Fireside Chat I Tarun Chitra, Robert Leshner and Kartik Talwar | Pragma New York 2023

foreign [Music] let's welcome tarun titra from Gangland and Robert leschner from superstate on stage we're going to cover a lot of things happening in defy Game Theory 50 other things investing everything that's going on so welcome hello yo don't fall asleep wake up it would be fun uh there's a lot of coffee on demand downstairs so I want to start with with you Robert um you are no longer working on compound Finance day to day what made you start the new Venture and can you tell us more about it well I think like a fish or an underwater organism if you are not moving forward you sink to the bottom and die and sometimes you got to switch things up a bit and for me compound is at a point where a lot of people can work on it without me being there and it continues to succeed uh Turin is one of those people there's a lot of people that are able to work on it in a dow decentralized format and it's working so I can leave I probably could not have left multiple years ago it probably would have died um and so I see bringing assets on chain is the biggest possible meatiest hairiest opportunity I can work on um simply because there's about a trillion dollars of crypto native assets and there's about 900 trillion dollars of assets that don't live on a blockchain yet and so by bringing that massive amount of assets that don't live on a blockchain yet onto a blockchain it's going to create a lot more opportunity for protocols like compound four other D5 applications when they're way bigger and way more assets they can interface with so is that a we have a new technology argument or the fact this is now just the opportunity and it's ready I think it's just the opportunity I think the world's ready for it so does that mean the institutions are coming the institutions are not coming yet I think the institutions are not going to really come until um the assets that they already hold are Unchained I don't think they're really that excited in my personal opinion to buy new tokens that were created recently so well so I guess so it helped me then simplify this contradiction so what is the the 900 billion or trillion asset that you you said we can come over and and that not overlapping what they want yeah well no so those are the assets they want is 900 trillion I can see it's all the stocks bonds currencies Commodities real estate every asset that's recorded in Old School inefficient way so it's recorded with spreadsheets it's recorded with legal contract it's recorded with lots of humans and lots of Courts to figure things out when they go wrong it's almost all the wealth in the entire world is in a really bad file format I mean that's a good premise for an opportunity this is this is all together enterprise software in web 2. um terennials I'm a big overlap with uh at least historically how compound has grown uh tell us about what Gauntlet is and sort of what made you start it and all together like what mechanism mechanism design falls into yeah I think uh you know each cycle of uh innovation in crypto always has an increasing number of parameters so if you think about Bitcoin when it started there's only really one parameter which was insulation schedule and once you pick kind of the end point uh and a time scale you've basically parametrized the entire Bitcoin Network more or less you know and then you you get to ethereum and then now you have to parameterize transaction fees you get to proof of stake now you have to figure out how to deal with how Mev gets redisruted each sort of cycle has more parameters more parameters means more efficiency but it also means more risk and more stuff to tune and from you know what we do is really look at all the different parameters within the ecosystem and try to optimize them uh dynamically you know I think over time as things get better hopefully these systems can optimize of themselves but you know it's a you know you first you have to figure out how to do the research do that and then you know run the experiments and then adjust things accordingly um and yeah I think the the interesting thing is that the mechanism design in the space has only gotten more complicated um you know yesterday I gave a talk at the ZK Summit which was you know I I can't think I were just talking about this he was like Hey were there like 20 people there I was like no they're like 500 there's way more people in ZK land than you think and a lot of it comes from the fact that it's become as easy to write ZK circuits as smart contracts were in 2017 and that's you know kind of a an interesting that's a great sign and then also increase the surface area because now you have to look at five other things that you didn't have to when you were just doing solidity um out of curiosity is there some like really prominent example that you can talk about on how it got the simulation or parametrization has helped identify a risk yeah for sure I think so there's two types of very broadly speaking I think you can think of like risks and defiant and decentralized systems is having two sort of categories one is sort of worst case risks worst case risks are you know ruined like the net proof of stake Network slashes everyone and goes to zero and now can't validate itself right like that's a really bad condition or a D5 protocol has cascading liquidations and goes to Sarah then there's average case bad behavior which is sort of like hey I'm an LP in a uniswap pool and I'm like slowly losing all my money even though I think I'm making money because I'm getting wrote certain types of Rewards and um you know something that you have to monitor is you know say you know uh a centralized exchange happens to suddenly shut down without any warning well there's going to be a ton of assets that are on that exchange that suddenly the D5 versions of those assets are going to have to be be at risk and you have to have emergency governance proposals to kind of figure out you know is this asset safe to borrow against how much risk is there with the collateral that's being used how much risk is there with the outstanding borrowed so that's the type of stuff we focus on but the average case stuff is like how do you choose incentives for your protocol how do you how do you bootstrap things or how do you you know adjust interest rates and protocols and and so there's a huge huge kind of gamut between them and it's only kind of getting more complicated you know Mev Mev is kind of like second law of Thermodynamics it's like only increase complexity every day right it's not like it's decreased it's never getting simpler out of curiosity What's um what's the new complexity on Mev with the introduction of a lot of l2s here yeah or like how's that evolved since just M1 or L1 yeah so I think the cross chain Mev world is actually even though you know people have talked about there's probably been more you know chatter about more ads sold on Twitter to people who are talking about cross chain Mev than there has been actual cross-chann Mev but part of the reason for that is that people aren't really doing things like cross-collateralizing across different l2s instead people are just like Hey we're gonna have the same token in three different places and they'll eventually get synchronized um but I think what we're starting to see in terms of l2mev is um you know a lot of the kind of data storage piece like the different l2s are actually competing in Mev auction so like to get in particular transactions on a particular M2 L2 you're starting to see the sequencers actually having to be competitive within the mempool you know six months ago the optimism and arbitrum sequencers were not competing to send fees or to to kind of finish reconciliation so the market as there are more Roll-Ups has actually bled into the the mempool and that that in and of itself creates a new game super interesting um you've both been in in D5 as Founders users also investors uh Robert you kind of just you hinted on what you're excited about here but overall kind of how do you think the D5 space is evolving and what do you kind of see the next few years fly well the biggest areas in defyer not in like the very Basics or like Financial things you can create I think you know for the last five years almost like the things you can do are basically the same it's like you can borrow you can trade you can do perps um that stuff is not the exciting thing what's blowing up in terms of like the surface area is how information moves with these basic things so there's you know so many more l1s l2s hybrid things in between and there's this massive surface area of interesting complexity between blockchains like how do you move assets between them how do you move transactions between them how do you move value between them and what are all the opportunities that that creates whether it's Mev or whether it's you know more interesting economics um there's a huge amount of work happening and opportunity in well how do you do all this stuff with different approaches to information itself how do you do the same type of transaction but in a way where we incorporate a zero knowledge um proof to let's say you know change the nature of a transaction itself how do you do the same stuff but you don't leak as much value or leak as much information and I don't think it's really the D5 building blocks that are interesting I think it's the different ways that that these building blocks are going to be thought about as a huge overall system not just one blockchain and it's Mev but all the different chains and transacting between them and the opportunities yeah I think um you know if you think about the last you know 2017 you had sort of this huge speculation bubble that funded a lot of infrastructure and public goods of different forms so that it was easy to for someone to create something like unisop or compound that was not sort of what the intended you know I think if you look at 90 of 2017 icos none of them really looked like anything particularly useful I think 90 is generous sorry yeah yeah 99.99 yeah yeah uh but but you know like the idea that the speculative bubble fueled infrastructure that made it easy for people to do very cheap experiments I think right now we're kind of seeing that the ZK side of the world and the L2 side of the world are having that kind of boom and inevitably there's going to be some certainly there's going to be new assets that come from The Real World like Sable coins treasuries things like that but there's also going to be kind of some notion of like how do you deal with assets on all these different domains uh and how do you kind of do that efficiently and I think people will now that people have these tools there's been this huge boom that funded creating all this infrastructure now that you have all these tools you're going to find very unexpected ways that people will make things that look like a uniswap LP share but takes advantage of either the ZK or multiple domains type of things so one thing that I feel like we we have seen an evolution on kind of how people think about it over the last few years has just been around what does it mean to be composable what does it mean to be uh kind of using these building blocks that are kind of D5 Primitives and and how does liquidity sort of come into play and we're going to see more and more of that fragmentation now with like multiple l2s multiple uh we're seeing l3s now too um has that changed kind of your definition of what D5 looks like or sort of how do you think that overall effects or or intersects with where where we would like to go but but like these things are making it harder now so if we think about the internet itself right the world was sort of singly threaded and synchronous for a very long time in fact it took like operating systems making it easier to do that for people to have asynchronous compute and like the cloud and things like that uh in some ways you could argue that Finance has a very similar path right like if we went to the 1920s there was only one stock exchange and everyone you know rushed to go trade in the stock exchange but over time this technology got better people started actually doing multi-threaded asynchronous not all in one place type of settlement and inevitably that always happens the more you have complex requirements for different types of assets so they'll certainly be fragmentation I think the key thing is figuring out what technologies make the cost of synchrony the cost of feeling like it's synchronous to the end user uh the cheapest and I think it's sort of again maybe this is not I was like in this case look at the commonality would still be the TCP packet is is what's shared but is that kind of we're betting on bridging between two chains becomes way too seamless or or do you think that sort of blocks T5 altogether from growing because we have to like make a zero-sum choice of where does my money go here or there or somewhere else I don't think that there's going to be one representation of an asset on each chain independently in the long run I think it's there is going to be a world in which synthetic representations are treated as effectively equal and you have high guarantees of security for that you know I think a lot of the kind of crazy stuff you saw in 2021 was people thinking that two synthetic assets on different places were worth the same amount and that the bridge was safe right and like now we've kind of Switched too far in the pendulum where it's like no no no Pepe on arbitrarm and pepe on optimism are really two different things and there's no way that we're going to try to reconcile them and I think the pendulum will swing back once we actually have a lot of these extra assurances whether they come from things like ZK or MPC whether they come from things like different user modalities for how users store things so one of the things I find the craziest of the last three months is kind of the rise of these telegram Bots that people used to interact with um I'm not sure if this blog post is out yet but I think flashbacks is running this blog post that shows like the different distribution of front end volumes from you know metamask versus um unibot no sir got a unibot but there's like 20 of them yeah and the different different telegram Bots the telegram Bots were like 12 to 15 percent of order flow going into Mev auctions right now and they didn't exist like three or four months ago right so significant so it's clear that like user how and some of the telegram Bots you might say hey I want to buy a bunch of Pepe and it it's it it basically has an NPC wallet that you might own some of the key shares too and it basically will do something like hey I bought 20 of Pepe on optimism 80 on arbitrum but like you the end user does don't care you're like I have X units of Pepe that is a different modality than thinking about having the bridging right and like changing the user modalities not only the transaction semantics is actually quite powerful and what we're seeing a lot of right now but they're not the same asset they're not they're not bad and some sharp finance-minded person at some point will find the opportunities um inherit in that and make a shitload of money is that talking about just there will be an Arbitrage regardless yeah at some point you know any synthetic asset stops being an a mirror of the underlying um due to some Edge case that will occur at some unpredicted unknown point in time um and will separate the uh the cream from the milk I mean I think we're seeing early hints of that with steak teeth there's already a few verses of it and you see uh a lot of nuance in how things are priced very good example um well I want to kind of do a similar version that same question for you Robert uh you work kind of a lot early into thinking about compound chain and and having an app specific uh sort of use case here how do you kind of now think about that space Maybe comment on sort of what happened with compound chain if you're if you're able to it and so I'm now seeing a lot of things around LP stack or the base or just seeing dydx what does that mean to you and how does that feature look like yeah I'll try to sum the whole journey up in like 90 seconds so long before there was a ton of l2s um before we got to this point you know around like 2019 2020 you know we started to say well what does the future of compound look like in a multi-chain world and the theory that we came up with was instead of having a version of the market deployed on every single L1 that might exist or every L2 that might exist where liquidities fragmented what if you could make one giant market with the assets that are native to each of those chains and it just worked and the approach we came up with was to have an app chain whether you called it an L1 or what an app chain that had a bridge to each one of these l1s or l2s and it would allow you to do really crazy cool things conceptually like having ether native on ethereum and locking that and then borrowing avax on Avalanche without having to even like go through and like Bridge your assets or send them anywhere you just lock it on one chain and borrow an asset on another chain um and this is like a crazy cool concept no one has built anything like this since mostly because what it's really really really really technically hard and two what actually prevented us from launching was we got uncomfortable with the security risk of having like six Bridges um this was ahead of its time but what could go wrong here what could go wrong I mean this is like a bad sort of like Tech strategy to even get into this position to have like six native Bridges um you know I would never wish anyone to have more than one Bridge um but we basically became uncomfortable at the risk of maintaining so many bridges um it was just a bad design in a sense and we basically got to test net ran on test and we're like we're never going to be able to run six Bridges safely like we're gonna get hacked and this before Bridges got hacked because there weren't a lot of bridges yet and we just said like this is too much of a security risk like we have to table this this is so do you kind of feel like that argument no longer applies because it's so easy to roll up your new chain now or do you feel like people are just gonna try it but then we'll realize that it doesn't actually work so we'll come back to a few big l2s yeah I don't think one project will be so ambitious um I think what you're actually seeing in a sense is like projects biting off smaller more manageable Endeavors um you know I think you'll see things where like you know one chain will get assets somehow from lots of different places but it still won't be like the crisp user experience of like oh you can do something on one chain like and it basically instantly happens on the other chain we're getting closer to that like there's like a lot of solutions that are like slowly cobbling us together um uh I mean both of you kind of brought up ux with just what that bridging interface looks like you with the unit bot or just telegram Bots um what are kind of some obvious things or maybe even some non-obvious things that can help make ux sort of the priority where we actually get more people to use stuff in defy I think For Better or Worse even though I think maybe purists may not be happy with this I think the NPC wallet it's our you know I I you know I I think there's kind of this unfortunate inevitability and that that's the middle ground between users actually paying attention to their seed trays and security um you know obviously I think there it would be great if everyone just used a smart contract wallet and you know they had recovery and all of these newfangled nice things but I think at the end of the day the MPC wallets just start about 10 times better for the end user right now so this is the argument for people not having to remember they're or having like a way to recover their seed phrase if they forget or the fact that they shouldn't be assigning things every time a couple things right like I think there's gnosis safe like functionality that people get out of these NPC wallets where I can say I can make rules that are like okay you're only allowed to move more than a thousand dollars if both shares sign um but if only if the transaction is less than ten dollars then you only need one of two and you know maybe that's you know the unibot thing rebalancing your portfolio for you or like the unibot thing going and getting yield for you somewhere and this idea of having rules-based interactions versus direct transactional interactions I think is is like the number one thing that these NPC wallets are getting right the problem with most smart contract wallets um is that they still are transaction based and there's there's still kind of like the the user has to specify everything it's very hard to do permissioning like obviously Nosa safe is like the king of this type of stuff like they figured out how to do a lot of this with safe modules but I don't think safe modules have reached the point that like the telegram Bots for instance could use them like the telegram Bots could do everything in nose is safe and there's a reason they don't and a lot of it has to do with just like the access control pieces are still a little bit annoying to do and I think NPC wallets are not going to go away uh and and yes they're weaker security wise like there's let's not kid ourselves on that but the average user of these telegram Bots does not know the difference between that and the first version of these things did the cardinal sin of making you type in your seat they already passed a test by putting the private key on telegram yeah yeah no I I'm not I'm not Advocate I'm just trying to point out that when you're talking about ux users already are doing this and this NPC wallet thing is already so much it's like the 80 20 solution for them like I want the smart contract wallets to win but I I just think the transaction level semantics of them are sort of like annoying and like I think there will be a way around it it just might take a bunch of development I have a really counter-intuitive and controversial answer and it's one that I feel partly responsible for getting to the state of the world I think the user experience of sending your money into a smart contract and hoping that it's safe sucks um I don't think D5 has to work that way I think you can rethink and re-architect things um to come up with a different Market structure so to speak for defy itself but I think the idea that everybody sends like two billion dollars into a contract um is a really really really bad structure this is already commenting on the the flow of how that happens or the fact that it shouldn't land in a smart contract I actually landed a smart contract I think you could have a smart contract that controls billions of dollars of financial transactions and routing and value but the idea of like everybody keep your money in a protocol like compound and hope that it's safe you know it can be done safely but like it's not a model that should be done is there an obvious better route uh I think eventually what we're going to wind up getting to and there's approaches to do this I've been spending some time on this a little bit um is I think eventually you wind up with a market structure that looks a little bit like oh assets stay in your own wallet or in your own custodian until they settle and you can have complex systems and complex logic where you can like metaphorically do all the same stuff you do with a compound the need as well but with like having more actual user control and custody over the assets themselves um you don't have to literally move the token into a contract to get the same outcome in the same logic you're very Pro MPC wallet though yeah I was thinking is uh three people controlling where the money goes well no not even that I mean like you can just like re you can just architect an actual D5 protocol differently that is a very controversial affiliate uh he may be interestingly I I promise people that won't go into technicals at the beginning but um you're effectively defining a lifetime of an asset which if you look at something like Cadence or any kind of that type of architecture you effectively separate the ownership from from the execution and that's like a a good mental model to think about this um you are also investors putting on your VC hats uh what is interesting in the the space right now as a upcoming opportunity or or things that you think that are are worth or finally ready for commercialization I mean not to sound like a you know dead horse but I really I do think the ZK stuff like application wise is the biggest open field right now um you know tens of probably like close to 10 billion dollars has been spent on infrastructure and we're going to have a glut of proving Supply in the next 12 to 18 months both in Hardware proving and proving for uh different types of applications and so far the only application that's going to use is billions of dollars of infrastructure is Roll-Ups and no offense ZK Roll-Ups are a very boring application of ZK all they're doing is effectively compressing a bunch of State transitions it's not even doing ZK right like you know it the system I did it yeah you're right a snark is not even using like privacy features there are no circuits in there yeah and and and and then the main point is that there's a lot of other stuff you can do with you there's a lot of stuff you can do mechanism design wise there's a lot of stuff you can do in terms of separating you know execution from ownership there's a lot of stuff that I think people were so focused on building this infrastructure to make Roll-Ups faster that they kind of like lost sight of like hey you could actually use it for other things and that is the ultimate subsidy that is being given out to everyone who's thinking of a hackathon project right there's just there's going to be so much proving Supply in the next 12 months I I personally am really motivated by the idea that you can actually use a blockchain to do Finance not in like a toy way but in like a Wall Street will run on this in 10 15 years way um and all the people that are looking to actually like legitimately Bridge The Divide between how the trillions of dollars moved and what we can actually do with the blockchain is the area where I'm most excited about maybe just to dig just a little bit deeper what's the most obvious thing that needs to be Bridges so I just convince them to see it this way or like a pure technical thing that's just missing for performance I mean there's technical things there's you know like security things there's infrastructure things there's user experience things there's bridging assets over to a blockchain things there's you know making things more private because honestly like the execution of how you do a trade on a public blockchain it seems silly in a lot of ways like that maybe shouldn't exist is that what we're saying I think MPV shouldn't exist like there's some impossibility results on that yeah but like you know like you know swap is the greatest thing ever but it's not Wall Street grade you know like I get from one word I use I'm like you know it there's just so many hundreds of things that I think like all sort of need to happen before like you know it goes from being like wow these are incredible proofs of concept and like they do work and like that's what's so exciting about this is like we are proving again and again that you can use a blockchain and contracts to do really valuable and important things like yes we've like proven that you can do it now we have to like Harden everything and like it actually happened yeah make it actually happen like where the state of it today is not like what wall Street's actually going to use like we're delusional if we think that another controversial opinion I remember actually that one's up no I actually agree with this one yeah yeah 100 agree I don't know why agrees with it it's not background yeah exactly ouch you know it's okay um well my kind of final question to you is can I have you been in this space for so many years and seen all these ups and downs any any parting advice to Founders or looking to start something in in this space this year uh have fun don't do something boring uh go wild uh like I said there every cycle in the booleans there's a lot of stuff funded and that stuff that's funded is free stuff for you to use to build new things and it's been heavily subsidized so you should definitely take advantage of those because whoever figures out the first application that is able to monetize get users whatever from that those new technologies will be the creator of the next wave right you know you could argue that that was the the first member advantage in this in this setting actually matters it really does so like take advantage of all the free that's you know VCS are funded for you all the free ZK compute ZK compute layer twos data availability all of these things are like you know those billions of dollars of subsidies so amazing Tyrone Robert thank you so much for taking the time today and uh really appreciate this conversation thanks Albert [Applause]