seem to be intersecting all that much. Or, you know, I can't think of like many trading lines or business lines within finance that are like, oh, we're losing money to DeFi or to crypto in some way.
Tracy:
Yeah, which kind of goes back to a wider point or one of the original criticisms of blockchain, which is that, you know, for all the excitement and all the hype, we didn't actually see that many real economy, applications of blockchain. And the fact that finance hasn't been able to make some of the technology from DeFi work, kind of hints at that issue. Although, I mean, I have to say there are regulatory hurdles to actually doing DeFi type things if you're a regulated financial institution.
Joe:
Hey Tracy, have you ever like played around or like, you know, a little while ago now, back in the spring we did that Hayden Adams one on Uniswap. Did you, have you ever like gone around or played around with Uniswap at all or seeing how it really works or anything?
Tracy:
I did. I started a MetaMask account and was playing around a little bit, but like I got to say, even starting a METAMASK account from Hong Kong is a nightmare in and of itself. Like just getting money onto it took me like a full day to figure out,
Joe:
I mean, like, it seems cool and it's actually very impressive. And there are some interesting breakthroughs, but like, it's just obvious it's not up to snuff for anything that's sort of like actual high-performance finance. You know, trading people are used to, like, executions, like millisecond-scale executions. The cost of trading is, for so many different assets, virtually free. And as everyone knows, you know, it's like, I saw you had a really good joke a while back about gas prices on Uniswap, like the cost of say, like trading on the Ethereum blockchain or whatever is like far above anything that would be like actual finance scale at this point.
Tracy:
Absolutely. And I mean, just the knowledge that you need in order to dip your toes in it. And again, going back to the MetaMask account creation, that took me ages, and I know like a little bit about what's going on. And then actually doing DeFi you know, picking what tokens you're interested in and then making the calculations for how much yield you can earn versus the gas fees. It kind of boggles the mind and it is worlds away from the kind of, I guess, service — is service the right term? — customer experience, that you would get on traditional financial applications.
Joe:
Right. And like, if there's a hot new NFT drop on Ethereum, gas fees, which I guess are more or less like the commissions, will shoot to the moon. Like obviously, if you want a high-performance trading environment, like serious trading, it's obviously not there yet. It's super interesting. It might get there, but it's not there yet.
Tracy:
Yeah. I would agree with that.
Joe:
So all of this is to say, to get to our discussion today that I'm very excited about, and we're going to be talking about a crypto blockchain initiative, which is extremely hot right now, lots of interest. And it's kind of like, what I would say is it's going after sort of like Wall Street-grade, finance-grade finance. And by that, I mean actually attempting to, you know, the idea that maybe you can have a blockchain, but it actually is at the speeds that Wall Street is used to. Actually at the trading costs Wall Street is used to, and, you know, I think it's interesting because one of the things that people say about blockchain is like, oh, it's a bad database. A blockchain is just a slow database. And for most use cases, it's not what you want. We're going to be talking about a blockchain project that actually is attempting to not just be a slow and efficient database, but to actually be high-performance.
Tracy:
Yeah, let's do it. I'm excited.
Joe:
All right. I'm super excited about this episode because it's super hot and people are into this stuff right now. We have two great guests on. We are going to be speaking with Anatoly Yakovenko. He is the CEO of Solana Labs, and he is the creator of the Solana protocol, which is this like smart contracting blockchain-crypto platform that's kind of like Ethereum — some similar ideas. And we're also going to be speaking to Kanav Kariya. He is the director of strategic projects at Jump Trading. So, a well-known trading firm, and they're doing a lot of interesting crypto stuff that we'll be getting into. And so it should be an interesting conversation on the sort of like marriage of technology finance, and an attempt to do crypto at true Wall Street-grade, finance-grade. So Anatoly and Kanav, thank you so much for joining us.
Anatoly Yakovenko:
Awesome to be here. Thank you Joe.
Kanav Kariya:
Thanks for having me.
Joe:
I'm super excited about this one. The timing is great. Well, you know, we're recording this August 18th. Most of the big coins are still well below their highs, but people are super into Solana. And I think people are going to be excited about hearing this one when this comes out. But you know, so Anatoly, I want to start with you. I mean, this is this cliche that people have said for a long time, blockchains are bad inefficient databases. And I think a lot of people are just sort of acceptive of that. It's true. That there's like this trade off that you have to make. It's like, okay, you can like be decentralized. You can be permissionless, you can be censorship free. But the price of that is high cost of execution and slow cost of execution. And that more or less, I would say characterizes Bitcoin and Ethereum, at least right now. And it seems like Solana essentially attempts to say, no, we don't have to accept the inherent inefficiency. Like, what is Solana, what is the goal and how do you think about that trade-off?
Anatoly:
Yeah. So the trade-off you described is kind of this thing that people call the trilemma, right? Decentralization. Performance. Security. And that trilemma really only applies if the network tries to exceed the bandwidth available to it. So bandwidth is, you know, what you get out of Time Warner or whatever, out of AOL, they give you one gigabit at home and a lot of places in the United States now, and that that's really what they mean by bandwidth. Well, if you look at Ethereum and kinda like these, you know, proof of work-based networks, they weren't designed to maximize the amount of bandwidth that each system can use. They really weren't designed to soak up one gigabit. That was really not something that they were built for.
So my background, if you folks don't know, I spent most of my career at Qualcomm. I was there from like flip phone days when really like there were these dinky little devices, 2003. And when I left in like 2015, my team was optimizing augmented reality of a supercomputer basically. So I saw this like massive improvement in hardware in just the span of 10 years. And I also saw what real bandwidth looks like. 5G, that we were, you know, was in R&D stages while I was there, is designed to give you one gigabit — well, two people are driving one in China, one in United States, they should have a one gigabit byte directional channel between them. And if you try to fit transactions over that channel, you can stuff about 700,000 transactions per second. Like Ethereum-size, Bitcoin-size transactions. So the only thing that's missing really is the hardware to handle it. And that's the challenge like, can we build a system, a fast database that could both process these messages, all the cryptographic signature verifications and retransmit them around world as fast as possible. So that that's really what we set out to do. We had some really clever insights, like using a verifiable delay function as a source of time income before consensus and, you know, using GPUs and AVX and a bunch of hardware optimizations for, you know, how the runtime works, how the execution environment works. But a lot of these things are engineering, like, you know, hardcore engineering challenges, but not computer science problems.
Tracy:
So this idea that you've sort of solved the trilemma or the idea that a blockchain can only outperform in two of three areas. So decentralization, security, and scalability, can you maybe go into a bit more detail about how exactly you do that? When you say it's more of a hardware issue versus computer science, what do you actually mean? And how does the whole thing work?
Anatoly:
So you can imagine a single computer that's really fast, right? You send it transactions, it gives you a response back, Hey, I'm done, right? So that's something everybody can imagine. If there was one computer that did this work. So there's this thing called time division multiple access, which is how 2G cellular networks work. And the way they work is you have your channel, your bandwidth, your physical frequency signal, and you have a bunch of subscribers that want to transmit at the same time over it. Well, if you allow them to just meet at the same time, you get noise because radio interferes. Same with blockchain. If you have to block producers, two computers, they're really, really fast — to try to produce a block in the LEDGER at the same time you get a fork. And the network is in a noisy state.
So this idea that we had really early on using a verifiable delay function or proof of history, if anyone has heard anything about Solana, they heard a proof of history. This is a clock that is outside of consensus, and it rotates the button of when any block producer can transmit a block. And it does it in a very predictable deterministic way. And it does it really, really fast. So it experimentally we've done it in 400 milliseconds. So every 400 milliseconds this button moves. Right now, the slots are 400 milliseconds, but we move the button basically every four slots.
So you can think of this different really, really fast computer around the world gets to be the next block producer every like roughly 1.6 seconds. And because everybody knows ahead of time, this is a really fast computer that's going to start, you know, start creating these blocks, you don't run into a lot of the bottlenecks that you run into on Ethereum and proof-of-work networks and all these other kind of random base coordination networks. And this is like 2G cellular networks. This is stuff that like basically I had to learn as part of my interview at Qualcomm.
Joe:
Kanav, why don't you come in, talk to us a little bit about what you do at Jump Trading with crypto. I mean, Jump Trading, I think people may or may not know, maybe you could just describe what it is, what it's doing in crypto and sort of like what the, I guess to you know dovetail with what we're talking about, Anatoly’s been talking about, some of like what needs to happen for crypto to get to the scale where it's worth it for serious finance players to be involved in it.
Kanav:
Jump is a coordinated trading firm that was founded about two decades ago in the pits of the CME, and is one of the largest participants across traditional markets, spanning most asset classes. And Jump’s crypto effort began as a skunkworks intern project six years ago at the University of Illinois where Jump has a research lab, and our involvement in the space has grown pretty dramatically over the last six years. And, you know, I kind of classify what we do probably in two buckets. So one is prop trading, which is exactly what we do on the other side of the house, where we connect to a lot of markets and participate very actively across the crypto landscape. The second piece where this conversation has felt really more interesting is the strategic bucket, where we've been more been partnering with and investing in pieces of infrastructure across the crypto space.
And that kind of started with centralized infrastructure, with exchanges, custodians, all the picks and shovels. And there's, you know, we walked into the much more exciting spaces participating with onchain protocols and projects, such as Solana and contributing, you know, much more hands on … projects like the Pyth network, which I'm sure we'll get a chance to chat about a little bit today, I guess. Tackling that question of, you know, what it's going to take to bring finance to crypto, well projects like the Pyth network are definitely one piece of it and I’ll save that for maybe a little bit later in the interview, but like Anatoly was, you know, mentioning, there's a lot of scalability that's yet to be had and that's needed in order to come out of this, the sandbox that we've been laying in with for the last five years. Right. And so if you want to build finance that can execute on open order books and process a lot of transactions every 400 milliseconds that you can facilitate a lot of meaningful risk transfer, you need a blockchain or, you know, something like Solana that can process and be that execution layer to facilitate that. You know, there's a lot of the stuff that you guys touched on earlier in terms of user experiences, and other, you know, problems that have been talked about a bunch, but I can tell you that through the course of our participation in this space, we've seen dramatic improvement in the quality of access that’s available. And as more and more folks like us get involved and as more projects like Solana continue to reach maturity, and more capital continues to enter the space, these are all, in my view, pretty solvable problems that are being dealt with.
Tracy:
Anatoly, I'm curious, just on that note, when you originally set out to found Solana, what was the goal or ambition of the product? So a lot of people have described it as an Ethereum killer. Was that the ultimate aim to create something that's faster, that's more scalable and exactly what applications did you have in mind for it?
Anatoly:
The slide deck, the seed-level slide deck literally said blockchain at Nasdaq speed. That was the tagline. And we were going after, like what I thought we would be going after are like these monopolies, like Nasdaq, like Nyse, like CME. Because, you know, for whatever reason, I started trading on like Interactive Brokers and a bunch of forex sites. And in my experience as an engineer, I was always like a little behind on when I got the data. When I got the information from these places. And when my orders got submitted, they're always a little later than everyone else's. So I always felt like I was always getting screwed by somebody else that had access to this financial backbone. And this core thing that blockchains are, that differentiate blockchains from databases is this idea of censorship resistance. So if we have a really fast blockchain and all you need is hardware to connect to it, and you're on the same level playing field as, you know, Jump Trading or all the best traders in the world, that's really something that I felt would be good.
You know, that's the product that I want it. I am blown away by the progress over the last year, having folks like Jump Trading, you know, to have their engineers start building lists and really take this seriously. And it, you know, it was a dream and kind of a silly idea maybe, a silly tagline in a slide deck, but now it really feels real. Like there's I think there's a chance that financial execution trading could actually run on a Solana in the next five to 10 years for the majority of things that are traded in the world.
Joe:
So you mentioned censorship resistance and of course, like, you know, this is a core value like in Bitcoin. And when I think about it in the Bitcoin realm, you know, Satoshi's vision, you know, I think about like, okay, I could send a transaction from here in New York to Tracy in Hong Kong and no third party needs to know it. And no third party can say no, and no third party can say, oh, you're not allowed to send that much across borders to Hong Kong because it's kind of like, they're, you know, it's like that cypherpunk vision that it's just between me and Tracy and no one else is involved in the transaction. It feels like it means something a little bit different in the sort of like pure finance context where you're talking about, oh, you don't like the fact that as a client of Interactive Brokers, it feels like there's someone who maybe has a computer in a server room in New Jersey, who's a millisecond closer to the exchange or whatever. Talk about what censorship resistance means in the context of like, you know, trading interest rate swaps.
Anatoly:
So there's, you know, these transactions, especially trades, is information that's propagating around the world, and you can think of it as kind of chasing news. So some newsworthy event happens in Singapore, that newswire, you know, fires off to a trader that's looking at that Bloomberg terminal, and they look at a market and then they make the trade. The goal for us is to have state transitions — like transactions — to propagate at that same speed, speed of light through fiber. So by the time that trader looks at the markets, they see the exact same price, CME or NYSE. And because of how the information propagates it's simultaneously to every computer that's part of the Solana network, and anybody can join that set. It means that me as a Joe Schmoe Anatoly, that wants to play around with deep learning and make my models, I get that data as fast as, you know, the best traders in the world. And therefore I can make my trades based on how good I am at the same kind of level playing field. So it's really like, I make this joke, it's kind of maybe a little morbid, but I say Solana is really great if you're building a nuclear first strike detector, you actually want to see the rockets launch, and you want that signal to fire. Bitcoin is really great for after that strike lands and you need to like rebuild society.
Tracy:
So, I mean, there is a tension with blockchain and regulated financial institutions that we kind of touched on earlier. And of course we've seen Gary Gensler come out recently and talk about how a lot of tokens look a lot like synthetic securities. So I'm just wondering, how does the permissionless aspect of a blockchain stack up against the highly, highly controlled and regulated world of financial transactions?
Anatoly:
So a blockchain, especially like one, like Solana, it's really like a very dumb packet switch. It's really doing nothing more interesting than AT&T does except it guarantees this censorship resistant piece that if I send a message it’s delivered to all the subscribers and the fault tolerance and all of these consensus algorithms guarantee that part, that there isn't any central party that can stop it. So that is really like very dumb work, right? The validators that do this, they're not aware of the bits that are sending. I feel like where regulation should step in is at the places where somebody is saying these bits in this computer, in this packet switch, represent something of value and I’m claiming that they represents something of value to the public. Because that's the thing that is kind of like, you know, effectuating the bits into something that has, you know, trust, assumptions that people will look at it and say, okay, is that really the UniSwap token? Is that really the Bitcoin token? Who am I sending my money to? That's a perfect place to regulate. The actual bits, how they're transmitted. That's really like dumb packet switch work, you know, like, I think what we've seen, especially in the last, you know, during the discussions around the amendment from the infra bill, I feel like those folks got that part and actually started moving towards that direction.
Joe:
Kanav, maybe you could come in and talk about it from the perspective of a trading firm that has to think about regulations or think about what is securities and has regulatory obligations. Because it does feel like, as Tracy said, this seems like it's going to come up more and more. It's like, what are these, are these securities properly registered? Like, how do you see this playing out from your perspective? And maybe you could just yeah, give us Jump's perspective or your perspective.
Kanav:
Yeah. So as you pointed out, Jump has a massive body of folks that are constantly monitoring the situation and adjust all this information as it comes in. And there's a lot of shades of gray, right? Our participation in this space has been defined by a lot of the activities that we've been able to get comfortable with. You know, one of them being contributing data to the Pyth network as a very neutral again, like Anatoly was saying, like kind of like a dumb packet switch thing, where we're contributing data that's helping bring pricing information in a high fidelity fashion to the blockchain. And that's a kind of neutral piece of infrastructure that can be leveraged to build a lot of things.
Joe:
You know, I'm very curious, if you could talk a little bit more about this, the Pyth network, because, you know, we did a DeFi episode, I think probably about a month and a half ago with Tom Schmidt, DragonFly Capital. And he talked of one of the ways in which this space could move forward is essentially sort of like through kind of like synthetic assets that use outside oracles to bring in pricing, bring outside pricing onto the chain. Talk to us like a little bit about what that is. What do you, what is Pyth? What are you contributing to it, and why? Like, why are you contributing data to this network?
Kanav:
Yeah. so Pyth is effectively a high-speed data bridge between the rest of the world and blockchain and in this case specifically Solana, right? And so blockchains for all their strengths don’t natively have the ability to access data that lives outside of the chain or off chain. And that means you can’t incorporate this data into the logics of smart contracts and applications. And that inhibits you from being able to build a lot of interesting stuff. And so Pyth is like a, almost like a decentralized marketplace, an aggregator that enables first party producers and owners of this data to contribute this price that helps build this piece of infrastructure to bring the state DeFi-enabled applications … The reason that us, and, you know, a lot of trading firms and a lot of the other guys that have announced participation in the Pyth network are really excited about this is, you know, quant firms have always been on the forefront of technological development, almost as a precursor to being relevant in this business. Rught?
And that's generally been more in the finance and kind of focused software space, but with the rise of blockchains and decentralized finance, it's an opportunity for us to be at the forefront of a completely new technological revolution. And we'd be, you know, we've been in this space for the last six years, but a lot of the other participants have been evaluating coming in in various different aspects. And this has been a really great way for a lot of people to get their hands dirty, own a pair of private keys, send a transaction through the peer-to-peer network, and start learning … [inaudible] … You can't really understand everything you do before you do that. So that's kind of piece one.
And then piece two is trading firms have generated a lot of data historically and managed all of this data in different ways and exchanges, as I generally model them are just fintech platforms that are looking to leverage their technology in the name of building other cool stuff. And so when you have something like Pyth that enables people to contribute this data to effectively make a blockchain data play, as you know, this oracle problem has become more and more predominant … [inaudible] ... It's an opportunity to contribute to something in a neutral fashion and get exposure to the space and have a play. And that's why, you know, we’re very excited and I can't speak for other people but, you know, through our conversations, that’s been one of the reasons why a lot of folks have been excited about it.
Tracy:
So one of the reasons people are very excited about Solana at the moment is because the native token, the price of it has basically gone kind of crazy recently. I don't have it right in front of me, but I think the spike was like bigger than Bitcoin recently. Just a lot of outperformance there. Yeah. So I'm curious, Anatoly, when you look at the price of Solana and it's going up this sharply, what is it saying to you? Is it saying that people are more value in the Solana network itself and they're willing to pay more for what is in effect like a processing storage fee, or is it pure speculation and people just sort of, you know, having fun with crypto?
Anatoly:
That's really tough, right? That's a tough question simply because I think the crypto markets have matured enough to recognize the value in smart contract platforms and that value isn't just in the processing or the data, it's in this like ecosystem, right? The shared state. The reason why Ethereum is so valuable is because there's so many companies that have built products that people want on top of it. And these products people want so much that they're willing to pay these exorbitant gas fees to use them, right. Like that is the value of Ethereum, that happened. Right. So what I think, if anything I think the price is reflecting that the ecosystem in Solana over the last year has grown really dramatically. Like we saw that in our hackathons, our first hackathon had a thousand registrations, maximum of 3000, last one had 13,000. 350 teams actually launched, you know, something. A bunch of them, I don't know the exact number, but I feel like it's getting close to 50, have raised funding just during the hackathons. That means that there's now an ecosystem of teams, right. Startups that have quit their jobs at Google or whatever, and created a product and have raised outside capital. Product-market-fit, eat glass, grow users. Right. That's really, I think where if anything, this is reflecting.
Joe:
Let me ask a question and I think it could actually be answered by both of you, I'd like to hear both of your perspectives on it because although we've talked about, okay, Ethereum, it's slower, it has high gas prices. There are the so-called layer two solutions that people are building on top of Ethereum. They're also building them on top of Bitcoin, which famously has the Lightening Network, which has been around for a few years. And these layer two solutions can sort of solve this problem of extremely high throughput, low cost transactions. I'm curious from both of your perspectives, what you see as the advantages of having it be on the layer one and why not just say, okay, well, if we want, you know, security and if we want high throughput, why not just use one of the layer two solutions that are being built right now on top of Ethereum?
Anatoly:
I think like the magic in crypto happens in this idea of composability where everybody is in the same kind of state. We're all playing the same game on the same server. It's all basically super-connected. Layer twos create these fractions, you know, and like, you know, little shards of places where a state lives. And financially, it's very obvious that if you have a market that now has to be split between two different execution environments, two different exchanges, that creates inefficiencies because now you have arbitrage between the prices between the two. Now you have to have capital in both exchanges, right. And manage that. And that's really not a great thing when if you have an alternative where everything can be in just one giant pot. So that's a very financial kind of explanation, but I think just developers, you know, as an engineer, sharding is a huge pain in the ass to deal with. A bunch of different roll-ups where I have to manage state from application.
It's a huge pain in the ass. All I want is like, oh my whatever, 10 million users do not worry about any of this stuff, right? Like as an engineer, it's just easier, right? I want a single CPU with as much throughput as I can. Right. Ideally with a single core that's as fast as possible. Dealing with multiple cores is still a pain in the ass, but still not as bad as dealing with a network of computers that now I have to scale. And those problems are really like something that takes a lot more time for devs to build products versus infra, you know?
Kanav:
Yeah. Just piggybacking off of that, you know, the magic word that people throw around in the crypto space often is composability, right? And, that's, you know, what Anatoly was talking about in being able to leverage pieces of state within the same ledger to build cool applications. And so you can take something like Pyth prices, the CME order book, put them together and build a derivatives trading platform. And if you have sharded state, you know, those kinds of exciting applications are not possible, and that takes away from a lot of the fun that these avenues, these platforms bring. One thing that's … [inaudible] ... is investing in what's called cross chain bridges, which are pieces of infrastructure that enable state to move between these chains. And so Ethereum has all these network effects that exist across a lot of these dimensions. And I think all of them eventually converge into building very useful and interesting states. And if you're able to use these bridges to then bring the state over to Solana and perform more interesting combinations in that state and create new state with very interesting security properties, that makes both Ethereum and Solana, you know, in my view, a lot more interesting.
Tracy:
So I have a weird question but it's sort of related to the last point, but I think there's so much excitement around crypto and a lot of it is rooted obviously in an excitement about technology and its ability to change the world. But on the other hand, there seems to be a lot of belief that once this technology is invented, like say Bitcoin, it's not going to be replaced by something else and it's going to sort of exist forever. So I'm thinking how to phrase this. So, you know, if you have something like Bitcoin and then people say its use cases is limited, so they go ahead and invent Ethereum, and then people try to improve on Ethereum and come up with alternatives like Solana or like Cardano, do you ever worry that like the next iteration of the blockchain is going to come along and compete effectively with Solana? And then secondly, how do you sort of balance the tension between building a big network — so you want a sort of first mover advantage, you want lots of people to be using Solana — but on the other hand, you know, you could have a new competitor come out of nowhere with a better proposition.
Anatoly:
My belief is that Solana is kind of like at a terminal design for censorship resistant real time, get the information bits as fast as we can around the world. There's Pareto efficient trade-offs where if you're building something that is trying to survive, you know, as a monetary system after World War Three, maybe Bitcoin is a better design. But if you're building something for real time trading, trying to disrupt Nasdaq, I don't see a better path. So the network that's going to beat us is going to be very similarly designed to us, but just executing faster, you know, people working harder. That's what keeps us working as hard as we can. There's also, I think kind of this other interesting aspect of this is that the tech itself, maybe not as important as just empowering people with cryptography.
That onboarding experience that you had with UniSwap that was really painful with MetaMask. Imagine that, you know, it gets to a point where you have two, 300, 500 million people that have done it and kind of get the idea of cryptography at the level that people understand what a browser does. Nothing more than that, right. But if they get it, then you have that many people actually all now able to self custody, all interact with any arbitrary blockchain. That space is going to be filled by technologies. You know, like somebody somewhere is smart enough to go build a network and get those people to go do something, you know. The tech is going to be less important than the actual getting those humans onboarded.
Joe:
I want to go back to this idea of like, okay, competing against the exchanges, blockchain at Nasdaq speed, you know, one of the big ways that a lot of these exchanges make money is the licensing of data. That data is extremely valuable. And so I'm curious, you know, Kanav from your perspective, but maybe from both of your perspectives, is that an area that you see as like, okay, this is prime for disruption. A handful of very powerful exchanges really just control, have a hammer lock on this data, and you could open this up. But also, you know, I'm also curious specifically say from the Jump Trading perspective, and there's a bunch of trading firms that are involved in this, I know like Virtu is one of them, what is the guarantee that the data you're contributing is clean or that it's high quality data? Because it's what, you know, like how would I, if I were a user of Pyth or, you know, if I had designed a smart contract that was contingent on Pyth data, how would I have any idea that it's clean high-quality data that's being contributed to it by you or the partner?
Kanav:
Couple of questions in that one, I'll go for the data model. I may not have like the most exciting answer here, but in reality, I think the feeds that are coming out of exchanges are primarily consumed by two classes of participants. One are forums like Jump, like Virtu, like GTS, DOW, a lot of other participants in the fintech world that are ingesting this data to execute on like high-frequency trading strategies that need to respond … [inaudible ] … And modern matching engines have determinism on the order of microseconds, or some banks, you know, a lot of them. And blockchains, you know by construction, Anatoly has just spoken about, it’s really as fast as it's going to get, because, you know, you're limited by how quickly you can communicate to a global network of computers and come to consensus, especially on the public internet.
And so you're talking about like hundreds of milliseconds, right? And a microsecond is to hundreds of milliseconds as a second is to a day. And so you're talking about some pretty dramatic differences here. And so the data feeds that are coming out of these exchanges, the thing that's probably not disrupted by a blockchain based data feed is the level of subscription that a … [inaudible] needs to consume to interact with high-frequency trading strategies. Now, there is a second class of data consumers that are more human or more realtime, right? And these are the guys, the analysts that are picking up the phone on the desk and providing commentary on the yield curve, it’s back office systems across the board, like all the classes of participants that don't need data every couple of hundred nanoseconds, but don't really want a 15 minute delay.
And that data model now definitely, you know, starts to get disrupted with something like Pyth. What I’ll caveat that by saying is that what's far more exciting with a blockchain-based data play like Pyth is the on chain applications that you know, you were talking about, some of the guests that you've had, were talking about the applications that could be built using this data. And I almost think of the disruption of that second class of data model as basically a side effect of what's being built, of its you know, primary motive.
Joe:
And then what about the data quality part? How do we know that what Jump is contributing to. How do we know it's good data?
Kanav:
Right. So the way the system works is as a network of independent data providers that are effectively are all publishing their prices as transactions into the slot of blockchain. And just to get into the weeds a little bit here, along with their prices, these data providers are also publishing uncertainties or confidence of that are both associated with these prices. And that's kind of a non-intuitive concept when it comes to the prices, depending on the market structure, any given grade that you're seeing, or any given snapshot of the liquidity that you see on any given venue, inherently is just an observation that has some uncertainty on a bid price. It's like a measurement that you make in the lab. And these data providers are publishing this uncertainty and this price to the blockchain, or they're like Solana, there's an on chain program on Solana thats ingesting these inputs and doing some slightly intelligent aggregation to handle outliers to a way back data and create like a final bid price.
And an uncertainty that's on the prices, right? And so developers that are using these prices are basically seeing the aggregate of a lot of this data that's being ingested. And additionally have this like extra field of uncertainty that adds a degree of freedom to the applications that they’re building. And having this big diversity of quarters is critically important, right? There's a lot of idiosyncrasies that come into play, right? Exchanges or like trading firms could have technology problems. You could have flash crashes on any independent … [inaudible] … You could have big fat fingerprints that cause, and you need, or either you want this oracle system to be highly robust to those kinds of events.
The second thin, the other thing that it does is when you have a network of these first party data providers, it's a consolidation of individuals across the entire landscape, right? And so depending on market structure of any given asset class, like the U.S. equities, you have Reg NMS that ensures that the 13 or so U.S. equity venues trade roughly in line. That's not the case for crypto, FX, Treasuries lots of other international equity markets. And that leads to fractured pools of liquidity that are often separated by geopolitical boundaries, right? And so pricing information that's being submitted by new participants as they're coming onto the bid network adds to a representation of like a global price for a lot of these sources. And so not only does it give you a good representation of the price, but also the distribution of those prices, which is potentially very interesting data. And as we see like KGI and GMO and FDX, and like all these participants across Hong Kong, Singapore, Japan, London, a lot of U.S. venues coming onto the network, it significantly strengthens that. So that's kind of like, you know, a big piece of what makes the data vetting very high quality and, you know, a much better source of pricing for application developers to use.
Tracy:
So I'm curious, just going back to the beginning of this conversation where we were talking about how Solana actually works. So if most of this is about expanding the network, upgrading hardware versus actual improvements on the computer science that underpins it, like, what is the next big thing for Solana? What are you working on right now? And like, where do you see it going?
Anatoly:
So the innovation and the computer science part we just got lucky with it. That's usually how it happens if it works. And that's been, you know, a blessing for us because we've been really focused on the hard work of like, realistically it takes 10 years to build a new operating system, a new database, the stuff doesn't happen overnight. And that's because you're trying to take this abstract idea and form it into the real world. CPUs, GPUs, you know, network cards that have, you know, their own behaviors. And to do that in a efficient way, it just takes a lot of, kind of, you know, blood, sweat, and tears. So a lot of the development that engineers do at kind of the core protocol level is optimizations that you would see folks doing if they're working in the Linux Kernel or, you know, a database like MySQL or something like that, you know, looking at memory, looking at throughput, block contention, and trying to see where do we have bottlenecks that if a validator that isn't working on software, if they add more hardware, they add more network cards, they should see that improvement, you know, from that investment. So that that's really kind of like our goal here is to make this thing elastic with the hardware that's given to it.
Joe:
I wanna like, you know, zoom out, you know, one of the things in the whole crypto space is each generation accuses the next generation of being insufficiently decentralized by some metric, right? So Bitcoiners look at Ethereum and they say, oh, Ethereum has Vitalik. And we, it was like, okay, no one knows who Satoshi was, he has no influence anymore though people try to still like understand his writings, but then Ethereum, they say, oh, you have Vitalik. He's the leader of it. And there's consensus. And that foundation, and also an individual can't run a full node. Like we can on a Raspberry PI. So it's insufficiently decentralized. And then the Ethereum's look at Solana and there's like, well, you have Anatoly who's a CEO. We don't even have a CEO, but there's a CEO and you need a big data center. And even maybe it's a little bit more tough to run a node on Ethereum, at leas it could still be done. Someone can't run a full Solana node on their home computer, you're insufficiently decentralized. And so I'm curious, like how concerned should one be or let's put it this way, why shouldn't one be concerned about the lack of decentralization on Solana that here's a company, you're the CEO of Solana Labs — I get that different than the protocol, but obviously you have a lot of influence — and there's no way I could run a full node at home. So why shouldn't I be worried about Solana decentralization?
Anatoly:
You could run a full node at your local data center, right? So I've always felt that it's not about like lowering the barrier to entry. It's about making the thing on the other side so valuable that you're willing to kind of crawl through broken glass to get there. So in this case, having access to financial data, access to the network where you can trade and participate in this next generation of finance at the same speed as jump trading. Like for me as an engineer, when I was working at Qualcomm, I would have drove to like, you know, a Hurricane Electric and set up my node like that day, like as soon as I would have learned about it, right? That's not a big deal for somebody that wants to do it.
In terms of decentralization that question specifically, the way we look at it is we try to address all the quantifiable variables that we can, and the specific one that we care about is this maximizing the minimum set of independent parties that can get to 33% of the stake weight and the network. And that's a very specific thing. Balaji [Srinivasan] talked about it, calling it the Nakamoto coefficient, you kind of look at the network and you try to find what is the smallest set of parties or participants however, you slice this network, that if they all colluded at the same time, they could shut it down. So they can never steal funds because the layer one doesn't take custody of your funds, your cryptographic keys that you own. Those things actually help hold custody. But if you're talking about trading, right, or even payments like a big payments company, right starts using Solana for payments, what they care about is that the service never gets interrupted, right?
It stays fair and transparent and censorship resistant. It's far more important in trading, but also in any financial use case interruptions cost money, right? So maximizing that minimum set is a quantifiable measurement of decentralization. You can start slicing it by data centers, by geographic locations, worldwide distribution, routers, BGP routing routes, right? Like everywhere you look at it, how do we make sure that that a problem is as hard for an attacker to pull off as possible? And in that sense, it's going back to this like nuclear strike analogy. Byzantine fault tolerant nuclear strike detector. How do we make sure that it's as hard as possible for an attacker to disrupt service? S that form of decentralization I think is, is like measurable. And if we succeed there, then we can deliver value to, you know, humans. And that's the real form of decentralization. How many humans actually care about this thing being alive?
Joe :
I just have one more question for you, Anatoly, you know, one thing that we haven't mentioned, Sam Bankman-Fried was an early investor, right? And we've had him on the podcast twice in 2021. It's like the year of SBF, like FTX, his exchange, has done incredibly well. You know, Alameda, his trading hedge fund, I think they've done phenomenally well. Can you talk a little bit about his role and his contribution to Solana and how helpful that has been this sort of like the dovetailing with FTX? I think everything that gets, you know, every new launch on Solana seems to get traded there very quickly. There was a token called Mango that just launched, it's already trading there. Talk about the sort of the synergies and the significance of that.
Anatoly:
Yeah. So FTX and our connection with Sam really started about a year ago. Not like super early in the life of, of Solana simply because, you know, we got connected to them even before that, but they were looking for something that worked. They didn't really care about our theoretical claims. Like when is this thing going to be live was like the first question out of Sam's mouth. And after we launched, we had this little game called Break. You know, you connect to the network, it sets up and you pay a little bit of fees, like a few cents, and then you can smash your keyboard and you see transactions fire off and get confirmed on your screen. It's incredibly dumb, but it really like showed to their engineers that like, okay, this thing is really live.
These are real smart contracts. You can go and build whatever you wanted and it's fast and cheap. And that's what kicked off their kind of internal team to go incubate Serum, build a central limit order book, which is something that they really wanted to do for years. Like really, as soon as they started trading on crypto and building FTX, there was this like, how do we do this in a decentralized way? Well, we don't want to do something cheesy, like an Ethereum layer two that like doesn't really like a full censorship resistant chain. So that was really symbiotic really from day one, from that moment, because the engineers saw how cool what we built [was] and that it actually worked.
At that time, people remember, like about a year and a half ago, FTX was not like the juggernaut that it is today right? Like it was a much smaller exchange. They were an up and coming exchange. Everyone loved Sam, but it was much, much smaller. And I, you know, we started both kind of taking off right around when Serum launched. A lot of developers started looking into Solana. The tools were really rough at that time and the work that the Serum team put in just building things and showing, okay, this is how things work. This is how the libraries work and how you get started. That had tremendous effect on onboarding new developers, just simply having another expert start generating in our code. These are examples. This is how you interact with Serum. And of course, you know, through that participation, you know, Jump and Pyth and all those guys really, I think their eyes opened to that. I think it's possible to build a next generation of finance in a decentralized way. So, you know, as much as you see Sam on Twitter, the folks at Pyth and Jump have been doing as much of the work behind the scenes, just not as loudly.
Joe:
Well Anatoly and Kanav that was a fantastic discussion. Really appreciate both of you joining us and thanks for coming on Odd Lots.
Anatoly:
Thank you.
Tracy:
Thank you so much. Cheers.
Joe:
I thought that was really interesting. You know what, actually, I think what really struck me is, I mean, there was a lot there. Anatoly’s point, you know, he said, well, it's like, anyone can go set up a Solana node at their local data center. And at first, when he said that, I was like, well, that's some sort of like weird joke, cause I'm never going to do that. But actually I guess the idea, it does kind of make sense. It's clearly like not a network that people can look run like as a hobby or like on their laptop. But his point, it's like, well, if anyone can do it equally, and we know that this is a problem in finance currently, which is this sort of like perceived inequality of who has the faster hardware or who has an antenna tower somewhere in New Jersey closer to the NYC data center or whatever. If anyone could do equally, it may not be available to everyone, but that does seem like an interesting potential solution or an interesting potential re-imagining of how finance could be made more fair.
Tracy:
Well, on a related note, I thought his point about, you know, it's not really about the technological innovation, but more about getting people acquainted with cryptography and getting people to understand. Yeah, that was very interesting. Although I gotta say like without some sort of improvement in the interface, I just, I find it hard to imagine that like millions of people are going to be doing DeFi but, you know, that could come in time.
Joe :
And of course, on the other hand, if you have these networks like Jump and Virtu and a bunch of others, then you know, they're interacting with the protocol at the sort of like API level, as opposed to like the, you know, the unicorn graphics on Uniswap. You know what’s interesting is like, hearing from hearing about this from jump perspective, I thought was super interesting too. Cause it's like, there's serious muscle in this space now. Like anyone who thinks this is going away or a fad or whatever, I think at this point is like missing a pretty big story.
Tracy:
Yeah. I mean, it seems like so many, well so much money and so many people are tied up in the industry now that it would be very, very difficult for it to go away. I agree with you.
Joe:
This idea, like for years, this idea, I always sort of just took it for granted, like the idea of, oh, blockchains have to be bad. Blockchains have to be expensive. They have to be slow. And it's interesting to think that Anatoly was just like, no, they don't. There's a different way to do it.
Tracy:
Yeah. I'm just thinking, you know, that notion that like, it's not so much about innovation in the computer science anymore, but more on the network and hardware side. I mean, I'm kind of, well, look, I'm not an engineer, so what do I know? But like, I know that technology changes sort of all the time. And so I do wonder if something could come. I mean, this to me is like a tension in the crypto space because you're trying to build a network and you want the network to be as ubiquitous as possible. But at the same time, people are trying to build better networks all the time and then make those bigger. And it just seems like, I don't know. It just seems like you're sort of getting constant change at the moment. And I don't know when we're going to settle or coalesce around one thing.
Joe:
I think that was a great question. And I do think that, like, it is a question for some of these like so-called like layer one, like smart contracting, platforms. Like the barriers to entry in them. Because like, you know, I'm sure if we talked to an Ethereum person, they would say like, oh, Ethereum has like 10X as much, you know, 15X as much money involved and you know, a thousand more developers etc. But it's not obvious to me, like the degree of moat, I guess, moat is the word I'm looking for. How do we know that a sustainable moat exists in this space? And it's not obvious to me that we know where that is or that that's been established by anyone yet.
Tracy:
Yeah, totally. I would agree with that. Okay. Should we leave it there?
Joe:
Let's leave it there.
You can follow Anatoly Yakovenko on Twitter at @aeyakovenko. Kanav Kariya is at @KariyaKanav.
