How Appchains Align DEX Incentives | Season 6 Episode 7

Transcript

think it does send a good message about the fact that we're taking this seriously and really do want to be aligned with the users the Traders specifically the Traders I actually don't care about anybody else in the ecosystem like the entire points of all of this stuff that we're building is to build the best possible product for our end users all right everyone we will be back to the program in just a moment but before we do I want to share something that blockworks has been cooking up for these last couple of months March of this coming year in London blockworks is hosting D London the largest institutionally focused conference in all of crypto Goldman JP Morgan 72 all in one room so you can know what the big money is doing so click the link at the bottom of this episode it'll take you

right over to the homepage and use Bell 20 for 20% off I will see you in Sunny London Town in March this episode is brought to you by Maverick protocol a suite of liquidity tools built around an Innovative amm Maverick helps token projects Dow treasuries LPS or basically anyone in defi shape their liquidity with efficiency and flexibility how you might ask stick around and you're going to be hearing about them more later now on with the show all right everyone welcome back to another episode of B curve before we jump in quick disclaimer the views expressed by my co-host today are their personal views and they do not represent the views of any organization with which the co-hosts are associated with uh nothing in the episode is construed or relied upon as Financial technical tax legal or other advice you

know the deal now let's jump into the episode hey everyone welcome back to another episode of bell curve in today's episode Dan and I interview Antonio Giuliano the founder in CEO at dydx this was a particularly fun episode not only because Antonio is such a forward-thinking and I think Visionary founder but also because we got to reexamine a lot of the themes that Dan and I have been exploring this season primarily through the lens of dap dexes that are building on ethereum from the perspective of the operator of an app chain so a lot of the questions that we've been asking our ourselves this season such as how Mev leaks um outside of dapex is to proposers in ethereum is very different when the validators are

actually part of dydx chain itself we also got into some really fun Nuance around some of the work that validators can do for app chain and some of what's possible to build on an app chain say having your validators actually run uh you know act as oracles and run an order book on your behalf um we also got Antonio's perspective on the relative market share of pers versus spot going into the future so overall just an extremely fun episode with a with a Visionary uh founder so hope you enjoy all right everyone welcome back to another episode of bell curve today Dan and I are joined by Antonio Giuliano the founder and CEO at dydx Antonio welcome to the program hey thanks so much for having me thanks we are we're really excited about this um because I think

this is going to be a fun sort of culmination of a bunch of the topics that Dan and I have already covered this season we're going to get deep on the idea of me um we're going to cover frequent bch auctions again uh because that's a part of V4 for dydx uh albeit from a from a totally new perspective because Dan and I have been approaching most of the season from the perspective of a a Dap Dex building on ethereum um and instead we're going to be talking about the app chain perspective here and hoping to get into the nitty-gritty granular weeds there um and I also maybe just want to say as a disclaimer this is um assuming a little bit of basic knowledge about dydx and per and all of that good stuff uh as well so just to dive in here Antonio Dan and I have been

one of the the big themes that Dan and I have been exploring this entire season uh is this idea of reducing uh Mev leakage um outside of Dex ecosystems uh and some of that loss ends up going to kind of the the proposer in the system of ethereum but obviously in an app chain context that's that's very different so I guess just from a very high level before we start to walk into um some of the architectures that you have for for mitigating or dealing with Meb and you know how do you as the operator of a of an app chain IND deex think about Meb that might be a little bit different from the ethereum sort of canonical perspective I think the most important thing to understand is that the design space is bigger in my opinion because you as the app chain own more of the

stack obviously in something like ethereum there's no preconception that validators should care at all about any particular app running on top of ethereum like it's not really the validators or potentially even the proposer responsibility to maybe do something like minimize me on Unis swap or whatever other random decks might exist on etherium but that could potentially be fundamentally different on an app chain um and I think it does make the design space bigger I won't say it's objectively better I think it comes with positives and negatives to it um but I think it allows us to do things like potentially align the incentives of

the validators the token holders and the users in a better way and more aligned way than what we could do on top of ethereum so I think there's a that's a pretty high level thought and there's a ton to dive into but that's what I'll start with yeah I mean I'd be curious um when you when you start to think about that alignment like one of the things that that stands out to me is you know in sort of ethereum land Dan and I spent a lot of time talking about how like really you know when you think about you know LPS and swappers on something like Unis swap they're actually very United today in that really what we want to do is there's an enormous amount of Mev that ends up leaking down to the ethereum proposer and folks on the app layer are sort of uh you know laser

focused on mitigating that that leakage and then just redirecting it back to the system of LPS and swappers and then eventually maybe somewhere down the line there's like a little bit of a battle for what's the right amount of uh fees to take Etc but we're just a long ways away from that um I would love to get your perspective on you sort of have this more explicit third stakeholder here where the validators and your ecosystem are running you know sort of consensus based uh code but also the application logic as well so how do you think about balancing sort of that Meb that gets created between those three stakeholders and then how much do you think about like redistributing meev versus just trying to mitigate it entirely so I think that's a really good characterization I would also argue that

there's actually four stakeholders in the ecosystem um so the three that you said effectively the makers and the takers and then the validators as well I would also argue that us as developers because we have access to more of the stack that we're building on are more we we have basically more tools to work with um for meev on top of an app chain than we would if we were just constrained to building smart contracts or other types of decentralized systems like proposers that could you know bundle up back to the base chain so with that in mind um I'll introduce two level two highle concepts that we're thinking about right now and the other highle thing that I'll say before I dive into literally what we're doing is that

although I think this is a really exciting large design space it's also very unexplored and I think it's much more unexplored than me on ethereum so far just because app chains are newer there haven't really been any projects with super significant trading volume that that happen on an app chain and we aim to kind of set the standard alongside other really good partners like Skip and others for what me can be or what me mitigation can be on an app chain so what are we actually doing the thing that we have built into the platform right now is at a high level what we're calling social slashing um based on we we effectively came up with this metric where you can score the

execution quality of the trades that each particular validator is executing relative to the mean or the average and then you can compute Statistics over the long term and be like ah aha this particular validator is executing at a price that's 10 basis points whatever farther away from almost all other validators they must either be doing me or potentially have non-performing Hardware software whatever else but whatever like you don't really want that particular validator to to be in your network um and I think just being able to come up with this one particular metric from an AUD auditability perspective is valuable now of course this is a really open-ended question and

a lot of opinions on whether this is actually going to work or not and we don't necessarily think it's going to be the be all and end all of me prevention but we think it's a good start and I think it does send a good message about the fact that we're taking this seriously and really do want to be aligned with the users the Traders specifically the Traders I actually don't care about anybody else in the ecosystem like the entire points of all of this stuff that we're building is to build the best possible product for our end users um so so just a little bit more on how this social mechanism works so effectively skip which is a thirdparty provider has built this dashboard which tracks this metric over time um and then you can look for each

particular validator how good or bad is their fil execution quality and then the theory is later either the token holders some governance committee still sort of being figured out by the community at a high level can come in do things like potentially slash these validators and their resp active stakers or take other action um based on how egregious the the uh execution scores are now this is really a game theoretic problem right like how is this going to end up is it going to be like over time all the validators maybe there's not one that's super egregious that's kind of obvious but maybe they all start doing a little bit more me over time to the point where they're effectively outsmarting the

metric ability to catch them that could happen I can't this is really a big experiment um but I also think it's possible that at least from a messaging perspective this one buys dydx a lot of time on the meev side um because these things the these games that you play on both sides from the of mitigator perspective and the extractor perspective take a long time to flesh out and the sophistication on both sides I think increases over time um so that's the thing that we've literally built or skip has literally built on top of the protocol right now um we can we can dive into that at a high level um the second thing that I'll just introduce really quickly and obviously we can dive more into this later is more protocol level

mechanisms to mitigate me um and I'll kind of introduce a concept at a high level that some who are familiar with Cosmos may be familiar with but probably most who aren't familiar with Cosmos might not be and this is the concept of vote extensions and what vote extensions are at a high level is a way for not just the validator or the leader that's proposing a specific block um to put information into the blockchain but at a high level it allows all validators or some subset of validators to put information into every block even if it's being led by another validator and this concept is actually really powerful and I think is a good example of what I

was talking about where we as the developers of the software behind idx can use this tool to put more information into the blockchain and effectively to use all the validators to come to consensus on what the actual orders that or what the actual trades that are put into the blockchain are going to be um and right now this does not exist on the dydx Chain this is something that our research team is working on right now from a high level um we've kind of really taken our research team at dydx from zero to one in the past six months but and I'll talk dive more into this later using this concept to not just have the one proposer leer have full say over all of the trades and all of the matching that

happens on the Chain but effectively use all the validators to work together I think is a powerful concept that we're exploring too yeah so maybe uh starting with I think it's a great overview starting with the social mitigation strategy I guess you know I come from a world where I'm kind of steeped in the belief that social mitigation for Mev is hopeless and so we should actually just steer in the other direction and accelerate it and encourage encourage me extraction within within certain bounds as much as possible um and I think I think you've gotten some push back from from that school for um for for sort of attempting something like this the sort of social shaming of of meev extraction um yeah I'm I'm curious how you react to yeah to

to that reaction okay well I first of all I just like don't really care obviously I respect you and respect all of the uh opinions of the people who have thought about this for a long time I mean Dan you know this that we have been talking about with Paradigm and Dan for example for a while strategies on how we can mitigate me um but the let me be more specific about what I don't actually care about um I don't think it's necessarily bad for these things to be experimented with in different ways from an ideological perspective again I think we do have more tools that are available to us through building our own app chain than most other platforms of significance have had so far so I think

that does warrant uh at least experimentation with this this social slashing in a different way again I think it's a big open question about whether that works long term and honestly I don't think that it's likely that's 100% of the solution in the long term I do like I was mentioning before I think it buys a good amount of time I do think it sends the right message and honestly it's just not very expensive to build um you know kind of is tasking the research team on how can we come up with this one metric and then how can skip Implement something to track this um I think it was certainly worth doing so I think there's a little bit too much of ideological Dogma in general in crypto which I've probably been well documented on on saying before really is where are you seeing that too many Nerds that are

upset about their technology yeah probably um but anyways like I try to be really agnostic to the technology and even let's extend that to the game theory of these mechanism design questions that we're coming up with so that's probably the main way I would respond like probably that's true um but I think it's worth experimenting with especially in a lowcost way like we've started to do yeah and the way I see it you're going to need some kind of mechan social mechanisms anyway to prevent collusion for example in the in the you know in the end game um I think you're still going to potentially be vulnerable if if two-thirds of validators were all to collude on censoring and ordering blocks and if you if we just uh encourage the idea that oh it's fine to

do anything that the economics that the computer science allows um and that the economics encourages then ultimately I think you're going to you're going to run yourself off a cliff no matter no matter how many mechanisms you build in so I do I do agree that I think a lot of the challenge is is it actually possible to to detect um uh collusion or you're just going to end up in an arms race with with uh Searchers but I think I think you've you've recognized that as much as anyone which is why again I think you're doing so much good research on the on the long the long term um and moving moving to other mechanisms yeah I would highight this as a difference that you tend to see in between ethereum and Cosmos just having spent a bunch of times in both ecosystems whereas ethereum is dealing

with a very different problem because it's generalized blocks based there are so many different validators um that has certain advantages but also certain constraints like you couldn't be like let's just kind of come up with this social Rule and give it a shot whereas in Cosmos you tend to see quite a bit more of that actually um which is sometimes very refreshing but also many Cosmos change don't have sort of the uh intense privilege of securing you know $250 billion worth of value um so it's just a very different problem I think that you're trying to I think it's a different problem and one of the decisions that we actually made on the dydx chain is at least to start we decided to make it very special purpose um and there was actually a pretty big open question about what do we enable things like cm wasm to enable general

purpose execution on top of the dydx chain in some sense at first glance it might be like oh yeah of course why would we not do that it's pretty free actually um but then you have to start worrying about things like how do you share block space between the generalized execution environments and anything special that you might do how do you mitigate the concerns of me with the generalized stuff and the specific stuff like how do we trade off against those two and I think likely we will move more towards generalized execution on the dydx Chain but in an intentional way and it's always easier to solve a lot of these problems and they're big problems when the design space is simar yeah maybe we should get a little into the um into the the batch auction solution that you that you announced last month

do you want to describe describe in a little more detail how that works and how it takes advantage at a high level right now it's the the way the dydx chain works is very similar to a normal order Book Exchange where whoever's the leader um as per the software gets effectively full authority to match the orders however they want and specifically however they want consistent with the limit prices of the orders that are put in the blockchain um so to some extent this is effectively fine this is more or less what exists on top of things like ethereum and basically every other decks right now it's a little bit different because it's an orderbook based system versus an amm one and that has different considerations um again we have this social mitigation feature right now um

but we think there's more to do here so the big idea about the frequent B batch auctions is that orders would no longer be matched on a onetoone basis but instead each particular Market would come up with one clearing price again still for the time being uh that clearing price would be put in by the leader of each particular block potentially over time you could include things like validators working together with vote extensions to come up with the clearing price for each particular Market um but then it's kind of like okay let's maybe take a step back here and just talk at a high level about how order books work it's kind of like there there are some ask orders that are selling assets there are some Buy orders

if you've ever looked at an exchange before you know what that looks like and then what is matching matching means that normally the ask orders are here and the bid orders are here and there's some spread between the two but if there's a bid that actually crosses this ask price or an ask that crosses this bid price then you have should have these two people trade with each other at this particular price but what price that's kind of the question here the the price that could be valid for these two orders to match together is anything within here and maybe numerically let's give an example of that so if there's an Ask order to sell at 10 and there's a buy order to buy at 15 then the leader right now as per the software can make those two orders clearer basically

anywhere between 10 and 15 and maybe we can come up with a better solution for that so the the problem here becomes it's really it's can be more susceptible to things like sandwich attacks where if one of the leaders injects their own order um and then they take a spread based off of that then they can profit um but it's a lot harder to do those types of attacks if you're using this frequent batch auction where there's just one one clearing price for every order that matches in our particular Block it's not a golden ticket to solve everything but we do believe it solves a good amount of classes of attacks on me so it's something we're seriously considering over time now let's Let Me Maybe talk from the other side because we've been talking a lot about from a

technical perspective why do we how do we solve me it is a big problem but there's trade-offs too um and one of the reasons that we didn't start with the frequent batch auction and it's one of the things I'm still honestly thinking about from a product perspective is how does this change the product experience of trading on dydx because one of the things that we've tried to do from a product perspective on idx is make a product that's very similar to normal exchanges or at least Web Two exchanges and make something where professional Traders and market makers and the like don't have to change their trading strategies ideally too much to trade on dydx at some level especially now that we've come fully decentralized we have to accept that that's probably

inevitable um and we should change the the strategies over time or kind of change the rules of the game to make the game as Fair as possible um but there is some cost to changing the game it's like market makers have to figure out how to quote now on frequent batch auctions instead of order books maybe they don't care that much about that maybe they do but that's what building great product is about you have to consider these product trade-offs get a lot lot of feedback from your users and then decide whether you ultimately want to make that change yeah so so when you've introduced once you've introduced frequent batch actions you do have the issue that if you still have individuals proposing blocks then then they can censor or um adjust blocks and and potentially still kind of effectively sandwich users or or

extract value from users um so you want to talk a little about how you solve you proposed to solve the censorship problem yeah and this is really where vote extensions start to come in and that's what I was discussing before um but the highl idea with vote extensions and we have a number of different ones that we're exploring on the research side I think we're we've mostly finished the ideation process so we're always open to new things there but our best idea right now um is at a high level like I was mentioning before all of the validators or at least some subset of the validators work together to put into vote extensions all of the orders that they see um and then this is a really powerful primitive because once you have in consensus all of the orders that should have matched that they see to be

more precise um once you have all of those orders or proposed trades in the blockchain you can have the subsequent leader look at the data that's in the blockchain and provably use that data to come up with what the matches should be on an ongoing basis um so there's potential trade-offs around this as well um and in kind of the worst case it could uh introduce a one block delay to finality is probably not the end of the world like the block times on ydx are on the order of a second but now maybe you're waiting for okay it's like order is place then block one all of these validators together puts all of the trades that should have happened into the blockchain and then block two um the the new leader actually uses that data

that's in the blockchain to come up with those matches together but this really does solve in a big way at least as per the constraints of how many validators the two-thirds consensus all that sort of stuff and in tender the problem of denial attacks um where it's like now no longer can a leader ignore certain orders or insert or they can still insert orders of their own but they're subject to the same rules that that everybody else is subject to um so this is the biggest I I think long-term solution that we'll probably have for meev or this is our best idea right now but one of the things we're pretty excited about in the long term um just talking about there's always trade-offs and trade-offs and trade-offs with all these sorts of things but it's one of

the questions that we're trying to solve right now is how do we make this happen from a performance perspective um because the way vote extensions work at a very high level is like the leader exists then they propose a block then they send that block out to all the other validators and then vote extensions which is something that's new within Cosmos um allows all the other validators the followers whatever to put data back to the leader that they have to put in the blockchain um this could be a lot of data like if we're saying it's order number of Trades that are happening on the dyd exchange for each block now it's multiplied by all of the validators um so before we kind of had this data problem of okay there's obviously we have to put all of the

trades on the blockchain into the blockchain but only one validator has to do that now at least for vote extensions it's like you multiply the number of Trades that are happening per block times however many validators you have so we're shooting for like 100 in the near future or so I forget what what it is off the top of my head right now I think like 50 or something but now it's like 100 times as much data um so one of the other problems that we're trying to solve is okay first of all uh how important is this like will this degrade the performance of the network um we think it's relatively important because honestly one of the problems with boat extensions right now is that they haven't proven yet to be very performant they're very new and I think the cosmos team skip others are doing good work in

terms of optimizing them um but for us from a mechanism design perspective are there ways that we can optimize this um and maybe have validators only put a subset of the trades into the block or only the ones that are different from the ones the leader proposes or something like that so there's a lot of different ideas that we're considering but that's the type of work that we're doing on the research side that's that's super helpful Antonio and I I want to dig uh deeply into there but I actually want to just take a moment here to connect some concepts with folks that might be a little bit more e uh similar with ethereum so a lot of what Antonio is describing here and what ABCI Plus+ enables you to do uh sort of rhymes with a lot of the conversations that get had uh in

ethereum around Pepsi um and sort of all stem from this idea of how to conduct a censorship resistant auction and to break the sort of proposer Monopoly so just to to break down in you did a great job of describing that but just to show whenever it's a complex topic I always think it's good to get pictures maybe that's just because I'm secretly five years old but I think it it's helpful for folks so just to go even into a little bit more detail about how uh you know one of the the differences in in Cosmos is that every uh validator runs both um you know the consensus client that's tenderman or I guess Comet bft now but also the application logic and then there's this thing that Bridges them which is ABCI um previously and in the and then uh in the ABCI regime uh

basically consensus comes up with a finalized block which is the input to the uh application layer and if you're following along via video which I would encourage you to do you can sort of see a diagram of what that looks like but in ABCI Plus+ it kind of inte it uh there like these three sort of touch points that happen as the block is being built where the application can actually have input um and more collaboratively build a block um and that's where sort of vote this idea of vote extensions come in when actually uh different uh validators in the chain can submit uh transactions which are then deterministically included within the next block hence how you can get the the censorship resistant uh the censorship resistant auction I think which is is is that correct um

Antony just to make sure I didn't butcher that um overall but I feel like that's the okay cool I think that's basically correct I mean honestly I'm not quite as familiar with the state-of-the-arts on me on ethereum as you guys probably are and potentially even some of the listeners are but I would make the high level point that a lot of the stuff is very similar and I really suspect that the endstate solution for app chains and uh layer one layer 2 systems on a bigger more generalizable chain like ethereum are going to be similar it's probably like the things on top of ethereum will end up building probably something pretty similar to whatever ends up working for the app chains but then also solving the problem of how do we have the settle

back to a base layer system it's just a little simpler again for us right now on dydx and some of the things we have going for us is we have the whole stack like I was talking about can be more tightly coupled with consensus and then it's also just honestly really nice that we have this decentralized network of entities that we can kind of make from a software perspective do whatever we want as long as there's mechanism designed around that and this topic this concept is really valuable so it's like this like proposer Builder like separation all this sort of stuff okay we just have like the proposers be the validators obviously um and then even for something that's totally different like Oracle prices normally if you're building a per product on top of more generalizable

chain you have to solve this problem of okay well great like your consensus is fully decentralized to the level that ethereum is but who cares if your Oracle network is less decentralized then your entire product is going to be less decentralized than that and you can use products like chain link or or others that are doing a decent job solving this problem but for us it's actually really nice like we literally just have like 50 to 100 already decentralized entities that we can make do extra stuff like report Oracle prices on the network um and I would argue that people have too narrow-minded of a view of what decentralization means I think people reduce this problem too much to what is the level of decentralization of the base chain itself obviously very

important since everything is clearly limited by that um but I would kind of argue that the full decentralization of the system is kind of the Min of all the decentralizations of the critical pieces of your network and for some like a spot exchange maybe there are fewer but for a derivatives exchange okay now we got to worry about me like we're talking about or we got to worry about Oracle prices or even if we're going to follow this to a conclusion we got to worry about like the decentralization of clients and front ends and stuff like this um so I think we just try to take a more holistic approach to what is decentralization and then you know I've actually been impressed with how little people have been lampooning us at least

for the past year or so on making the decision to switch to our own app chain because hopefully we're do a good job communicating it um but there still are those out there that are like oh well you have like 50 to 100 validators this thing is clearly not decentralized at all um and yes like that that's a Min on the amount of decentralization that we can have but I would argue the entire system is more decentralized still than at least any other per exch decentralized per exchange that's out there and that's really our goal all right everyone we will be back to the program in just a moment but before we do I want to share something that blockworks has been cooking up for these last couple of months March of this coming year in London blockworks is hosting D London the largest institutionally focused conference in

all of crypto we are gathering 1,00 of the world's largest asset manager so than trafi macro funds Crypton native funds big allocators and financial institutions so Banks payment processors Etc all in one spot it's very rare to get the likes of Goldman JP Morgan 72 whatever all in one room so you can know what the big money is doing we're diving into the themes that they care about so we're talking about the intersection of macro and crypto where we are in the cycle real world assets so everything from stable coins to onchain treasur IES to tokenize assets it's going to be a blast but the other reason you really want to go it's London baby center of the world at one point you got Pub culture you got fish and chips great beer it's going to be a blast so because you're such great listeners to bell

curve there's a Code Bell 20 that's going to get you 20% off so click the link at the bottom of this episode it'll take you right over to the homepage you'll see all of our speakers and use Bell 20 for 20% off ticket price are going up soon make sure you go use that code I will see you in Sunny London Town in March May that's a good place to segue and talk a little about DX's transition from originally being an L1 ethereum application to being an L2 to being um a Cosmos app chain um so what was some of the trade-offs uh and and factors that made you make those uh decisions as a project yeah so we've been around for a while I mean dydx is like six and a half years old at this point so we were in D5 before def5 was

even a thing um so obviously we were building on layer one ethereum um because that was the only thing that existed in town and that made a lot of sense at the time the reason that we went to L2 is pretty simple honestly it's we were getting destroyed by the amount of gas fees that our users and kind of the company to some extent on behalf of users was were having to pay um and this didn't used to be a problem at all um but then in the bull market of 20 or 20 2020 or so when people really started trading on Unis swap compound using actually using these products the gas Fe just went like a THX through the roof which was hard to foresee and then honestly the decision at that point in time was like oh my God like this is

terrible like who's going to use an exchange where you have to pay $100 to make every single trade on the exchange and I think that really hit home more for us at dydx at least sooner than some of the other decentralized exchanges of the time because our users are more professional than the users I would say of pretty much any other decentralized Exchange um and when you're kind of YOLO in to some random shitcoin on Unis swap or whatever you have to pay $50 gas fee or whatever but the thing could kind of go up 100% in the next two seconds um you're probably not going to care quite as much about gas Fe you just swap users obviously still do care about gas fees but probably less than dydx users do is

the point I'm making here we had to solve this problem um and we took a look around at what was possible to build on uh at this point in time in 2020 entered into a partnership with starkware to build the well I'm used to saying current dydx products I guess it's not the current dydx product anymore um but the V3 version of the dydx product and that's what exists today and there certainly are trade-offs there and the thing that we had to do to enable that level of performance at the time given the limitations of the technology was introduced somewhere centralization on specifically effectively the sequencer side through starkware and then on V3 and all the way up until V4 uh us at the

company behind dydx were operating parts of the exchange and we're operating things like the matching engine and that worked pretty well and that's taken us to where we are today with about a billion dollar a day or so traded on the platform and you can trade on dydx with almost no gas fees basically no gas fees and that's a pretty awesome trading experience and again we have to really look at who are our users they're different than most other dexes but I think there is a really fundamental reason why we were I would argue the first major project to really Embrace L2 in a really big way and at least by trading volume the biggest product on L2 today so that's why L1 to L2 I don't think that should be huge news to

anybody at this point I think the rest of the market has finally caught up that this is something that is exciting I mean I think people are always excited about it to some extent but it's also feasible now with things like optimism base um Stark net all those other other stuff and now like the L2 Revolution is really on so you might ask the question why yeah yeah I I do think it's interesting because I think you were you were somewhat early on that and particularly on the question of composability because I think early on other than dydx having I think for a while the cheapest flash loans on ethereum L1 I think you've you've kind of issued composability um or or Atomic interoperability with other applications as a as a major goal um and I think with with moving to too and then uh and then

moving to L1 you're sort of moving further away from collocation with other ethereum apps um do you think has that been a major sacrifice for the application um it's a good question and just a brief aside on the flash loan piece I would argue we actually kind of invented flash loans at dydx we didn't call them that at the time and Brandon and I our chief engineer and myself we were like oh this just seems like a cool thing we should obviously build but we're not going to take a fee on this or anything and we'll just enable them for the entire ecosystem because like why not we want to use it um and I can't I don't even know how much has been put through in flashland on dydx like a trillion dollars or something at this point in time so I think we were right

that these were G to actually be big but probably should have put some more uh potential value capture whatever around that um whatever we're we're moving on um so that was Flash loans and I think you really we really did think a lot about composability and I think that that flash L thing is an example of us building that is a first and foremost priority on what we used to build at dyx and I think the reason why we've moved away from that is we realized that it's hard and it really distracts you from what you're building with the enduser facing products not that it's impossible to do both but it's very hard um and I really believe and I think people

probably see this come through and a lot of the communication that I do publicly it's all about the product our product first and foremost and then it's all about the end user experience um and we just have as you say I agree kind of moved away from composability over time I think that's gone well for us um and I don't think this will always be the case for dydx but I think I see it as a sequencing thing where it's really important for us to get the main use case for the product to be great before we move out to enabling well or at least like encouraging are seeing as a major priority other people building other products on top of the protocol as well um and this is not to say that this is

the objectively right way to do it but I think this is the right way to do it for dydx and just with the DNA of our team I think again our users have really high expectations because they are more professional Traders so it's just it's really hard to build derivatives exchange like you got to think about all these things like performance like liquidity like oracles and everything else in between um and it's just the design space the complexity is harder than for just a spot exchange um so I think it warrants us spending a lot of time focusing on it and then going forwards we already have started to do this like one of the things that I'm excited about even with the dydx chain

is other people building either code directly into the dydx chain or things on top of it I think the thing I described before that skip is building is a great example of this I think the dydx token holders having literal more control of the exchange because okay now they don't have to have this centralized entity Implement some of the things that they want to do the whole thing is decentralized it's open source and they can just enact effectively any change to the active software that they want to um so I think things like that do move us more towards composability in the abstract sense um and then eventually we'll get to the point where we really are to the level of like a Unis swap or something like that being like Hey we're

unlap you can use unis.org but really it's about just building stuff on top of unlap and that being a really powerful thing so ton one thing I'd love to get a sense of um is you know you've been I think very uh vocal and and probably very forward-looking I would say amongst founders of I remember at a panel you were at our our Das conference must have been like a year or two ago but you've said it many other times which is like I wish Founders would stop looking at these ecosystems distribution and more look at what it can enable in terms of your product which I think you probably ahead of your time and that might still even be a somewhat um you know non-mainstream consensus take but I would love to get a sense of you know

when you think about what being an app chain enables you to do like more concretely uh as opposed to what you could do even as a as a layer on the layer two like the starware um I'd be curious like what those things are like what are sort of sort of some of the advantage of like for instance having each Val Val Ator being able to run their own order book and you mentioned I was actually going to ask you about oracles as well because that seems like it's a Hot Topic in ethereum land of using an oracle versus not or how do you use the Oracle and you kind of have this very neat answer uh to that um so I guess for folks who aren't as like in the weeds in terms of you know what you're thinking about building from a product standpoint and what were some of

like the needle moving decisions that you were like wow I can just build this and this solves a lot of my actual product level challenges being an app chain as opposed to being a a Dap can you just give us maybe like your top two or three in terms of what moves the needle for you yeah I think that's a good question and I think whenever you're building a really big new product it's worth articulating in very simple terms why are you doing this what's the one thing maybe the two or three things that you're trying to solve by building this new product like what's the problem on your existing product if you really feel the need to build something fundamentally new I touched on this before but just to clearly articulate it for our move from layer one to L2 it was performance that was it um it's like we

had to build something we was getting so destroyed by the gas fees the performance was so bad we had to make that change before others in the space did and we figured it out and the big change from V3 to V4 or the move from an L2 to an app chain is decentralization um and we had to solve this problem of how do we get rid of these Central components of what we are operating on the chain and how do we make the system as decentralized as possible both now and then have a good platform to improve decentralization over time um so with that in mind I touched on this before but I really do believe this is the most decentralized system that we could have built that's sufficiently performant as

well um and that's hard actually it's hard to not have to trade off against decentralization in performance and I think to some level you do but I think great software Engineers realize that everything is a trade-off but you can make you can kind of minimize the effect of tradeoffs over time or like what's the best trade-off that you can actually make um and we felt like we need to to as a goal at least maintain a similar level of performance to what exists on dydx V3 um and we articulated that in terms of the amount of orders placed per second that we saw on the platform and on dydx V3 was about 500 to 1,000 orders placed per second so again at least to start and we'll continue to improve this

over time but we need to have the dydx chain process a th000 orders per second ideally with low or no gas fees because like who wants to pay gas fees when you're a market maker and placing orders that's terrible um so I've said this before but just briefly I'll say it again the big thing that we did about two years ago when we started looking into building the dydx chain was take a look around and ask ourselves which blockchain or L2 or whatever can support a thousand operations per second and not just operations but these are pretty sophisticated operations actually it's not just like you're moving funds from one place to another you got to do margining checks you got to do all this other stuff that's pretty

computationally hard um so a thousand of those types of transactions per second again with ideally low or no gas fees and as close to instant finality as we could achieve and we came back with the answer none of them not even close um and you know maybe that's changing to some extent now but I I really do think that a thousand transactions per seconds is just where we're starting like if really our goal is as I've articulated it to become one of the biggest exchanges in crypto period centralized or decentralized we have to continue growing from here like you think binance is only doing a thousand transactions per second like no I guarantee you it's orders of magnitude more than that so what's kind of an architecture that we could build that will continue to scale

in that exponential way um so we took a look around the answer was none of them Cosmos included um and I won't go too much into this right now because I've talked about it elsewhere but we built this kind of differentiation between being able to place orders in an offchain way the order book is run offchain and then only the orders that actually match have to to go into the blockchain so that was the biggest reason behind the decision was how do we solve this problem of decentralizing the order book and the matching engine and then we kind of got a lot of cool stuff for not for free but at least in a better design space in a better way along the way um so we started with that and then we were like oh should we keep using chain link as we do on uid xv3 for

oracles or should we build our own Oracle system um it's actually not that hard to build this Oracle system or the hard part of building an oracle system is having this decentralized network of reporters and as I've said before I would argue that decentralization is a Min so already by definition we're meing our decentralization by the amount of decentralization of our validator set so one we just have exactly the same level of decentralization for the Oracle system um obviously there's there's some other considerations around like where are the sources coming from and stuff like that but we try to do an industry leading job with that too um so kind of got that without too much problem um we got this bigger design space for MV that we've been talking about as an advantage

and I think it really just does come back to you know I didn't honestly think a lot about oh me is going to be so much easier to solve um an app chain um and and that being one of the really big reasons that we decided to build the thing that we built but it does come back to okay let's pick the one thing like what are we trying to do here we're trying to make it more decentralized and that will have if you believe that that one thing that you're trying to do is really important that's going to have a lot of impacts that you maybe would not have even foreseen at the beginning if it is a really important thing to do hey everyone wanted to take a quick second to shout out this season's partner Maverick protocol now many of you probably know Maverick is an in

Innovative amm which they are but in reality they're a lot more than that as well Maverick is a suite of tools for def users and builders that allows them to put liquidity where it will get the most work done since Maverick launched in March theyve been gobbling up market share and at the time of this recording which is the end of September on a trailing 7-Day volume basis Maverick is now a top three decks by volume and they support over 50% of the volume on the L2 ZK sync a chain Maverick enables LPS and token pairs to process higher volume with limited tvl which allows them to support some of the highest levels of capital efficiency for lsts like WAP Ste another very cool feature is something called Maverick boosted positions so

that allows protocols looking to bootstrap their token liquidity to Target the shape of liquidity of any token pair with surgical Precision Maverick is backed by some of the leading institutions in crypto Founders fund panta coinbase Ventures Finance labs they are all backing Maverick in their Vision to revolutionize the next next generation of defi daps and helping them build their liquidity in all market conditions click the link at the bottom of this episode let them know that I sent you thanks guys I want to get to discussing uh perss here as well um it's something that we haven't really covered in in the season thus far but the maybe the last question on on this subject is how do you feel about you know what absolutely should be onchain to ensure decentral like whatever your sort of

threshold is for decentralization that you feel like is necessary versus which components are you sort of comfortable moving offchain which it seems like a lot of dexs are starting to experiment with that particular lever as well so be curious to get your thoughts I think it really matters what you mean by offchain onchain I think is pretty well understood people think about that a lot it's like okay this is blockchain it's limited by the decentralization of the validator minor set itself consensus whatever but offchain could kind of be anything um and I think a lot of people when they hear offchain immediately think of centralized which is an option but it doesn't have to be or potentially there could be another network of offchain participants like a keeper Network or something like that that does some job together but then again you

have to really think about what's the level of decentralization of this offchain thing um and as long as that's similar to or greater than the decentralization of your base chain amazing you're not sacrificing any decentralization when you're moving something from onchain to offchain so that's really the way that I think about it um and I do really think a lot about decentralization at each step in the chain as you've heard me talk about um but it matters what you're doing right like potentially something like what are literally the balances of each particular user and how are they being Modified by each particular trade that's really important right like obviously that's the key thing that has to be

super decentralized like probably that has to be more decentralized than like okay I place a trade and maybe some validator like censors me whatever like great that's bad like we want to solve that too but do we need to solve that with as much urgency as some validator can literally change your balances by by themsel or with some non- decentralized network doing that like that's clearly terrible like we can't have that happen um so I think it's worth thinking about from like a product perspective being honest about what level of decentralization does each particular point in the transaction and Abstract abstract sense chain need to have um and then using that to inform your decision making for the tradeoffs that

you're going to make between performance decentralization product whatever else I think one other thing that I think is is really decentralization can have different effects on is between spot dexes and perss I think it's generally a lot considered a lot harder to do a a secure build a secure per needless to say in part because you know with spot you can ensure non-custodial of of uh everyone's assets but with thep ultimately you're subject to to all the risk controls of the of the system um so how decentralized do you think perss can get well I think the sky is the limit for all this stuff right and that's one of the really exciting things about blockchain is all of the stuff is now

technologically limited rather than limited by humans and we can go a lot farther with technology and I think it'll continue to increase exponentially at least for the time being or the next five or 10 years or so so I think they can get very decentralized I mean I think the state-ofthe-art is is decent right now um and I think it's I mean I think we all believe this but it's worth noting that a lot of these like decentralized per systems have been more or less as robust as a lot of the centralized ones at least for the leading ones um or at least within range of that it's a little easier because there's less open interest all this sort of stuff on dexes than and Sexes especially for derivatives um but we're growing and it's significant right now

so I think it's just something where the level of decentralization or safety and risk is good right now but is a major thing that we are going to continue to improve on the systems over time I think from a financial perspective it's pretty similar between centralized and decentralized exchanges the main difference is you have to really deeply understand each like the way each participant's Network operates and how that's going to be different from a centralized exchange um like let's give a specific example on a centralized exchange obviously the operator of the exchange the matching engine is the one doing all of the liquidations and on the dydx Chain now there is no Central

entity so we have to rely on all of the validators to do the liquidations what are considerations around this okay well as long as we set up the incentives correctly we can make this into a really good system but we have to be honest like uh there's incentives for once an account becomes liquid liid atable uh that the first validator actually liquidates this account but you know what if they are offline or what if they just decide not to or there's some weird incentive that we didn't think about that maybe this particular validator doesn't want to liquidate this accounts or whatever and then maybe you have to wait for two three four five validators and the block the next block spaces to actually do that liquidation well okay

it's probably going to happen a little bit slower than the liquidation if it's on binance and that's at kind of the millisecond level and we have to set risk controls appropriately now can we improve that over time yes definitely with things like better mechanism designs shorter block times all all the rest of this sort of stuff um and like we we have to think about like the question okay is doing a Liquidation in 10 milliseconds really that different from a risk control perspective uh from doing it in five to 10 seconds with really high confidence maybe but we actually have to look at this statistically and that's something our research team has done done like what if we had a 5 to 10 second lag on all of our liquidations with at least with the data we have on dydx V3 right now and

that historical Market data like would the Insurance Fund have lost money um and we came back with the answer no um so and again like we can never exactly know what the volatility of markets are going to be or if there are certain attacks and stuff like that but the best we can do is approach this from uh good research risk statistical basis and understand the technology deeply and that's something we think about nice just out of curiosity what are the block times on DX it's about a second right now I think interesting um I don't want to go di that could be a whole other rabbit hole but um I I want to ask you a little bit about just how do you see the market share in between uh perp shaking out with spot I suppose on the decentralized exchange front and then

one of the other questions that uh Dan and I have asked a little bit this season is the the popularity of perss as opposed to something like option options um which options if you look at the tradis and analog extremely extremely popular product and they just haven't really taken off in crypto for whatever reason so maybe there are path dependency sort of issues there or maybe for whatever reason it's just a better Construction in crypto specifically but yeah I would love to just get your thought on you know purp versus spot as market share and then perp versus options as a product okay well obviously I'm kind of biased on the first one but I will make my points as I believe it um uh clearly I think that perp especially in the Dex space are going to become

really significant over time especially from a relative basis where spot dexes and per dexes are starting right now literally if we just look at the market share of dexes within pers versus spot it's way less in perss and this is the main metric that we look at at dydx to decide if we're successful or not um is what is our market share of all perpetuals or more broadly derivatives trading that happens within crypto to give people a sense that's about one to one and a half% right now pretty good pretty significant and this is a massive massive Market with huge companies like binance um now coinbase others playing within it um and we're by far the largest decks that's playing in this space but there's a lot of room to grow right like it just doesn't make sense to

me that forever crypto is going to be traded on centralized exchanges the really hard problems to solve this we've started to discuss but like is the market share of derivatives trading for Dex say always going to be like two % we're like one one and a half perent and then you have more Mark more volume than all the other per dexes combined more or less so the entire deck Market space for purposes about 2% um I actually don't remember or don't have off the top of my head uh what kind of the the spots deck market share is but I think it's a lot higher probably like an order of magnitude higher at least um and I think exchanges like Unis swap have really started to take out more of like a Dan you can correct me around 15% yeah

that's kind of thinking like 10 to 20% of like the entire spot market share so again like in order of magnitude more than what dydx has in the per space right now so even if we just look at those two things like it just makes logical sense to me that that should grow that being said there's a lot of really really hard problems to solve as we started to say within per specifically and it's just harder than building spot exchanges especially from a technical perspective so um it does make sense to me that it's taking longer than than spot um but we as kind of the market leaders and Dex curs have to be the on to take ownership of that and really push things forward which I think we try to do um and then the second question on why pers over options I

think it's just that the market structure of crypto is really fundamentally different than traditional finance and I've said this for years but my main point continues to be that in crypto really the thing the one thing that that crypto has enabled that has given it product Market fit on a global scale is enabling retail traders to trade and have access to more professional products than they would otherwise have access to like that's it that's like the M like you could attribute almost everything that's happened within crypto so far to this main thesis and that's really fundamentally different from traditional Finance where now there's things like Robin Hood and Retail is starting to come in in a bigger way but traditional Finance was first and foremost built to

serve institutions crypto is first and foremost built to serve retail um and this is really fundamentally different doesn't mean that institutions aren't in crypto but it means that if you are an exchange you have to build products that retail Traders want to use um and that's the entire game and sure the institutions will come in they will actually be most of the volume on your exchange like on dydx I think 85 to 90% of our volume happens from institutions and it's like 10 to 15% that happens from retail but I really just see institutions as a multiplier effect on the retail trading volume that exists so if we look at this from a financial design perspective on what derivatives

is the space going to be most likely to trade probably going to be a derivative that's different it's going to be a derivative that's more targeted at retail and I'm not talking about it's your friend that bought their first Bitcoin yesterday when I see retail here I'm really just talking about sophisticated individual Traders in jurisdictions where they have access to these types of products um but for those Traders they really want something that does solve their needs and their need if you're a sophisticated Trader trading derivatives is effectively you want to trade on Leverage um and you want to do that in as simple a way as possible it was really interesting actually we had Arthur Hayes in for a fireside chat at dydx a couple months ago and I asked him this very question like how did bitmax

come to do per rather than options and he was like actually we started with options not that many people know this about bitmex and they built this whole thing out and and everybody kept he Arthur was doing support himself and everybody kept writing in and asking him uh like what is xere like I don't understand this thing I just want to trade on Leverage like let me just trade on Leverage simply and he kept answering and it's like no this's like the xery here's how you price options here's how you think about them this is well understood people don't care they just want a product that is simple to use and that's how he effectively came up with the per product for bitmex and I think that's really Ive as to why um pers rather than options within crypto and I

used to think that this was just a step in the evolution of derivatives in crypto and I think probably to some extent it is I do think options will gain more market share over time but I think it's going to take way longer than people think like way longer um and and this has kind of been empirically proven for the past few years um but perss are really here to stay there's a bunch of liquidity behind it there's a bunch of retail interest behind it and these products are now better understood at least for our Target users than a lot of the options dated Futures and stuff that exists in traditional Finance yeah um that I honestly there's a whole other conversation that we could have about that actually perss were a a construction that existed pre- crypto a

lot of people don't don't know that there's a great episode that hosu did with Sue about that on on Common Core but uh unfortunately this is all the time we have Antonio this has been a ton of fun uh if folks want to find out more about you follow the work that you're doing at dydx what's the best way to do that so we are dydx on Twitter dx. exchange on the web and I am Antonio M Juliano if you like me in particular thank you very much for having me this is a good conversation ton of funks lot for coming on thanks for coming on all right all right Dan that was a that was a great episode I feel like we could have gone for double the length of time there um I think yeah I think the only the only problem with it was I think um Antonio

is so good at explaining things and has such a clear um kind of style that it was it was uh hard to get too back and forth and and um uh into too much discussion with him because I think he just had a lot of a lot of really interesting clear things to say yeah I agree I I wanted to start with this this idea that he brought up about the almost like social slashing because this was this was an interesting proposal it got uh it got a lot of attention I would say in both the ethereum and Cosmos communities the Mev communities I think you you pointed out sort of the the ethereum perspective which is that if if I had to summarize it and be pretty reductive about it it's like hey guys this kind of feels like cheating here you know there's there's

no mechanism design here there's nothing on chain this is just saying like Hey we're gonna slash the people who are are bad and I feel like this is um you know we unders squid it a little bit during the during the episode but it's this is indicative of a difference uh in between some of the solutions I've seen tried in Cosmos versus ethereum but I I love maybe you say a little bit more about how you how you view that yeah well so first of all I think it I I so I think it's I think it's very interesting and maybe gets a bit of a bad rap in the ethereum community um or around some people whove reacted to it I think so one thing and I noted this on the on the show is I think any mechanism um in crypto almost Mo most mechanisms for

consensus or for economics for a lot of these things depend on some form of of these Norms or or social sanction or or weak subjectivity or or uh call it what you will um and I think we see this in ethereum most clearly when you talk about something like reorgs for example so it's accepted socially in ethereum that it's okay now to extract most kinds of Mev from a block that you are the proposer for um but it's generally not considered not acceptable to reor other people's blocks um as proposer in order to extract to extract more Mev and then there's there's areas in between that are that are you know sort of where people are learning the um the Norms around around like oh should we allow someone to propose a block a little late um or these or these other things but

ultimately I think those are enforced by social norms and we as a community made something of a decision that actually for various reasons we going to accept single block um meth extraction uh uh for and for various reasons many of which I sort of I tend to agree with or or think Mak sense but I'm not sure it's the only choice that you can make and I think saying we're not going to accept it actually does make a big difference it means you know for example um you know people people sort of mock it but it means like if if uh if you didn't have this kind of Norm maybe just dydx valid is would be like I'm the most me extract you so so delegate to me and I'll give you kickbacks from the meev that I extract right like you would actually get you you you're able to do a

lot more if it's socially if it's socially allowed and if you if you um have to hide it it actually does limit what you can do um now I think again clever validators often would be able uh will be able I think to extract Mev in subtler ways um and possibly in the long run they can you know it's this kind of mouse game they can extract all the same kinds of Mev but um but I think there there are other components to their solution which includes like this dashboard and actually measuring measuring metrics of it that I think you know look it's it's at least if it is a cat and mouse game you know so there there's um it's at least you know you know they're they're trying to be the cat for at least a second and then of course they are taking um ultimately uh

they're they're working toward working toward trying to design other systems um to to either supplement or replace or replace the social sanction But ultimately I think you are going to depend on some kind of social norms whether it's around you know not colluding with other other people whether it's around just not destabilizing the entire consensus um uh you go any any system ultimately I think will fall back on some of those yeah you know the reorg example is a really good one there was a there's an old episode of zero knowledge where um it was a dean Igman his partner at project Blanc but he used to be in Flash Bots I think his name is Edgar I'm I'll look it up but uh yeah and he was describing how he sort of embarked on this uh this project that basically would have been extracting Mev

through reorgs and he said the thing that got him to not do it was a DM from balik you know so um there is there are social pressures everywhere and I tend to agree with you I actually don't think it's such a negative thing because at the end of the day You' you'd be shocked about how much just everyday average life in every individual country is enforced by not necessarily the law but uh social norms that we all abide by and one of the the reason that this just makes me smile because it's such a funny difference between uh I think the ethereum and Cosmos communities is again it's because I think Cosmos doesn't have this pressure of securing the $250 billion that ethereum does but ethereum tends to be much more research oriented and uh Cosmos sometimes will just be like let's just try this and you

actually saw this you know the the example that I love to point to is the Lio well what's the right amount of Stak for one staking pool operator to have is it 20% is it 15% let's do some studies and in over in Cosmos land they implemented something called liquid staking module which is a form of Regulation essentially for the Hub and they're like let's just drive like 20 per. yeah feels feels about right it's just funny um I it's right it has I think it can have downsides but I think it also um yeah like often you end up it's interesting because often in you end up with a solution that actually isn't isn't very well researched because like we didn't expect this to happen all the research was around some other Edge case and this is and then this is the one that happened um and so like unless

you have a really like Ironclad proof that your system works which again in in crypto only only a few things have that um yeah I think sometimes you sort of got to try it and then and then patch what you see rather than um trying to anticipate like sort of or plan everything out theoretically ahead of time I agree I also think that's why it's actually although people myself included like bone or push back against tribalism it's actually not a bad way to organize people to solve disperate problems like there's some amount of group think that happens in every crypto community and there are definitely cons to that but the pro is that everyone is like laser focused on solving one thing right or sort of in the same way of thinking and actually having these different communities with different perspectives and ways of approaching

problems is actually not a bad way of solving a really large uh related set of problems at scale um would be my I that's my sort of thought okay but uh one of the things that I appreciated about Antonio that I'd love to get your your take on is he I was thinking about this as we were recording he he's very thoughtful about treating decentralizations as a means instead of an end so what I mean by that is like you often hear in crypto like I'm a you know decentralization maxxine it's like why well why are you decentralized because if you ask why do I need to be decentralized you have a good understanding of how decentralized I need to be so that I can optimize for giving a good product and trading it off against decentralization so you know for something like ethereum and Bitcoin it

was the explicit use cases censorship resistance whatever uh for a money you want to be basically just as decentralized as you possibly can but I feel like Antonio has been at the Forefront of pushing for well actually for an exchange maybe I don't need to be as decentralized as money itself um and instead I'm G to ask myself for what exact purposes do I need to be decentralized and I thought that's where a lot of the really in interesting insights came out around um you know like validators running their own order book and being able to act as oracles very interesting do you need a thousand of those like probably not I thought that was a very interesting he clearly thought quite a bit about why he needed decentralization and to what degree I thought that was interesting oh yeah yeah I think it's a theme we've talked

about a bit in this in this season but like what do what do we actually want of decentralization and I think there are different different layers to it in different kinds of things you want for me like one of the one of the really hard lines is the the ability at least to trade on a on a system um non-custodial the ability at least for me if I if I if I choose to to um to trade participate in system and in a system and trade with it without without having to have another party custody my funds is really important but that doesn't necessarily mean that everybody has to be it doesn't necessarily mean that things like order books um can't be can't be more centralized and I think if you have more centralized order books sometimes that can mean that can mean worse outcomes for for example trade uh

price execution but sometimes it can mean better outcomes um sometimes like putting some partial trust in in some entities can mean that in most cases you actually end up um with better execution I think I think you know being a realist about that is I think Antonio Antonio historically has has been um and and thinking like okay how are we actually going to slice this what are the what are the components what roles is decentralization play in our system I think is really important for for Designing any of these rather than yeah exactly like at every at every step being like well it must be better to be more decentralized um at this at this step or like I agree or or what are we even doing here and I'm like well you know I think actually let's let's figure out what we could figure out what we're

doing here um and then figure out the best way to do it maybe you know what that I love that you said that because so so often it's like well maybe we don't need to be decentralized and then people almost like their hackles go up and it's like what's the point of even doing this it's like there's a lot of Shades of Gray in between complete and utter decentralization and one blackbox server right Spectrum tradeoffs yeah you know yeah um but that you often do get a very emotional response oh one thing um I actually wanted to ask you is actually the question that we LED with so you and I have been approaching this all season from we are going to optimize actually the experience of both LPS and swappers to uh you know by by redirecting some of the Meb that gets redirected to

proposers today um and that is very different uh in an app chain system where it's a little bit more of a closed loop or the stakeholders are sort of more are very different it was also interesting to hear him say I really care about the swappers uh more so than anything else that was interesting that's right so so I think one thing one thing worth pointing out which we didn't even harp on in the in the call but just by doing an app chain automatically you're capturing a lot more Mev for the application um and and token holders than you are you know as opposed to what's leaking in um in a system like ethereum I mean even if suppose suppose every is just maximally capturing Mev and there's just a just a ruthless race to the bottom on that ultimately that means that's that's not just going to go

to the proposers it's actually going to end up you'd expect um or a lot of it at least you'd expect to go to to to token to stakers because whoever can extract the most Mev is able to then provide the best rewards to uh the stakers and they'll get the most people most stake delegated to them and so you are you do have then even in this like maximally extractive like like you know sort of one of the worst ways you could may I would maybe think of you could design the system you already are getting like Mev extraction for in some sense the protocol or for or for token um for token holders although again I think there's some there's some areas where that leaks out so I think that's important that like actually you do get some benefits out of the gate but we

don't want to you know um uh I think that that's not actually everything that we want one of the things we might want is that may just not be extracted from users in part because I think if you if you give users a bad experience then maybe they just won't use your application and um and then you actually end up with a lot of dead weight loss like the application might be better off if it was providing better better execution for for users and then I think there's there's just like what do what are yeah what are the goals of the system and I think generally yeah I've sort of talked about I think with um with dexas I tend to agree with um with Antonio that what matters actually the the constituency being served is the swapper and we should just be maximally um uh benefiting them and everything

else is just Downstream from that including what we do for liquidity providers and stakers and and all these other parties they're all there because some somebody wants to show up and get a good price by trading yeah it's a really good point I um oh Sho where was it oh one one interesting thing I just on uh fees you know he he was obviously very laser focused on fees when we talked about what what made you make the switch from starkware and why Cosmos and you know obviously uh he's dependent on like high frequency Traders it's just his market makers and it actually made me think of this um this post from Dave OHA blanking he has a co-author so I I apologize but I uh he on on fee credits this idea of fee credits um and the the the the key Insight of this paper is that fees

people people describe them as a form of Revenue um but but really the the purpose of fees as they originally con see it's a form of a Cil resistance right it's like you want to prevent spam from just uh you know pulsating through the network and the the Insight that that David and his co-author had was well actually you know instead of just charging people that's a very crude form of cival resistance maybe you could do something like if you are proving that you're a good actor for the network by staking then maybe you could just not charge fees so I I do think there are even like kind of new interesting ideas in the world of app chains where you could reduce and there's there's like gasless fees and stuff I know going on on ethereum we talked to um um von Beethoven about that too but you know I

just there there're even cool I just didn't we didn't have enough time to get into it so I didn't want to go down that that rabbit hole but I thought that been I do think it's and and this is part of what makes my colleague Charlie um has historically made him excited about appchance is I do think there's a lot more that you can do um in terms of setting the rules in order to actually make make me exraction more instead of compatible um or limit or limit toxic mag extraction in a general purpose chain it may just like there are sort of it may just be impossible to actually limit these things um uh or or much much harder and when you're cohabitating with a lot of other applications the problem just get much harder but when you're working on an app chain um yeah I think

I think there actually are there's a lot you can do and a batch auction is one example where you're sort of having this seep into the execution layer of how we execute these trades um this rule at the consensus layer in a in a way that potentially yeah potentially makes both more secure and and ends up with better prices for users so I think yeah I think application uh app chains and applic application specific rollups as well which we haven't seen as many of in this uh recently for for whatever reason is in like app specific nonm um rollups but I think we might we might start to see um more of at some point I think uh yeah I think you can you can actually do a lot of other rules that make that maybe make mic action a little more aligned yeah I I think there's a this is

actually a debate I was having earlier today but with with one of our analysts um but Anton originally moved there was the fee reason but then there were things that you could do in an app chain that you couldn't necessarily do having your validators serve as an Oracle and and hosting an order book is kind of the primary one that he talks a lot about with reaking you actually sort of open yourself up and there might be the opportunity to do that right like you could actually opt into this actively validated service network of people that would essentially act as your um set of validators as a as a rollup on eth I guess and and you can actually look um like reaking has a bunch of different use cases Ian da is sort of the first wedge use case that they've tried to do

but they you they also have like um you know partial block building as a as an application that reaking highlights and you know you you could actually see probably in theory a lot of what we talked about today with Antonio could be built on a lay or two I guess the question is would you and how practical is it I'm just not in the weeds enough to be able to answer that question yeah right and what benefit you get from an app chain um that you don't really get from reaking is this this alignment of token holders and you know stakers ultimately actually being um subject to you like to this to this token that is specifically tied to success of your app right so one one thing and again I think it's you could still end up with a bad equilibrium where everybody extracts

until they kill the chain um but in some sense like everybody you know stakers on dydx don't want dydx the the the um application to to fail to be to be this like degenerate you know um thing that that that ends up ends up killing off its use they're more long-term aligned because because ultimately they're holding dydx token um or and or running validated or something like that and so in some sense that it's it's a way to basically tie everybody together here whereas sakers of of eth may not actually care about as much about the success of uh of one individual application so I think that is an interesting thing we talked about it my colleague Charlie and I talked about it a few years ago in a post who wrote called the cosmos thesis um but it is it is interesting potentially to to try to

align people using using individual staking token yeah yeah I think so I that I I think that alignment is and he actually called it out he said there's the there's a fourth um fourth stakeholder as well which is dox's shame just yeah I'm a huge nerd for Cosmos and I I love uh it's always nice to get people from the cosmos ecosystem invol because they have very different um some orthogonal to eth way of thinking um one one uh sort of if if you're a huge nerd for this I we'll actually link it in the show notes but I don't want to go too deep into this but the just the link with with Pepsi um for for people want to figure it out barnabe who is the the kind of uh author of The Pepsi idea over in ethereum which stands for protocol enforced proposer commitments uh it's

actually very similar to to ABCI Plus+ and it I think there's um if if you squint at it and obviously Antonio is trying something more directly with this it's a way of Hosting censorship resistant auctions um in a way where you're not questioning like who is the auctioneer you know risking auction or exceptable value so that's something cool we don't need to get into the weeds but we'll link it in the show notes here so you can check it out because he actually highlights how these ideas are similar um last question for for you Dan I think we talked about this in our opening episode of the season but you know it was interesting to get his thoughts on the part Market that was a cool little anecdote about Arthur Hayes actually trying to start out with options which makes a lot of sense now

that he just said it like yeah obviously you want to start out with something like options instead of some weird derivative that had basically been theorized but never took off in trafi but he was clearly and I'm kind of I'm kind of on this page too that perss look like they're going to be the winning derivative instrument at least for the time being what did you think about his thoughts about just relative spot versus per Market share and what what do you think about all that yeah well I think I think I I agree I basically agree with what he said I think I there's a there that there's a lot of room for decentralized perss to grow um relative to relative to centralized ones and I think you know some of I I'm not I'm not sure that there are um I I think in some

ways decentralized pers it might even be more important um relative decentralized because in part per markets as we've seen historically can be can be subject to a lot of abuse um and can be and can be very risky and I think there some a a a properly designed and fully transparent onch and decentralized uh uh PR you think should be actually a lot more auditable a lot more for for um uh for ordinary people a lot more um or at least for yeah at least for highly technical people um a lot more yeah a lot more transparent um and ultimately yeah potentially safer I think in the long run it should be it should be safer I'm not sure we're at that point um today although again depends on depends on the centralized prep that you're comparing it to um I think one more interesting one that I think is uh uh uh

is worth noting is I think longtail perss um you when when you're trading like uh uh crypto assets I think like you know spot markets versus versus per markets you know if you're on chain like one nice thing about trading spot is like you're literally trading the thing itself like it's it's just as close to the metal as you can get but um one thing onchain onchain perss might be able to unlock is actually like just trading on this like really long tail of of you know of things what we would normally considered to be prediction markets and I think prediction markets have been having an interesting moment we haven't talked about them much this season um but uh you know I think we saw we've seen some centralized per exchange sort of uh uh uh play with okay like can I offer something that's a little more

like Trump Futures right uh prediction markets on that I think you could see decentralized ones do it as well um and they might be a natural place to actually just trying okay like can we actually can we get derivatives on like there's many more things that you could do a derivative on like literally any number than there are actually just assets in the world and so this longtail um uh very very longtail I think crypto's been historically good at um serving this this extremely longtail crypto Solutions have been and I think you could see yeah you could see per serving that as well yeah I you know what else crypto does well it's like the best and worst thing about it right for when it comes to finance it moves quickly which is very uh that's not how it works in trafi for good reason a lot

of the time it turns out but you also do get a lot of very interesting experimentation like one of the things I wanted to ask Antonio was you know if you look at something like Avo they have uh pre-t toen Futures that have launched so you can actually trade something before it actually trades is there a ton of value to that I don't really know but yeah I don't know it's kind of scary um but uh like how do you set your mark what do you what you know there's no there's no sort of spot price to to reference so it's like an entirely circular um construct right but I think yeah certainly an interest interesting concept yeah uh and all right last last question for you here because Antonio I feel like has come out he this is I think a very differentiating thing with

Antonio as a Founder as opposed to many other founders in the space and ch's come out and said like look I don't want to be judged based on the ecosystem it should only be about the product and honestly putting my own Founder Hat on I everyone admires that right like yes it should be about the product should be about the users but then I put my own Founder Hat on it's like well yes but at the same time you know eth has all of these users right and you can just clearly see it based on volumes uh and the amount of assets that have migrated up to rollups like yeah you got a little bit of a leg up from a distribution standpoint and the counter to Antonio's perspective is first time Founders think about product second time Founders think about distribution so I'd be curious

like if you're just putting on your own you know you were starting a deck uh taking some of that research that you've done at Unis swap putting it to work you're like I'm I'm launching a Robinson exchange here how would you think about doing that would it be all about the product would you be trying to go to where the users are how would you balance that I think you I think you certainly want to go with Jason to where the users are um one sort of uh thing I've thought about about Cosmos for a little bit is that ultimately what's the what's the cosmos what's the cosmos Hub really like the cosmos Hub is actually ethereum possibly like most you know to some extent a lot of a lot of Cosmos projects have become successful by by largely by having functioning Bridges through and from ethereum um and by like

you know by by sort of centering themselves relatively close to the ethereum community the assets that are in the ethereum ecosystem um rather than trying to be Central to to um sort of Cosmos and so to the extent that you have actually um you know I think dydx very much um my understanding is like there you know there's the bridge obviously from um from ethereum there there's also um I think usdc for for onbody on to it like it's it's not designed I think to be that much harder to get into it you can you know you can use like metamask and everything it's not it's designed not to be much harder than to get into it then you can than it is to get into an L2 right and so there something like what is what makes something an L2 rather than um a cosmos

chain like some of it is some of it is a security model some of it is um is sort of is The Vibes and The Branding but some of it is just accessibility and and bridging and I think that's you know having I if you have a a cosmos chain that actually has a very it's very easy to get on into and out of ethereum then I think it may actually not matter that it's oh it's technically it's technically in the cosmos ecosystem okay but from a user experience perspective it's it's um and in terms of like being able to attract users it may actually not be that different you know what I think you're I completely agree with that and actually one thing I was thinking about was that it's almost like ethereum and Cosmos clearly have such a similar road map um there are a couple

key differences like the Hub has token voting as opposed to ethereum which doesn't have so there's some key differences but there's a path dependence that ethereum started with the asset now they're trying to go to this multi-chain future of a bunch of different now the benefit of that first of all obviously that was at least in the short term the right decision they have many more users much more liquidity um and so that's their starting point the the disadvantage of that is that you actually have a bunch of fragmented layer twos and ethereum users don't have this understanding that that's all supposed to be one thing and so it's like I go on arbitrum or I go on optimism and it's a the user experien is you start on eth and then you kind of go here or you go up and down and that's

how you learn as a user but on Cosmos it's very different everyone in Cosmos understands there's the shared vision of this interoperable network of chains and actually it makes the user experience at least from my expect perspective today very challenging but every chain understands that it's supposed to be this one chain amidst this interoperable network of other chains and it actually makes them weirdly even though they fight a lot that that was the understanding from the beginning whereas a lot of the l2s are like I think if you ask them privately they'd be like I want to be the only L2 or like one of two you know like it's just a different um perspective and standpoint and yeah I they're they're kind of building the same thing in Reverse with admittedly with some minor some differences but yeah um yeah I think I think a lot of

the time like these sort of religious differences in in approach like oh you know we do it one way and they do it another way are actually just like right now we're making this strategic or tactical decision and then like ultimately oh it turns out actually we could just do that too um or we want to do that too and so ultimately you know I think I think a lot of and the the projects that survive tend to be the ones that don't get too much we talked about this in the Dex context too that don't get too much religion around a particular type of solution yeah my personal theory for why it tends to feel that religious is because you're often trying to decide between really technical decisions that no one understands so you have to wrap narratives about like good bad what do

yeah what do you think like the original stuff that there were schisms about in the um in like the the the Catholic and Protestant churches right it's like it's off to these incredibly incredibly esoteric religious debates and then you sort of manage to manage to create this massive cultural around these gotta win hearts and Minds baby uh it's like that's the only way you're gonna get everybody to agree to your definition of the number of angels that can fit on a pin or whatever right is you sort of have to make it make it like core to your cultural identity that you're that you believe in this one rather than another one but ultimately yeah I think you you're going to be led you're ultimately not going to make a hopefully I think as a project not going to make a bad decision based on like oh we we

previously you know we we rallied ourselves around this but actually we we decided to make this difference this different change yeah I agree with that um all right Dan well we can we can end it here uh just on the subject of Catholicism that feels appropriate and uh yeah this was a really fun one and I will I'll see you next [Music] week