Hashing It Out
Hashing It Out

Episode 91 · 2 years ago

Hashing It Out #91-Paradigm Dan Robinson

ABOUT THIS EPISODE

Dan Robinson is a Research Partner at Paradigm.xyz, an investment fund active in the DeFi space. Reflecting his interests, our conversation covers a broad range of topics, including frontrunning and MEV (the mempool as a dark forest), Uniswap and building a defensible business on-chain, and Yield protocol a novel design for fixed rate on-chain lending using Automated Market Makers.

Links:Paradigm

Hashing It Out Podcast Social Media

Join-Slack

Donate to Hashing It Out

Hey, what's up? So Avalanche, let's talk about it. What's an avalanche? Snow comes down real fast, fierce gains momentum. But I'm not talking about the natural disaster. Or if it's not really disaster, I guess it no one's around. But anyways, avalanche. What is it? You've heard about it. Now you're gonna hear some more. It's an open source platform for launching decentralized finance applications. Right, defy, that's what you want. Developers who build on avalanche can easily create powerful, reliable, secure applications and custom block chain networks with complex rule sets, or build an existing private or public subnet. Right. I think what you should do right now is stop what you're doing, even if it's listening to this podcast. Stop, pull over, go to the gas station. If you need to go to a subway, there's a subway, like everywhere. There's always a subway. All right, all right, there's always a kroger. Just stopping a parking lot somewhere. Go to Alva lava, Alva LABS DOT ORG, to learn more. All right, stop, go to Alva labs. That's a va labs labs dot org. Now entering pecast work. Welcome to hashing it out, a podcast where we talked to the tech innovators behind blocked in infrastructure and decentralized networks. We dive into the weeds to get at why and how people build this technology the problems they face along the way. Come listen and learn from the best in the business so you can join their ranks. All Right, okay, welcome back, everybody, to hashing it out podcast, and I'm your host, to a John Marlin, with cohost J harrold. Say, everybody, J everyone's morning, and we have with us today Dan Robinson, Research Partner at Paradigm Fund. Dan, say hi everybody. Hi Everyone, it's good to be on here and we'll start this off with the standard Crypto podcast intro question. What's your background? How did you get into Crypto? Yeah, so I started out as a lawyer and I went to law school. To be honest, I can't exactly remember why, but I didn't. I didn't enjoy very much being a lawyer and had sort of discovered that I like programming a lot more. And around the same time that I was planning to rich quit my legal job, I got interested in Baco and and then etherium when that white paper first came out. So when I when I left my law job, I ended up working at a private permission blockchain company called chain, which ended up being acquired by stellar, and then, about a year and a half ago, Joint Paradigm as as a research partner, and I'm here representing my own views and not those of paradigm and nothing that I say in this podcast is going to be investing advice. Awesome. Yeah, I find the transition like how did you even you know, go from like how did you learn to code from having been a lawyer? That's not a typical path that I've seen very much. Yeah, so I've been I've been programming for a while and just I think there's something about, until about two thousand and ten, which is when I graduated from college, being going into going working in startups, are going into attack wasn't as Randy, wasn't as Vahtional, just didn't feel as much of a normal choice as it does now for me for whatever reason. And so so, like I never really thought about it as actually something that I would want to do professionally. And then when I was in law school I had a start up that I was working on and realized, Oh, like just a day to day of programming is so much fun that I sort of will hate to miss this when I'm when I'm working as a lawyer, and indeed I did. And then there's there's something about this is this is more about it the transition. But in law, when you have an idea, you have to go find someone who's had it before and if you if you can't find someone who's had that idea before, it's worthless. And it's almost exactly the opposite from Texas. There's there's some room for innovation in law, but it's really at the margins and for the most part, if it felt just like a much more conservative industry to be in, then then tech and particularly crypto. How did you come upon the hearing white paper? Like at what stage was it, because I...

...think it was. Yeah, I think I was. I think I was the first. This is a humble brag, but I think I was the first person to post it actually on hacker news. So pretty early on I just came across a randomly on like bitcoin twitter and it I thought, at least from the from the beginning, it sort of like seemed a lot more serious than a lot of these other kind of the others are like the coin clones, which are, you know, it's like like I need the corner. Around the time was kind of like sort of interested in so this would have been done after, etc. MIAMI. Yes, it would have been, like me, I think, late January, two thousand and fourteen UNC yeah, guess it was not in the community at any in any way at that point. Yeah, that's that's that's still pretty early. Yeah, that's it's really that's really early. I think. I think there was. It really started. People's really started to sort of talk about this idea of smart contrast. Also round that time into a February. That's for still. Yeah, and so it was. It was I mean it was just it was felt like it was actually trying something different from Bitcoin, which is very cool. And what about so you're, know, a research partner Paradigm Fun. Tell us a little bit about paradigm and what it means to be received partner at a fun yes, yeah, so I was lucky because I was close friends with math Wang, was one of the CO founders of paradigm, since childhood, and he and Fred essentially designed this role for with with me and mine and sort of with my help, I'm trying to figure out what would be most effective for me to do and what I most enjoyed doing. So I really get a kick out of it, although it may not be for everyone. So a lot of what I do is just sort of normal parts of the investment team function, and that includes sort of reaching out to two projects that are starting to be successful, diligence in potential new investments, evaluating them with the rest of the investment to making decisions. Another big part of the job is helping out portfolio companies, and there I think one thing parrot one way paradigm maybe is different from other funds, as we get pretty hands on helping out with things like mechanism design and so on, some of the technical issues with with overt folio. So I was a CO author on the unit swap to two white paper I just read. Just recently published a white paper with that with another portfolio company, yield, on a new AMM that they're using and and and just generally helping our portfolio. And then the third area is this kind of independent research, and that's where I think I just really appreciated having a having a place where they would welcome to doing that, and that's publishing. So I publish sort of you can call them research papers, but really it's sort of like it's like a sketches of ideas for things that people could build. And so I came up with one of them called Rainbow, about a year ago, and then another just under a year ago called yield, which eventually turned into a project that we're incubating that's implementing these these white tokens. And so that was a case where I think we weren't sure exactly how publishing these papers would help the firm, but we thought, you know, like one of the one possible outcome would be that we end up incubating a company based on it, which happened here. Yeah, I find that very interesting and did see that progression happened over the last year. It looked like you you put out the White Paper, a few people read it, someone came and think, you said, suggested improvements even and that led to, you know, you basically handing the project over them center. Correct, that's right. So so Alan Alan, who's the CEO of field now, came to me. I think he fixed one of the core problems in the protocol, which was settlement, and he hadn't he figured out a way to settle, let essentially directly, to die, more directly to a to a makeer faults, as opposed to settling using using an auction or oracle or something. So that was one of the big missing pieces. And the other was just we wanted somebody who could really own it as a as a founder CEO, and we think Allan's that owns that person owns another former lawyer for some reason. Yeah, I think, I guess I maybe I got to get along with him for that reason. Okay, makes sense. Yeah, I think it's just the power of putting your ideas out there and actually seeing it a comeback, improved and better and without providing opportunities for for you and in the fund. Makes a lot of sense. Yeah, it's extremely rewarding, although sometimes when other so there there's another project, at least one other project, implementing the yield protocol, and sometimes I think like this, why do it? Why do I do this to myself? But honestly, none of this would have happened without publishing the paper. So I can't actually massall. You regret it that much? Yeah, that's that's that's interesting. I think that's a like see you talk about, you know, putting ideas out there, like we're living in an open source world and this maybe is...

...going to get to like something I thought would we take a little bit longer to get to, but is interesting to me. So it's everything's open source. It's all forkable. You can just copy and paste it. UNE SWAP is probably the most high profile company in your portfolio and we've seen in just the past couple weeks we're seeing basically an explosion of copy and pasting in defy, with mixed results. As you mix and match Legos, some of them, I don't know, have some surprising results. But so, yeah, I'm interested in your thoughts on these, these competitors that are emerging basically taking pre existing ideas and how one one can build a motor on their business or protect against competition in this world. Yeah, I think it's I think it's a really interesting problem. It's somewhat different from so this problem is has been an issue in open source software development for for decades and in fact a lot of opens of potentially big open source businesses what kind of eaten by by something that Amazon and Microsoft was like their cloud computing systems. Where they end up. Basically, you know, they they tried to do this this model for monetization where they were that they would provide support for people, but then turns out that Amazon and Microsoft can actually provide support for this open source software better than the projects themselves, in many cases, on and infrastructure around it and running hosted posted instances. So that's that's a long way. Tony know this. I think this isn't a new problem and open source software and monitor figuring out how to monetize it. It's arguably easier and cryptout, and that's because of in some ways easier because of tokenomics, because in a lot of in a lot of cases you can you can have this sort of like concrete steak in the protocol that anyone can own and therefore kind of has a you know, you can get different kinds of stickiness from that, I think. So that's that's that's one way to do and that's just under sort of a natural modetization method right for a particular protocol. That doesn't necessarily stop someone from working their way so and you just obviously does not have a token. I think it's I think it's one of the most important questions. That's that we're going to it's gonna have to be addressed in the new future. I think with with someone like you know swap. I think there's these are these are maybe more short term motes, but I think there's a very strong motor around brand equity and if you see people kind of dunking on a lot of the copycat clones on twitter, I think people have a lot more respect for you to swap and that leads in many cases not necessarily to it's more liquidity, because a lot of liquidity is very flighting and just goes to whoever's bribing people the most. But but the higher volume into enter sort of stickier liquidity. Another I think you know. You just have those sort of fantastic work external to the protocol on the on the interface and the the and analytics sites right now and, like a lot of these projects, obviously your copy copycasing those, but I think we're going to see more infrastructure sort of built around around you a swap that's harder to harder to fork out. integrations are another area where I think you know if you have if protocols, if and centralized exchanges are and the verticals are all providing direct support for, you know, swap. It's hard for another project and yourself, for example, like your self, accuidity, pool shares as collateral and lending protocols, it becomes it becomes, you know that that's that's another barrier to scale before someone can consider do it for and the one other one in this kind of intangible is I think we've seen with most of these forks that or copycats that people aren't really happy just just copying it, copying, for example, and just like removing the fee. That because because you know sp you too as a protocol fe switch that could be turned on. And the funny thing is just like no one will. No one will do that because there's no way to make money off of it. So people always going to they're always gonna get a little greedy or they're going to get a little too creative. So there, you know, they put in something where they get to capture rewards from this, from this new protocol, and that makes it. You know, then that becomes that becomes more rent of attraction than the your shop is doing or they or they try to try to fix UN swap by adding some feature that isn't really a feature. You mentioned this idea of of sort of an easier way to monetize open source. I'm I've always kind of been the opinion that it's an interesting way to like if you're acting in an open source manner, using tokenomics, it's a it ultimately brings an incentive to find a way to make it work. It sounds like that's kind of what you're talking here, but you know, what are some of those things that you really need to focus on? I mean, obviously, me coming from a branding background, I see UN swab looking into that and really knocking out of the park. So I'm sure...

...that it's not. It can't just be token on its or is that a huge part of it as I think. I think the memes are probably the most important thing, and what we're crypto is kind of change things is that there's I think there's a lot of relatively strong memes in even in like open source software. I don't know, people like a lot of people really like reactor, whatever, have strong opinions about language wars, but tokens have effectively a monetize these memes. Right. So in a lot of cases, and this is we've seen, obviously, just like a weaponization of this recently, where people are just coming up with every kind of food they can't in order to grab this. I think we're going to see we're going to see diminishing returns from that. But a really strong meme is sort of indistinguishable from a brand, and I think you know swap has a fantastic fana's a brand and a fantastic meme. I mean the story. Had founded it with less than us a hundred thousand dollars in grants before launch, most of which went to a smart contract audit. It was his first programming project ever. He launched it with no governance, no no fee, and now ended up this becoming the the largest deck by volume by far on etherium. I think it's a it's a really good story. It's very hard to replicate that kind of story. It's similar to projects that I think a few years ago people thought you could just fork bitcoin by, you know, I'm having slightly faster block times or having or having sightly larger blocks, and it turns out you, you, there's basically no way to fork that kind of founded with yeah, I would agree on that. Yeah, it's very it it's very hard to capture that kind of leg ahead. In itself is is a bit of a mote, but it's still open stores. Yeah, very interesting. So it's get right to it. Damn, we were just before the show. We were talking a little bit. I can't think of the segue other way than to say, hey, tell us, tell us about what if you want. You, you, you were, you were, you know, whatever was up to last night? Yeah, yeah, so, yeah, so, I'm a little tired now because I had I had, you know, sort of one of my first kind of late night mad etherium hack nights. And unfortunately this wasn't it wasn't an it wasn't an exploit and it wasn't it wasn't really a vulnerability any project. But what had happened was, and I'm planning, I'm planning to publish a block post on this at some point, once I get my thoughts straight on it, a someone had locked up money in a contract by accident in a way that seemed unrecoverable, unrecoverable, and I heard about this in a discord and, you know, I was I was just thinking about it for a while and then, like late last night, I realized that there was actually a way to recover it, but the by just calling a function on this contract. The problem is anyone could recover it. So it wasn't just the owner of the funds. I could do it, which means I could steal it. I could white hat recover it and and return it to the to the owner, before someone else did. But anyone else could as well. And this is an issue because etherium is is basically a dark forest. The etherium men pool is. This is a war zone and if you do anything, if you submit a transaction to the men pool, and I knew this, luckily, from having heard this from they'll die. On who's experience, as himself as well and as is one of the experts on it, if you if you try to submit a transaction, anyone can see it and they can inspect it, see what it does and if it would be profitable for them to front run it and and you change it in some way, they do that. So by submitting this transit as Whitehat transaction, I would effectively be putting up a big flashing sign saying here's some Mev for somebody to capture. So I asked the Security Community, I talked I talked about to it, to some researchers who focus on this and I tried to implement an obfuscation to and this is involved a couple contracts, involved multiple transactions and obfuscation that are basically prevent these, these generalized front runners from seeing this transaction and and to me because, again, like this doesn't require any human action. Nobody's actually looking at my transactions. It's just a program is just checking whether it could replace it. And it's really devilish because they it's not just the transaction itself. Because, you know, I had like a contracts with authentication in the way it checks the entire internal call stack of the of the contract. Now, I'd actually never seen this happen and I I was certain know, I knew intellectually that a fossil, which is why we went to all this trouble to try to get around. This topic covers my first person, but I had never really seen it. And but when we actually, when we deployed this route golden machine to try to capture it, sure enough the gout front run as soon as as soon as it was submitted. And so it was. It was a very visceral, you...

...know, like that moment at the beginning of a quiet place where the monster like eats the kid because he because he makes a sound. That sort of like sets up the stakes for the rest of the movie, because otherwise you'd be like these people are being silly, right, you know, you know intellectually that there's danger, but you wouldn't actually really feel it. And so it was that it was that kind of moment from you. I thought this, yeah, like like this is actually is a real dangerous environment. And you know, I think actually some of our portfolio companies are are working hard on on ways to try to mitigate that. But I think it's a very fundamental issue. Mev and sort of this these couple wars for far theorem, I'm their insecurity. So it's I mean to that the analogy where it's like you knew that this was an adversarial environment and you've heard about it and you get here the the like, out of the chaos and violence outside, but until you like stuck out your hand and had it shot off, it didn't wasn't really didn't really feel real to you. That's right. Yeah, and and so I do. I also want to kind of so unpacked that a little bit so it's a first of all, just maybe for listeners who aren't aren't familiar MeV stands for minor extractable extracted value, and so the idea is there that by being a minor and having the ability to determine transaction ordering, you basically get, I I'd suppose, the first rate of refusal to make a transaction. And so so what's happening is a minor can simply, like, if they see a an arbitrage play or something that somebody submits a trade that does a flashlan trades, you know, do you know? Swap and like arbitrage has to balance or some other, some other amm the minor can simulate this, see what would happen if they sent it from their own address and if it if profit, then then go for it. So what you were trying to do, I think you were trying to hide this, basically make it so that you know, you probably predeployed a contract, you like preloaded it with a commitment to the transaction we're going to submit, and and so just try to make it a few steps and too complex for for a minor to pull off or even understand the outcome of what you were submitting such true, you're muted. That's right, and I think specifically in this case, I think in most cases right now, Mev isn't actually directly extracted by miners, although that seems like a frontier that's potentially opening up. It's mostly these, mostly pots, and the money often ends up in the hands. So some of the money ends up in the hands of minors, because I've high gas prices in these gas auctions that that happened between these spots. But but that's that's I think that's that's right, and there's quite a lot of money, I think, to be made from from me by these bots. And like I this this transaction, this production, was large enough for me to stay up till three am trying to this amount of money, trying to trying to find it for somebody was small potatoes in that and the MEV game in general, and mostly so, I think you know, it would seem just like really hyper competitive mentor races. And I think ultimately, yes, we'll probably see minors extracting some of this value directly, and right now you know this. This MeV is definitely not benign. What's what happened, for example, yesterday, although again they will only be able to exploit it because of because of basically this mistake that had happened by it. There are kinds of MeV that are that are really catastrophics or something. A double spend could be an example. Is basically an example of MeV. When you think about it, and and sensory transactions. You could, if you center transactions, you can manipulate all kinds of protocols on chain, and so that becomes a really not just not just kind of a an inconvenience, but really a threat potentially to have this ability of a theorium itself. Yeah, and so key like it's maybe hard to get into over audio, but I act like I want to understand. Like they did. They actually like so that you did get front run right and they got the value. And so did they like trace back, like they basically redeployed the contract that you had had set up and everything, or so they didn't have to. So my on my office station was basically to hide the fact that at the bottom of this call stack it does this particular call to a contract and that called to a contract is just to give me the free money call. And so what they could do is just is just run my transaction and and just and just and basically like...

...look at every internal call that happens and see if if I did this call and replaced the two are the address that's in this argument with my address, what would what? I make money. And apparently that's what happened. Okay, so it's a fairly basic just like swap out. So even though the transaction you submit was fairly complex, the end result was was profit. And so this this front runner, went for it right at the end result of this, of this one particular call. If if it tried out this call, I see. And so we only did the call. Like it's still sort of like or did do exactly what you had planned. It didn't do it, you know. Yeah, so it couldn't do exactly what I do because I had this multip right projecting. It just skipped that. It did some other stuff too. It was like it was kind of wild. I did not expect it to actually be this specificit hip, to be honest, and yeah, friends of gas token, it'd like was definitely, know, definitely pros running this anything. From from what I've heard from others, like this particular front running about hasn't seen around in the wild before. But you know, we we only know about this software from it's like a black hole, and we know about it because of its effects. Nobody, nobody know. Researchers like you know. I don't think fell as ever, as ever really seen, for example, like the the the actual software these guys are running. You just have to infer it from the behavior of what happens on chain. Yeah, that's so fascinating. We've seen. We have, like we did. I can sense it's diligence. Did a bit of research into like if we could kind of like honey pot by like, you know, deploying what looks like an obviously vulnerable contract, you know, so you eat, there's like one eth in it. You have to submit one eath to extract it, and then we would try to front run you on it. I think what like it ended up just that it didn't quite work out in the end, and trying to remember exactly why, I think it was that the front runners are yeah, just well, actually, I think I described the mechanism slightly wrong, but the thing was that we thought that, like, we would make it look sneaky at like a human might read the contract and Think, Oh, I can get money out of this. But the Bo it's not even being careful, reading carefully, it's just simulating it and realizing quickly all this is enough. There's nothing here that's right. That's right and and a lot of ways, I think the a eyes that are doing this are too stupid to fall for a lot of the tricks. And that's, yeah, I know. I think we discarded a bunch of ideas that we're sort of like obfuscate this, and there's ways by making it look like we were withdrawing different tokens or something, just because if the dumb strategy of look at every call and see if I made that call, what I what I make money, which just it just, it just cuts through all of them. And this is this, it's a really difficult problem because it's, you know, like this, this fundamentally, yeah, and like for this transaction to be executed on chain, it has to be public and in theory someone could just construct the block that it's going to be included in, see that it would make this money, evaluate this and and find that call in the middle of it. And there's a call that's like the call itself is not permission at all. So there's really sort of like essentially no way to avoid that. So the only right now the mitigation is to try to make it as computationally more difficult, which is what we were trying to do for that for the attacker to to find that, yeah, fascinating. On keeping on the topic of Mev for a moment, this is something I haven't seen talked about a lot, but but so optimism, the builders of optimistic one of the they're using optimistic roll ups to put an Evm in your evm, which I think is kind of mind blowing. And so, you know, on the topic of scaling and through layer two solutions, I have this sort of doubt or concern that, like, the more top heavy we make etherium, the just the more incentive we create to to execute a fifty one percent attack. And it seems to become to me that like if you can, that you with enough financial engineering, enough large hint transactions on enough player two's, an enough different tokens, even you can think of a token, is a simple layer two almost in some sense right like when when the when the transaction value of any given block is starting to eclipse the the security budget of eith issuance this starts to feel like to me like a real risk. Is Is Mev somehow like actually possibly a solution to this? I'll just there's a lot there. I'll just I'll just leave it open to you. I agree with that. So you agree with one in particular? Just this idea of...

...too many layer two's on top of themselves potentially leaves the stack of vulnerable on the underlying chain. Yeah, so, yeah, so I think. I think this is a concern long term for the sustainability of generalized smart contract platforms. Is that. Yeah, people can basically build so much stuff that becomes top heavy and only, you know, the bill gets. A definition of a of a good platform is something where the value created is greater than the value captured and that's in some ways sauce by a definition of what could be an insecurity situation. So I think, did you know? I think is a fundamental problem and I think there's some solution. So one that wouldn't involves changing the general chain architecture is something like Cosmos, where you have these app specific chains, where each chain only sort of permits the the behavior associate with that particular APP and then can extract sort of a reasonable amount of the value from it that is needed to secure this taking token for that particular chain. And so that's that's a little easier, it's a little more tractable of a problem then the trying to capture enough of the value on a turn complete chain. And so so that's that's that's one approach, although you do lose, you lose, I think, a lot of the benefits of having having a turn complete smart contract chain. The some other approaches that I think are promising that layer to include the one proposed by optimism, which is a portfolio company of ours, and optimisms approach is to basically have a sequencer that's responsible for for ordering transactions, and this, you know, this would have avoided the problem last night if there was an honest sequencer that we could have submitted this transaction through that that controlled access to this this contract. Basically, then our transaction be confirmed before anyone else would would rather secrets before anyone else would see it. So that's so, that's that sort of one approach. And then that there's potential for that sequencer to basically fund itself and so being centralized by extracting benign MeV potentially. And then, but I know maybe include stuff like being the first trade after a union to swap after the after the true market prices changed. Someone has to do it. It's a good service for it, and if that could be directly captured by a minors pins that as a security. So I could see what you were saying. Is I think in some ways Mev is is a solution to the problem through to its own problem, which is that if Mev is quite valuable, then continuing to extract it, to be able to extract, especially if you can attract it, you know, in a benign way, could basically subsidize the the security of the entire chain. You mentioned ordering. Do you what are some of your opinions with like? Obviously there's not enough information known out there, but if we look at sort of the two in terms of them pool, do you think, with beacon chain being essentially a metronome for finality and ordering, do you think that could could help that? Or is that it was? It's still the same same problem is is more or less there. So I think proof of stake certainly helps with the limited problem of fifty one percent reversion. I think that, you know, I just just this slashing improve stake. I think helps enormously with that, which is that after some point you can sort of be guaranteed finality with a very high degree of economic security. I think the these other problems with transactions, emission. I mean I think beef to architecture isn't, really isn't necessarily final enough for to be able to evaluate that. Personally, I think that layer choose, like optimism, are going to play a really big role in youth to and what these two goes very close together very nicely with optimistic as UK roll up, which of these these systems that depend on the data being available but not necessarily executed by everyone, because what easy to phase one really excels at is ensuring data availability. You can have to get massive throughput and very on data availability. And then if you handle execution at the at the second layer essentially, then then potentially that that improves some of the scalability issues. And then, yes, I think, execution the second layer and then potentially these MeV for yeah, preventing prevented militia's extraction of MeV could hopefully be done at the second layer. So that could then involves having having sequencers like optimism plans to have where you have some some extra process, either either a semi trusted party or a or some kind of external can other consensus process for ordering transactions, and the worst they can do is kind of is at some point exits can by changing the order of some transactions, but basically you have to have some kind of exit valve, escape valve for that's all right, let's see. Let's let's talk about quickly about some some of the the other projects you're involved in. I'm particularly curious about...

Hyde or yield. I think it is in general it sounds like a very interesting project, but definitely something that takes a little while to the wrapper wrap ahead of heads around. So let's hear about that. Yeah, so to some degree I think we're coming at financial and and be if I were coming at the financial innovation somewhat backward, in that this innovation of the perpetual swap and crypto trading, which you know, started on centralized exchanges like Bo Max and I think it's starting to see more decentralized things with that, is a really novel, interesting approach to get to the leverage and lending and it's perpetual. It does it at this at a profetual because it has this floating rate, floating interest rate, and it's extremely powerful thing. And what we've seen with most like defy lend borrowing and lending system so far and synthetics, as that we synthetic assets that is like like die, is that. Yeah, they all have floating interest rates and the interest rate is basically set either by governance or by a formula in order to or, in reflexes case, by by an algorithm. And I can let's talk about that reflects or another portfolio company of ours. But the what what's actually most common in traditional finance and landing is is fixed rate borrowing and lending. It's by far the majority of the of the market and we haven't really seen that in defy and they've been some approaches to it. I thought that the approach that had been tried gave up too much in terms of, say, fungibility, like loans with different interest rates weren't fungibles. It was basically you had to find a have a peer to peer market for for we had to find someone willing to borrow for the exact same term at the exact rate that you wanted. And there's no easy way to get out of the position early. So the you know. So I think that was the that was basically that was the impetus for designing why Tokens, why dye, which acts more like how how fixed rate markets work and tradition finds, although I think it's to be honest, I think it simplifies himself about that as well. And why dye does this basically about being a similar to a analogous to a zero coupon bond. So the price of y dye floats until expiration of matures. They when API charity, it's worth one die and up until then it floats. It trades at a floating price and the market just tells you basically what this prices and the price implies a particular interest rate. The price in the time to maturity implying interest rate, and so that's a way to easily get market determined interest rate if I fixed rate interest rates for any particular time period. And the Nice thing about this is that lending y dye is essentially just just buying it and holding it. You just buy hy Dye with die and you just hold it and it appreciates over time because if you buy it at the discount and borrowing is just it's similar to to and maker. You just meant Wy die and then you sell it so you don't have to find and this means that all these solutions that we've come up with for improving the quidity between ear see twenty tokens, so like you know, stop another, amims are example. They all can be leveraged for for for improving the quidity in borrowing and lending, because it is just really your see, your see twenty trading. And but in fact we did actually come out with a paper that that improves potentially on this amm who's on his name and designs for this very limited use case of being the wide eyed eye or why token target, token trading pair. So you're just you're really just using the MM as an oracle to get a price to gage demand, for to borrow assets, and using that to set the interest rate. Well, so, so, I mean the the price, is the interest right or really like it determines? It's like we're not right. We're not looking at the price and saying like Ah, that's what the interest rate should be. It's a literally the price of which you can buy. This tells you what interest rate you can you can borrow a line it at right. So I will pay s today to get a dollar in a year, and that's guaranteed. Yep, and the AMM IT'S A it's a fun paper, other one of the more esoteric ones that I've that I've read in because it it shows how you can derive these invariants, like you know, swap or like or like or like this one with a by solving a differential equation. Right, I actually think that there's some, a lot behind what you just said, right, which is fairly matthy and technical. So you just said like you can solve. Can You repeat that? And Yeah, you can. Let the implications are so you can. You can find an invariant like like you know as well, by the satisfies particular property, by expressing that property as a differential equation and then solve it, solving it. So...

...a differential equation is an equation that relates a function to its own derivatives. And the Nice thing about any constant function market maker, like you know, swapper or balancer curve, is that the slope at any point. So the derivative of this curve at any point is equal to the negation of the price, because it's you know, it's this. It's this. It's hard. It's hard to describe visually here, but at any given point the amount that you send in to the amount that a ratio of the Amalius End, the amount you get out is the price. And so for you to swap the the property that it satisfies that it always the fifty allocation at the current price, at the current price that it offers. And if you express it as differential equation and then solve it, it tells you x times y goals. K is the is the relationship between the reserves that satisfies that particular property, for for it. And something similar happens when you try to derive the the one that we use, the yield space and M and that we describe in the paper. And that's that makes that satisfies the property. Got The ratio of the of why? Over, x is the interest rate that is off for the current time. So it's an it's that's why it's called yield space. It's an automated market maker in yield space and interest rate space rather than in quote, price space. Okay, yeah, that is a complex idea, I you know, for listeners, if that didn't make unrepresent sense. Don't feel bad about it, but it does. It does sound like it is pretty fascinating. What you're basically saying is, you know, if you have a relationship that you're looking to achieve, you can. It provides you a sort of a mathematical way to say, okay, well, this is the relationship by I'm looking for between these these assets. Here's what the I don't know if ideal, but like with the equation that I need to put into my for my pricing function in the market maker. That's right, and I think you know this one turns out to be a pretty simple one. Because why? Why? Tokens are not that complicated. I think the holy grail in this potentially would be a made a market maker. That could happen marketing options. And so basically it you if you could imagine, like expressing like the black roles equation for the actual's formula for for valuing options and express it essentially as this differential equation and then solve it, you could have an automated market maker that takes in time to mature or timed expiration and like underlying asset price of an option and prices it along a curve that the tries to sort of minimize the amount of lass that currently amms have as a result of time decay for their for the four options that they trade. Yes, because right now, right now training options on, like you know, swap, it is pretty difficult. You've end up losing, losing a decent amount to sort our treasures just because essentially have all little options are and that's potentially predictable. But I think that's that's much more difficult math than I'm right now able to do. So not that I haven't tried to look at it. So I'm curiously what your what your vision is like. You are pretty deep into the defy space and part of your job is to, you know, come up with ideas with the future and where this is headed. What do you if you had to, you know, not make concrete predictions as sarily, but think about how we're evolving, how the various primitives will combine and continue to morph and breed, if you will. What sort of things do you think we should look forward to? So it's a lazy answer, but I do think there can be there's a lot to learn from traditional finance and whitetop and take a lot of inspiration from how fixed rate markets work in the the real world. Obviously, I think a lot of people have been trying to do option projects opens an example, and I think so. I think there's there's a decent amount where I've just sort of catching up to do there. But then I think there's also a kind of extraordinary amount of of innovation that can happen at the edges of these kinds of things. So I think that's to my own aren't too much. But I think, like the the yield space, AMM is not something that we took from traditional finance, just kind of pops out of how these of how these particular instruments work, but it's something where, you know, anyone can can come up with the mechanism like this, which is pretty cool. Like I I was, I was a lawyer, but at no point was I in a position to like to kind of like invent new new new mechanisms, new new products and this particular way.

And so I think I think necessarily, you know, we're not gonna it's impossible to predict some of the some of the cool stuff we're going to see because the reason it will happen, the reason there and the reason they will be a uniquely enabled by tripped out is that it'll be so creative that no one could have come up with it before, including me. Are you worried that some of these token mechanisms are are not necessarily built for regular people? You know, they're they're taking from finance and they're they're great products and obviously there's a chance for people to learn and make their own mechanism design, but for the the vast majority of people it's, you know, it's it's going to it's just going to be like, oh, that's another die for one which just used die. Yeah, so I think the the answer is basically yes, and I think there are a limited number of of applications that have gone what we would consider encrypted out to be, to be more quote, mains stream and Bitcoin, bitcoin any themselves. I've I think are I think, you know, shops in that category. I think die is in that category. I think a lot of the innovation is going to be under the hood and really in Cryptoi. Know we're still quite early in sort of building out this infrastructure. I think a lot of these primitives are not going to be they're not going to be user friendly on day one or essentially, ever, and the the there to basically serve as tools for the rest of the ecosystem to be able to deserve take advantage of them. So, like you know, the traditional financial systems enormously complex and but like my interaction with it by having a mortgage is is somewhat complex. Gott understand my mortgage. But actually, under the hood on, this is like is a very complicated financial product that allows like a something like a thirty year fixed mortgage, because there's all kinds of weird optionality and everything embedded in that. That is that is just abstracted away. And this is only possible because there were these degenerate sort of a option markets and people have put a ton of work into figuring out a price all these strange things. And so, yes, I think you know, there's there's the hope is that eventually have a better financial system from the ground up, but I'm I don't have too much hope that it will be much more understandable than that than the current one. Yeah, I think that that's it's I've kind of walked that journey myself recently of frustration with defy and how hard understand it is, and then realizing, oh wait, I absolutely do not understand finance whatsoever. And and almost the reason that it's like because there's like an uncanny valley or something where you can get close enough to understanding defy or for me at least, it's like there's this achievable possibility that I could understand what's going on with these things, and so I find it terribly frustrating that it requires so much effort, whereas you just give up and let go with the finance system right now. Do you think that there's possibilities like, perhaps because of the audibility, auditability that on chain finance provides? Are you aware of, or of you thought about, technologies to make it more approachable and, you know, perhaps like visualization tools to enable normal human beings to understand what they're interacting with? Yeah, absolutely so. I think I think you know it, solidity has come a long way on this. I sort of like automated tools for from that, from sort of a Pro Iman perspective, it would be I do think it would be kind of fantastic to have tools that could sort of help help make these things more like automated visualizations of it. Yeah, exactly. Smart contract systems and especially from a financial perspective. I think. I think we're a long way from that. It's the kind of thing, though, that could be built on this because it's it's amenable to Annalysis, because it's because it's written in Code. And that, I think, is that's one of the most promising things, is just that we can end up with this, with a set of developer tools that like people would, you know, dream to have in in traditional financiering, traditional law. You know, I think like it would be really fascinating. I would I would love if I was like working on I was not, I was not a contract lawyer, but if I was like working on a contract as a lawyer, I can actually know paper, contract, signed blood or whatever, to be able to fuzz it right, to be able to say like, Oh, what's you know what would happen in they're all these situations and just try out a bunch of random scenarios to it right. And you know, in financial and finance they do have they do this. They have models and simulations that can show what would happen in various situations because they've modeled the problem space. But in Crypto you could actually you can. You don't have to you doesn't have to be a model. You can use the actual contracts themselves as the model that you that you can runs up against, and potentially that could yeah, but potentially that could reduce the kind...

...of errors that cause to the that you as a financial crisis. I think that's very speculatve, to be honest. I think that kind of financial crisis are very difficult to avoid and in any system, and I think actually the to be honest, I'm pretty bullish on how the Fed has has mostly acted for actually sensitive since the crisis, to be honest. But I think, you know, there's there's the potential, that has potential to sort of design the system in a way where it's more robust of these things. That would be the that would be the happy case. I think you're a good point here. You know, yes, it's pretty DJ in in the back background and you don't don't really know what's going on. But you know, two thousand and eight really did prove that. Well, we really don't know what's going on and it was really unclear what was how all of this was connected. I think bringing visibility, even partial visibility, like let's say one percent of everything gets is slightly visual. You know, there's visual visibility, they're a little bit of analytics and suddenly have a lot better decisionmaking power real time. Or even if it's not not, you know, not the best, it's still far better than what was in existence just over ten years ago. That's definitely the how I think you but there's this cost of transparency as well. Like this, there's than yesterday. That happened to me right is is basically an example of this where radical transparency can be very bad for you. Any smart contract exploit very often could. It would have, it would have been prevented if, yeah, if you couldn't, if you if, certainly, if you could't read the code of the contract and if you couldn't run automated tests against it. You know and and that that's just a that's sort of a tradeoff there were making. One Nice thing about that trade off is that it's sort of forcing us to really harden the infrastructure now, while everything still a little bit silly. I think the just the adversarial environment is so harsh that hopefully what emerges out of it are these are these relatively robust and susons. I think, for example, like governance minimization. I think a lot of people had kind of this instinct, or I'll go including you, in a swamp. I think we've really seen it play out. That, like governance, is a liability that should be minimized as much as possible or localized to a particular you know, to solve the problem, and that, like general purpose governance, it that's the increases the complexity of what you're of what you're dealing with. Awesome. I think that's a pretty good place to trap things up. It's been a fantastic conversation, Dan is there. Is there anything else you'd like to share with the listeners or just let them know where they can find you and follow you? Yeah, the best place to find me is is twitter. So a lot of these snarky takes I try on twitter to see before I bring them on to podcast. So I'm at Dan Robinson on twitter. Thanks. Thank you so much. Thank you so much, great conversation. Thanks having me.

In-Stream Audio Search

NEW

Search across all episodes within this podcast

Episodes (118)