Shining A Light On MEV | Tarun Chitra, Justin Drake

Transcript

all right everyone welcome back to another episode of bell 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 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 before we get into it today just want to give a quick shout out to this season's sponsor Rook close to a billion dollars worth of Mev has been taken out of users pockets and that's just on ethereum and that number is only getting larger unfortunately Rook thinks that it's time for a change and they've

built a solution which is going to automatically redirect that Mev back to where it belongs into your the user's Pockets so you're going to be hearing all about them later in the show I'm a huge fan of this team and what they're building so stay tuned to find out more alright hasu episode one here I think uh you know that you and I intro we'll call that episode zero today we're going to be talking to tarun chidra and Justin Drake uh very very excited for this one you want to talk a little bit about why we're having these two folks on I mean so I just like starting from the very high level and then just branching out from there and and we can and kind of zoom in over time and I think after this episode zero that we did I think tarun and Justin uh two great guests because

they both I would classify them both as some of the most holistic thinkers in in crypto they they both are working on many different things so Justin knows everything about the ethereum protocol um not just that he's he's also now operating a relay uh the ultrasound relay that's making um important uh contributions to the mevs to their Meb boost supply chain and tarun is one of the leading auction experts and kind of academic contributors on the topic of Mev and so I I hope that we get a really balanced perspective on the big questions of Mev from them means well I thought it was a it was a great conversation and that was

a perfect lead in so without any further Ado let's just uh let's get right into it all right everyone welcome back to another episode of bell curve today I'm joined by my Fearless co-host hasu and today we've got Justin Drake and tarun chitron with us welcome everyone thanks for having us hey yeah welcome guys this is um this is going to be a ton of fun guys I I I'd love to uh Justin and Turin while we kind of have your brains here uh kind of work through a bunch of the issues uh in Mev today or kind of touch on a lot of the topics that that haasu and I approached in the kind of kickoff episode of this season um but to start with you know I think with something as as complicated uh as Mev you almost can't go too high level

I'm sure we'll get deep in the weeds later in this episode but I actually want to start with just asking you a question that we kind of started by asking hasu in the first episode which is what is your sort of definition of Mev so I have a very simple definition which is that um when you consider an economic system and its economic activity on top of this system it's basically the value that the system can extract from the activity it's just that's that that simple now there's a few like things to to mention here one is that oftentimes the term Mev is used in the context of blockchain so blockchain is the economic system and there's various agents at play you know this could be

the proposer could be the The Searchers the Builder Etc subtlety is around the the word e so e stands for extractable but oftentimes people you mean mean extracted and it's kind of unfortunate that these two words start with E and so people sometimes use the word realize so there's realized value which is the extracted value and then extractable value EV the value that could be extracted but may not have been perfectly extracted and then the m is just the idea of uh you know a maximization problem where you have some well-defined economic activity as input for example some set

of ethereum transactions and you want to do something with with that for example build a block by ordering them in such a way that you you build you extract the most nav um I would have a follow-up question to your answer Justin which is he said that you think Mev is the value that the system can extract from the activity um so going by that analogy wouldn't Mev rather be value that some external actors can extract uh from the system doing the activity so it's almost like somebody undermining or eroding the guarantees that the system is providing and and kind of making them worse for users right so

um this is this is an alignment problem the question is like where does the Mev um go and really you want it to go in the most aligned ways and you wanted to to give it to the people who are maybe the closest the most proximate to the system so in the case of ethereum the blockchain I'd argue that the most proximate are the ethereum holders and so if you can direct them if you there that's that's the the optimal outcome maybe the next most proximate would be proposes and this is why the Mev is currently going to and then after that there's agents like relays and Builders and searches and as you progress you become further and further away and you could argue that maybe you know there's

a misalignment there and it's it's leaching value you um but you know I'm I'm confident that in the case of ethereum specifically we can design Mev in such a way that most of it really goes to the core of of ethereum um and uh and doesn't get leached out uh Justin you you just dig in a little bit more to what you were just describing there I I caught your recent talk where you kind of outlined this idea of Mev precedence where you explicitly sort of outlined this this stack of where Med should go can you dig into that a little bit yeah absolutely so um basically you have this value which can be extracted and the question is

where does it go and you know around this question of you know alignments you basically want to come up with a precedence list of who is the most aligned and who's the least aligned and really at the top of the list arguably is user generated me so when a user makes a transaction that generates Mev really the ideal outcome is that they get a rebate a refund and it goes to them but then one of the things you understand is that not all any V is user generated and so the next best thing after that is to take so-called latent Mev which is just available from Arbitrage for example and and you know ideally give it to uh to

someone very aligned to ethereum it could be a proposal it could be an if holder um and yeah what you know it's a little bit subjective in terms of how do you define alignment so if everyone might have their own you know ideal precedence list um but uh you know in in my opinion the way that I think about it is that you have this Mev pipeline starting from the user um and ending with the protocol and it goes through the the wallet the interface the Searchers the builders the the relay and then the protocol and really other than the user that that's at the very beginning that should

receive a rebate for the activity that they generate on the other end kind of economic energy that is latent to the system should be should go to to to to the system itself and the basically you can just walk back this Pipeline and and rebuild the Precedence list from that so the holder should have precedence over the proposal that has presidents over the the Builder that has presidents over the relay that has presidents over over the search something like that it's still you know this idea of Precedence is very new to me and you know we can try and and come up with a methodology for building This president's list but and maybe one kind

of heuristic is looking at this Mev pipeline which starts you know with uh with the user then you go through the interface the The Searchers the builders the relays the proposers and maybe in the case of Mev which is latent to the system which really originates from the existence of ethereum itself you kind of want to start with the end Which is closest to ethereum the holders and then walk back so the the holders should have precedence over the proposals the idea of the Precedence list is a little bit subjective um but in my opinion um what you could try and do as a heuristic to build this precedence list is to start

um with the Mev pipeline so you have the users and the the interface the Searchers the builders the relay the proposal and the holder and you kind of want to start from one end of the Emu of this Mev pipeline specifically the end Which is closest to ethereum and the reason you want to start there is because the existence of ethereum itself is what leads to this latent Mev to be there and so it it it's kind of best if the holders grab it first and they have precedence over the proposals that have precedence over the relays that have precedence over the builders that have presidents over the searches and you kind of walk your your way back like that

yeah tarun I would love to would love to get your your thoughts on that and I think we're going to kind of get into you know Justin what you were just alluding to which is some of the some of the philosophy that sort of sits at the heart of ethereum's design and Mev and like just exactly what that intersection is but you know kind of what we want to try to tease apart a little bit in this episode is like what is the um you know sort of the philosophies that we're using to design these systems and then what are the natural economic forces that exist in these systems and like where do those two sort of bump up against each other so Turin would love to just sort of get your response to what Justin said and also kind of tease apart your get your thoughts on those

that sort of attention yeah I mean I think there's definitely this um this uh you know the Supreme Court definition of Mev which is uh you know when you see it in reference to of course the famous uh ruling just declaring porn to be the thing that is you know when you see it but for some reason that mbv seems to to um aggrieve people in the same way I suppose uh the a good point that Justin made I think that's actually oftentimes not brought up is that you know people always think of it as maximal

extractable value um but a lot of Mev doesn't actually realize the true maximum people just try to find the simplest Arbitrage or the simplest liquidation or the simplest thing that they can kind of get some profit from and so the realize actual Mev might not be the maximum might be quite far from that um but I actually like to think of Mev as defined as the m stands for monopolist and you know in leader based blockchains with single leader the leader the reason that you even have this is that there's a single leader who is a monopolist for that block they get

to choose some features of that block such as the ordering or which transactions are included or perhaps which ones are excluded depending on on how things are set up but the idea is that in in every economic system a monopolist has sort of some extractable power of value because there's just demand for good that they control the precise configuration of the supply for um you know obviously there are designs that have multiple proposers per block um those of course in your other types of problems in the network you know like the latency get goes up or slashing goes

up or you know there's no free lunch in some sense but for Simplicity especially given that the majority of muv takes place on blockchains I have single leaders who are chosen I I've come to the personal belief and it's easier to think about it as monopolist extractable value because it really reflects the idea of hey you know in a lot of different circumstances monopolists just naturally have some form of value they can extract and it they may not extract the maximum necessarily uh I think the this idea of Precedence is actually quite interesting um

because there's sort of the sense in which it's not it's it is definitely more of a philosophical decision of who should be receiving the benefits or who should be whose losses should be capped Maybe uh in these systems and in some ways I think mev's sort of the way of defining the the theory of liability and in some ways for blockchains so yeah I you know I could give you a more strict definition you know when writing papers or proving some properties of it you you inevitably can't use the most broad definition

um and I think that's another reason it's very hard to reason about because technically you know in in most other systems when you think about a monopolist extracting value there's sort of only a couple things they can do they can like restrict the supply of the good to make it artificially scarce they can fake demand to make the price look like it's going up like wash Trading um but usually they can't change the behavior of the goods they're selling but the the thing about blockchains that's interesting is that the computation being run does change the actual state which is sort of the good being sold and so there's this very weird thing where the space of possibilities for a

monopolist to extract values much much higher the the set of actions they can take is is much more it feels almost Limitless relative to traditional economic theory of how monopolists extract values so um you know I maybe that was a little too too high level but hopefully that no I mean I think it touches on a very interesting rabbit hole which is that um the proposal or the block builder in this that they don't just run a single option even though we sometimes frame it that way we actually almost always frame it that way but they run many many options options in in parallel as well as sequentially right because there's a separate option from what goes on in the

first slot of the block and then there's yet another option for what happens in the second slot of the block and the third option for what happens in the thread side of the block but the attractiveness and the outcomes of these options are always contingent on what happened in the previous auction so they're in some sense sequential and in some sense also all happening in parallel and I don't know if there's any precedence for that what what do you think to run in traditional non-crypto systems yeah I mean I think I think this is one of the reasons I don't view Mev as that similar to say high frequency trading um

is that yeah the set of possible things you can do and what and the idea that like yeah there's this like inherent sequential nature where an action can change the particular good that you're selling on the next you know step a next event that occurs um is is is is usually quite foreign I think the closest thing is at auctions but even then it's usually more like sure there's a new product made on every auction like I want to Target you know people who are who live in New York and skateboard and buy Gatorade right and that's a particular product that might only exist

for an hour because like the product says anyone who fits in that bucket who goes on Instagram in the next hour is the thing I'm selling right and and that product is there's many of those right there's one every time but they don't have as much of the sequential interaction other than there's a fixed amount of AD space and they're competing for the ad space but there's not a sense in which the uh New York skateboard Gatorade person if that ad gets served it changes the New York surf border Red Bull persons realized ads in some sense right like they're not there's not this

interaction between them whereas executing a particular Dex trade May suddenly make a liquidation not hold anymore and so the there's there's that sort of interaction between these Goods is one of the kind of like unique characteristics I think of this that um you know yeah to your point that it is somehow like missed when people kind of think about a lot of people just think about it as like oh I got a worse price but there's also this this extra stuff in the middle that changes exactly why you got the worst price one uh just while we're on the at the New York analogy that was that was uh really interesting tarun yeah how would you

respond I think one of the the big philosophical questions in in a lot of mbb debates is is Mev avoidable should we be taking steps to avoid some of the harmful uh you know effects of Mev or should we just accept Mev as a fact of life and we should try to maximize the extraction and then make sure we're Distributing it to the proper the proper folks in the pipeline that Justin was outlining so there was a uh there was an op-ed piece that was written by Ari Jewels I think who is the PHD advisor to uh to Phil Diane and he published this in coindesk and he kind of used this little a bit of a hyperbolic sort of example but he said look the way that crime happens in New York is not efficient right there's a

lot of expense that gets spared right on policing but instead what we should do is basically auction off the right to burglarize and and mug people or their homes uh and then once we will kind of get our share of the loot and then redistribute that to the people of New York and I think the reason for that hyperbolic example is to kind of paint a very Vivid picture in people's heads of what is sort of going on in these Mev systems where we allow people maybe not explicitly a lot but we say sandwiching and front running is a fact of life for users but we're going to make sure that we see that that that Mev is redistributed to either holders of ethereum or stakers or whatever it is so I guess uh both Justin and tarun how would you respond or maybe do you agree with that sort of

example where does that example break down or what side of that philosophical debate do do both of you fall on oh I I was just going to add one line of historical context so that article came out I think about two to three years ago when the flashbots auction launched and uh introduced the idea of a bande which decoupled the position in a block relative to other transactions from the price that you're willing to pay and this in in many ways made sandwiching transactions easier um so that was that was one of the externalities introduced by by the freshbots option so we believe this it was happening anyway right so cementing was already running while in the in the

public memory but that was exactly the point that Ari Jules was criticizing here yeah so I mean I I guess I have uh quite strong views on this uh this article as being a bit of a not only a red herring but um very self-interested form of marketing for for Ferrari Jewels um and his co-authors uh so you know a lot of their research at that time especially when he was the head um researcher chain link was on Fair sequencing and fair ordering and and I think there's this there was this sort of idealistic idea that there would be

purely cryptographic solutions to to reduce Mev and you know this it happened to be that the flashbots auction came out at the same time as the the original research on this and I I view I view that piece as a very disingenuous marketing Ploy uh to promote one's own algorithm um but you know hey maybe it works because we're talking about it today in 2023 right uh but you know I I'm very much uh of the idea that it there's no real way to um to kind of remove these types of uh value transfers in these systems

um you know I think a lot of people you know some of some of that analogy is like sort of a bit of a just so story um because the the policing methodology is sort of a little bit you know like hey there's like a way of post-hoc policing things at all whereas in Mev there's not really like once the monopolist executes it it's over you know unless there's a fork or something but like you know let's if realistically it's basically over at that point um one thing I think that's worth sort of pointing out is that almost all the Alternatives that um are being you know kind of promoted like like what

he himself was doing in terms of fair ordering type of things they just preference a different user um so you know fully competitive Mev markets mean that every different type of actor who is part who is rational will sort of participate in bid in order to to either make their own value not Decay or you know extract a profit now obviously in in reality the set of users includes unsophisticated non-rational users and of course the highly rational users who are submitting bids in every auction right um

and so there's sort of this uh this sort of interesting thing that if you actually spend a lot of time trying to understand Fair ordering and I guess it took me a while to actually finally even like show some of these results that what you're doing instead of preferencing particular users like in a sandwich attack and preferencing the Strategic user over the unstrategic user right the the rational user who's monitoring everything to the user who's like I did a metamask swap I don't even care about paying the 80 basis points to metamask um all you're doing is you're preferencing a different type of actor in the system and it turns out these Fair ordering protocols preference different

applications over one another so instead of it being different users getting better you know utility it's certainly applications get better utility or better performance and other applications get worse utility or Worse performance and so the question is do you want to preference users or do you want a preference uh particular applications or application Developers and I would sort of argue that the former where it's something where you know different users are competing uh but you know any application sort of gets a favorable ground censorship resistance for the types of value payoffs that an application can offer is

more fair uh but that's just you know that's just my personal belief philosophically but this is where I think Justin's concept of Mev precedence is kind of important because this is this is exactly this question of like do you want to discriminate at the user level or the application Level and your particular design choices will inevitably leak down to doing some form of discrimination um but I think the most transparent form of discrimination is probably much better than the much more difficult to reason about forms of discrimination and the options are much closer to transparent forms and so if you have that transparency then you could think about redistribution right in a credible way that you really trust that it's

actually all being redistributed and there's not sort of like you know the uh the the dictator stealing all the the UN food dollars for their mansion you're and I want to kick this back over to to Justin in a second but true and you reminding me of something I think it was very very Plunkett of skip who actually made this analogy that is this idea in public policy where you can make a very small change and have very large very unforeseen consequences and the example being there's something in the U.S called uh third strike third strike policies where basically it doesn't matter that a third strike

offender of uh when you commit a felony for the third time it doesn't really matter what that third felony is but it carries with it like a very long uh prison sentencing and the idea there being hey you know everyone deserves a second chance but if you're a third third time offender there's probably something wrong we wanna we want to incentivize against uh you know crime like that but what actually happened is in these small counties where they implemented these these laws murder rate went up an enormous amount and the the rationale being that if you're about to commit crime for a third time and you know you've got you know like a minimum 20-year sentence you're going to kill the witnesses you know uh so just kind of uh you know a pretty

vivid example of how these even interfering in like a very small way to try to do some good can have these enormous adverse and completely unforeseen consequences so maybe with that example in mind you know Justin what you were describing before is almost a philosophical decision right like let's let's start from the fact that regardless if it's fair ordering or if you're an auction maximalist like some amount of economic value is going to be generated in this system and maybe this pipeline is more of a way to allocate you know where that Med goes instead of actually like making something that's the difference between this Mev Supply Chain versus like the supply chain that Apple uses or Toys R

Us uses to make so with that in mind Justin like walk us through kind of some of the design philosophies that you have um you know when making these decisions right so Tyrone brought up a great point which is that there's there tends to be these Monopoly powers and so really the President's List is is only really important to focus on the the top one or two kind of items because they they gather you know 99 of the of the value now um on the topic of you know monopolies I think uh Tyrone you're right that you know the status quo right now is that there is this this extreme Monopoly but

I'm actually optimistic that we can dilute the Monopoly we can decentralize it and the way that we can do that is by basically giving sovereignty to The Originators of this enemy so we've talked about the user if a user for example makes a uni swap trade they have sovereignty over this transaction because they originated it and they can you know we can Design Systems whereby almost all dmev goes goes back to them and so you know one of the questions that I have is how much of the Mev you know initially originates from somewhere and can be given back to the originator and maybe

it's way more than we expect and let me give you a couple examples one is Oracles if you if you're maintaining Oracle for example your chain link you're maintaining the ifusd Oracle price and my understanding is that it's a three or five multi-stick like these Originators this committee they get to decide what price they put on chain and when to put it on chain and so ultimately any Mev that's created from this change of price on chain from this Oracle should go back to them as the originator and so you're kind of decentralizing this Monopoly uh that that you know currently the proposal has and instead

kind of munching at it at various angles and you can maybe think about uh you know the the centralized decentralized Exchange um you know Arbitrage ultimately like these Arbitrage opportunities originate from somewhere right there's someone making a trade or placing a limit order or whatever it is on the centralized Exchange and maybe they should be given sovereignty over the Mev and so it's possible that over time we're just going to carve out the power of this Monopoly and just return back to the originator and I don't think it's a philosophical question I actually think it's an incentive thing whereby the originator is in the position of power they have Monopoly over their own you know economic activity flow and so they will

be incentivized to try and claim it back now another part of your question was around the externalities of Mev so we're doing crypto economics with blockchains and we're using incentives to to try and shape various outcomes and the really annoying thing about Mev is that they just distort all the incentives that you put at play you know and so um one of the the big research questions is how do you basically mitigate or completely remove these economic distortions and the good news is that from a research standpoint we've actually is is my personal belief that we have designs that completely remove these economic

distortions and and and remove the the negative externalities so you know we can talk about censorship we can talk about front running we can talk about incentivizing reorgs we can talk about you know potentially Mev draining economic bandwidth there's all sorts of distortions that can happen but for each and one of them that I've identified so far I believe that they can be fixed with uh with a clever design and these clever designs include inclusion lists there's encrypted mempools there's enshrine PBS um and and the whole slew of Apple designs like Mev burn as well hey guys quick break from the show here I want

you to imagine something for me imagine swapping two stable coins on chain paying zero dollars in gas and instead getting a rebate of two thousand dollars the something that's actually happened on chain to understand how I want to introduce and thank the season sponsor Rook zooming out for a second the current state of affairs at Mev is billions of dollars so far have been extracted from users Pockets using Mev Rook is coming in and saying enough is enough blockchain should drive value to their users and the applications they use it is time to leave the hobbyist era behind us if we want to move forward and we want to get this right that's why Rook has built a custom blockchain settlement Network and it's one that gives you full control over the entire transaction life cycle today you can

connect to an open source rough node the work protocol will automatically match bundle and auction your orders and transactions in seconds with zero gas overhead also any Mev that's discoverable along the way will be returned to you the user created as a collaboration between the industry's top mechanism designers and Mev Engineers Rook was built from the ground up to be scalable safe and programmable you can get your own mempool choose searches and Builders and link your mempool with others to discover even more Mev you can Define how the Mev is shared and delivered as well and what could basically process anything from transactions to meta transactions and more this is the way that blockchains basically should have been from day one so if you're a user listening to this here's what I want you

to do I want you to go to your wallets go to your favorite app your node provider and say Hey I want you to be working with these guys rock I want the Mev that I create to be redistributed back to me if you're a developer and you want to stay ahead of the game the best way to do that is to follow them on Twitter they are at Rook or even better yet slide into their DMS they are lighting responsive they'll get you set up today and if you do slide into those DMS as always please tell them that I sent you Justin could you could you talk a little bit more about the idea of Mev burn because I think most of the community is kind of wrapped their head around the 1559 sort of burn but Mev burn is still something that's relatively relatively

new um and you were starting to get into some of the reasons why burning Mev is actually a little bit preferable to kicking more of that Mev back to proposers you are getting into some of those economic distortions so can you just kind of layout that uh go into a little bit more detail about why Mev burn might be a be a good idea and how it can actually help avoid get into some of the specific economic distortions right so actually right now I'm preparing a whole you know bankless episode dedicated to Mev burn so I'll be speaking in another one and a half or two hours on the topic but um just just like running bankless basically podcast extractable value yeah podcast pev yeah you heard it here first depends

when this is going to come out I guess okay so basically the question is what are the bad things that happen when you give Mev to the to the proposer um and then one of the good things that could happen if you don't give it to the proposal if you give it to the to the eve holder so let's start with the the bad things one of the bad things is this idea of short-term reworks so if you have a spike of nav imagine a 1000 each block um and you and you give it to a very specific uh proposer think of it the analogy that I have in my mind is throw a piece of bread in central Paris what happens you have you know a

bazillion pigeons that come and try and eat the bread and they're all fighting for it and it's kind of the the same thing here um and basically what happens is the Searchers are fighting for this piece of bread and then the builders are fighting for this piece of bread and then what we really don't want to happen is for the the attesters the validators to be fighting for that piece of bread because what does fighting for the piece of bread mean for a testers it means you know trying to to reorg the chain and we don't like reorgs and so instead of having the proposals be the recipient of the Mev if it's if it's just burnt if it's just destroyed and effectively given to the if holders uh then we we just remove this

competition and this this incentive to to do the short-term reorg um another significant and I think underrated aspect of giving the Mev to the proposal is this idea of a lottery so when you play a lottery almost everyone is a loser um you know there's going to be a million ticket buyers and you know all of them except one is going to be the the the jackpot winner and it's a little bit like that with me I mean he's extremely spiky and what that means is that as a validator if you're just a solo validator more likely than not you're going to be earning less than the average APR and the reason is that you know some very

very lucky validator is going to be earning a whatever it is you know million percent APR at the expense of everyone else who's going to be like below average and so if you remove this Lottery aspect um to to to ethereum you remove the incentive of pooling so it's the same thing with with Bitcoin mining today right Bitcoin mining is this this crazy fact which is that if you buy a Bitcoin machine you you purchase uh you know you spend ten thousand dollars on the Bitcoin machine more likely than not this machine over its entire lifetime which is let's say five years will not mine a single block right so and and so

the the fact that it's just so so spiky for Bitcoin you know that you have this massive Spike every 10 minutes and you have just one lucky winner out of millions and millions of rigs that means that you need to pull and I think we're gonna see similar Dynamics especially if we enter a new Bull Run and Mev kind of goes through the roof in comparison to the issuance which is very very smooth so there's this centralization Vector through volatility and variance and then there's there's like a few other kind of interesting ones so for example if you're trying to build a pool um a staking pool like Lido or like rocket pool

um really you want to you want to be in a position where you smooth out the the uh the the Mev and in order to do that you're kind of expecting that the these operators decentralized operators will uh will send you the Mev now the way that you design these pools is that there's some amount of collateral it could be reputational collateral for example in the case of Lido or it could actually be if collateral in the case of Rocket pool but what if the Meg in a specific block is significantly larger than the collateral imagine there's a 100 000 if block um and it's just worth more than your reputation and instead of giving that

1000 Eve block to the pool you just run away and take it for yourself so for example I'm I'm just taking an example attestant is one of the operators in in in Lido and most of the time you know you know their reputation is worth uh you know more than the the block value but there might be this one one block whereas is just so much more valuable for them to just run away with the money then give it to to Lido and for Rocket pull is even more of a problem because you don't really have this this reputation at stake um and you know they're moving to to lower amounts of collateral only a teeth and so if the block value is more than eight if you're just incentivized to run away uh

run away with the money instead of giving it to to the pool and so the the idea of burning kind of makes it much much easier to to build robust and trustless uh liquid sticking derivatives and then another um kind of interesting um thing that happens when you give the Mev to the proposers is that you're you're in in you have uncertainty around um economic bandwidth so there's two um there's two public goods that are credibly neutral that you can fund through Mev public good number one is economic security and public good number

two is economic bandwidth and these two things in my opinion are kind of disjoint you can't really you don't want to mix the two together because otherwise you you get kind of this lower quality economic bandwidth but basically what I'm trying to say here is that if can be used for two things it can be used for staking or it can be used in D5 for example to collateralize a stablecoin now if we have a massive Bull Run and mov goes to the moon what will happen all the if will be sucked out of the evm and it will go to the beacon chain for staking and that just won't be left that much economic bandwidth to power for example decentralized stablecoin

and so you have you're going to have this economic drought which dries up the the land the fertile land of of Mev and it's almost self-defeating whereby you know ethereum is doing so well that it's producing so much Mev that it dries up its own land and so if you if you burn the the Mev instead of giving it to the proposal then you have very predictable um amounts of economic bandwidth that you can work with in the context of the evm I would be I would be very interested to to hear your perspective on I think the underlying for me anyway question is kind of how much should we try to minimize Mev at the protocol layer uh for example with techniques such as

privacy um versus encouraging application developers to do better because I feel like if we go too much into um if if we go too much um into just saying um we use techniques that minimize Mev as much as possible at the application they are then discuss almost like a blank check to application developers to do whatever they want because they know they have this floor under the second floor underneath them right that basically bails them out for any mistake that they're going to do and so the result is they build applications that create a lot of Med that then has to be redistributed or protected by some in protocol mechanism

and so we kind of layer on more and more of this economic pressure on top of the protocol so I wonder if if you thought about this before because a lot of what you have written about in the past with the kind of staking versus uh lending Etc um it kind of resonates with this trade-off between the application and protocol layer yeah that's a good question I mean it's funny to think of like better Mev redistribution is acting as sort of like FDIC or the FED that you know perhaps that that's not the wrong analogy um I think one kind of important question for all of

these things boils down to you know how well can you actually audit in you know post talk that like all of these processes have taken place correctly for instance in the the example Justin had of like there's just this hundred thousand eighth Mev opportunity in one block and then the validator is part of a pool doesn't doesn't redistribute it um you know having ways of really being sure that those types of transactions which are technically happening in the transaction supply chain before you know people have come to consensus or agreement uh

are somehow very it's verifiable that that that transaction existed and it got redistributed and the burn was done correctly is um extremely important I do think an interesting this this idea that application developers or incentivized to build very high Mev applications is a very Fair Point you know arguably you could say that uniswap became very big in 2019 because it was the only application generating like copious amounts of Mev at that time right like most of the other applications at that time that maybe generated some value for users

were like lending protocols and liquidations are just not that frequent Etc and so there's sort of this feedback loop between the lending protocols and uniswap where like they'd both kind of feed create Mev and feed off each other and that would generate almost like transactions incentivized watch trading right on the back of a piece yeah yeah yeah yeah yeah exactly um and this is before there were farming rewards I think the the contracts to do that were only really written oh there were farming rewards the IPS were the farming rewards yes yes but you know you know I think those LPS probably still did well because most of them were in the make or die pool or

east dipole which you know probably still amongst the top earning pools in history um fee-wise but uh there's sort of this yeah there's sort of this kind of uh in thing that like there is some amount at which you want the applications actually generating Mev because they just generate transaction fees in general um arguably you know Bitcoin as much as it's hated Mev the the the Wizards coming out generating transaction fees and then sort of you know bitcoin's very unfortunate case this off-chain Discord Mev of of like how the transactions are being

orchestrated uh you need to post a shower video in order to get one yeah exactly uh you know like there's definitely this this interesting question of whether there's moral hazard introduced by by um by by burning Mev but I think the idea that you can reduce your issuance um which you know we have a lot of evidence that that these kind of burning mechanisms are actually good for is probably worth it um I guess the real question is when you have so many applications and so much usage that the only usage is the parasitic applications that are generating a ton of Mev

um and to be honest no blockchain has ever reached that point yet in my mind uh where where it's like you know 90 of blocks are filled just solely for Mev um actually that's that's a lie that's a lie I think Luna during the crash probably qualified um I think like when Luna was when when UST was losing its Peg I remember seeing some some like interesting mental statistics of what percentage of transactions were withdrawals versus what percentage we're trading against the the Luna amm Oracle and it was it was like 98 but that that being said it's like an exceptional time like the

usdcd peg was probably close but even then it wasn't like 90 of a block um so I guess my point is you know the the [Music] I'm not sure it's a good idea to totally I feel like application developers should be free to do what they want philosophically to me because I I don't think that there should be sort of some idea that like hey we chose a mechanism that somehow restricted the sets of payoffs they could offer users now of course they're they're externalities that exist but I think one important piece of

you know correctly figuring how to do something with those externalities is you know one of the reasons I think personally auctions are much more fair than latency games and this is sort of something that people like to talk about a lot is like which is a more fair system first come first serve latency game or something that's an auction so at the very highest level auction Theory sort of states the following whatever prices people pay in an auction if everyone's rational no of course that you know who knows if that's true but but for Mev auctions people are extremely incentivized to be only the

rational participants um that the price that people pay is equal to the total is at least the total externality that you cause on all the other players all the other users and so auctions have this sort of nice property that at least formally if you could audit and verify all the steps leading up to them that the price cleared is actually a true measure of the negative externality caused on other users by a particular user just to clarify you you yeah you mean the for me the pain of not getting the

item yes exactly so the you know if you probably one of the older ideas and auction theory is something called the Vickery Clark's Groves mechanism um and the the the vcg mechanism is exactly this it is you make an auction that charges if you knew if you were the auctioneer and you were also omniscient and you knew the utilities of the users given the item or not getting the item uh you basically charge the winner the sum of the negatives of the utilities that they like sort of the losses you implicitly caused um and in some ways you can't give such a guarantee with the latency games uh

there there are some like very extreme counter examples that show that like you you really can't um you can't force the winner to pay something like that negative externality and so but the the way the ability to do that depends on how well you can verify and audit the entire chain of TR you know State transitions so that you knew the utilities of those users to cause that that sort of payment so I'm not saying it's like we know exactly the entire step but you can see theoretically at least formally that there's a sense in which auctions have this property that I think is much more fundamental to Fair systems of being able to to to charge sort of the

net loss uh to the winner and and sort of make things more fair and so as Justin alluded to like all these little mechanisms put together should should get to that hopefully but I think there is this necessary thing of like really being able to audit and verify that everything happened was true and doing that amidst a a transaction supply chain that's getting way more complicated right now right the those two things I think are are sort of somewhat at odds with each other there's so much interesting stuff here to dive into I'm for this season I'm really hell-bent on making um privacy and latency a big topic um and I would like to dive into both

you already touched on both uh to ruin that I would like to ask you first Justin so on the role of latency so you are running a relay or at least you're one of the the people behind the ultrasound relay and your relay is is leading a lot of innovation in latency optimization in in math boost right now so would you walk us through the role of latency in the Mev supply chain and um what steps you're taking in that field but but also if you see any risks here for ethereum right um just before that I just want to make a very quick point on something that

tarun said around the the depth themselves being the recipients of Mev I think certain types types of dapps are very natural recipients of Meb because they are the originator of the economic activity so for example if you take a roll up you know they have they have sequences for example and so because that sequencer is the one that receives the economic activity and puts it on chain and so from the perspective I feel that's the originator they're the ones that that will receive it and I think that will contribute to this idea of breaking down the the Monopoly position that the proposal might might have uh might have

today and so you can think of Mev burn that's not being you know burning all the Mev across the whole world is only burning the leftover muv that uh you know is provided by default to the to the yeah exactly I the way I like to think about Mev brand is basically it's just a redistribution scheme from proposers to hold us uh so it doesn't affect Builders it doesn't affect Searchers and it doesn't affect applications wallets and users so anybody who basically comes before the proposal in the Mev um in the Mev supply chain right and um the reason why I'm I'm personally personally don't see the debate about uh Mev berners as like so

interesting also fundamental is that I think the line between proposers and hold us is blurring anyway I think um I think that more and more proposers slash stakers are now also holders in the sense that they are using their uh their stake as collateral somewhere they can use it in D5 and whatnot and from the other direction more and more if we'll be staked uh because it's becoming uh more and more liquid right so this is the the role that these liquid staking Protocols are playing and and I the more these two roles converge the the less I see the need to make any protocol changes

um in the realm of of burning mevp because it's it's you basically just moving at left pocket right pocket yeah agreed so one thing that can be done is you can do Mev smoothing which removes a lot of these um economic distortions and if we live in a world where 99 of if holders are staking then yes then you know burn versus smoothing is is irrelevant but the reality of today is that 14.5 percent of all Eve is staked and that is a minority it is a significant minority and it will grow but I my guess is that for the foreseeable future let's say the next few decades it will be less than

the majority so it'll be less than 50 of all if being State and so as such it is a an important debate now in terms of latency um you write uh right now we have this uh this incredible latency game that's happening at all layers of the Mev pipeline and we're we're relays are also part of this this latency game and we've been shaving milliseconds here and there and just to give you a few uh you know fun examples of things that we've done one is um with this we disabled cloudflare because cloudflare was costing us 10 milliseconds of latency another thing

that we've discovered is that if you restart Gaff periodically for example every hour it just runs faster so for some reason if you leave gef running for a long period of time it accumulates some sort of garbage that we don't completely know where it comes from but we just restarted periodically another thing that we're doing which is a door crazy is that we're running a gef on a ram disk so we have a half a terabyte of RAM and we put the whole ethereum state on RAM so that you don't pay so much of an i o cost when you go read and write to the state and you can do late simulation faster we're also

using like state-of-the-art AMD epic CPUs and at the software level we're doing things like optimistic relaying where we invite Builders to post collateral at most one if uh to to to the relay and basically any block that comes in which has a value less than one if we can go ahead and forward the bid without verifying without simulating the block um we forward a bit to the to the proposal and if it so happens that the block is invalid and the proposer missed that slot well we have the collateral if need be to go refund the proposal now we've

been running optimistic relaying in production on mainnet for for about 10 days now and that has allowed us to shave about 100 milliseconds at 100 milliseconds is enough for us to win a very very significant portion of the block so we're winning about half of the auctions that we participate in despite you know pretty big competition is about uh you know uh 10 relays that are competing in in in in in in these blocks and and we still happen to to win to win half of them do you see any risks from the ultrasound relay taking too much market share in the relay market and if if yes are you thinking about any steps to mitigate that

yeah absolutely so from day one uh when we launched uh we publicly said that we would self-cap to one third of the block so we're kind of self-imposing a social Norm um so that we we encourage and Foster diversity um and one of the things that we're we're doing our mission as the ultrasound relay is to try and um remove uh relay censorship and so we're helping some of the other non-uh centering relays for example diagnostic relaying uh getting set up with all the tips and you know optimizations that we found uh so that they can they can be in in the top position uh with us um I mean on that point of censorship

it's kind of interesting because if you are a censoring relay and you want to guarantee that none of your blocks have any censorship then you must do the simulation and So structurally speaking censoring relays are at the at the fundamental disadvantage and I don't think sensoring relays are gonna be winning latency races um censoring relays are only going to be relevant when a uh a validator has opted to only connect to non to censoring relays which is a a small minority um now in terms of the the risks of latency like the the obvious one is that

it becomes a you know a winner take most a game whoever is just best at doing this kind of thing can can can win it all and so we need to start thinking okay what can a monopolist uh do for example what can a builder that has a hundred percent uh market share uh do and like one of the biggest concerns that we have is censorship now it turns out that there's this idea called inclusion list that basically Force the Builder to um to to to include transactions and so that that specific concern uh is it is a is a short-lived one because we have a design that should be uh on on

chain hopefully in in in in a few years now um one of the interesting kind of uh feces that I have around latency is that it's actually going to lead to every single layer of the Mev pipeline to be distributed now what do you mean by distributed I mean geographically distributed so for example the ultrasound relay we have plans to have an instance of our relay in multiple geographies we might have one in Asia one in Europe one in America and the reason is that um we want to be close to the validators right and it turns out that validators

are geographically distributed and so if we want to win the races let's say in Asia well we need to be close to Asia if we want to win their aces in Europe we need to be close to Europe and so in a way um this latency race is going to increase robustness meaning that we won't be in a single jurisdiction we'll be in all the jurisdictions we'll be everywhere and nowhere and so that's kind of that's kind of interesting and it will it will not only happen at the relay level but the Builder level and we we're already seeing that at the Builder level so each Builder can have multiple Builder Pub keys and what tends to happen is that the builders want to be close to the relays and so the flashbot relay is in the US and so Builders usually have one

pup key and one machine which is dedicated to flashbots uh that is co-located with flashbots and that produces censoring blocks and now we're starting to see is Builders also having a machine and The Pub Key which is dedicated to to building blocks which are non-censored for the ultrasound and the agnostic relays and that are co-located to to to to these two relays I think it's a bit of a stretch to say that latency makes the system here more robust I think it's the the fact that latency plays such a huge role in determining who wins and loses in these systems means that you have to take much more deliberate design decisions on where

like how massively replicated your software is and where in the world exactly you located right um wouldn't it rather be that way 100 yeah there are risks and you know I I it's more of an optimistic take that okay okay and I mean on the Searcher side there's you know Searchers are it's interesting because they want to be connected to two different types of parties on the one hand they want to be very very close to the mempool and the users whenever a transaction is broadcasted they want to know about it as soon as possible and they want to be close to to the builders and I think the set of users is extremely geographically uh distributed

and I think that is one reason to be very very optimistic which is that you know depending on the time of day or depending on you know the economic activity that happens in one country versus another one um the the the set of relevant searches and Builders will shift and and be somewhat decentralized uh through the decentralization of the of the user base itself tarun I would love to Loop you in here and and hear your take um on the role of latency in blockchain systems especially if we move to a more layer 2 execution uh like more execution happening on layer twos

one risk that I have personally is that users have gotten so used to having much faster latency on these layer 2 systems that now they kind of back themselves into a corner where they practically have already entrenched um certain Market structures without knowing it without like deliberately designing for it very curious uh what you think about that yeah um I think the you know the the sort of natural view of blockchains in 2016 was hey there's sort of a notion of like how many transactions are in a block or some bound on that and the notion of there's a single time scale that's a

confirmation time or you know distribution of confirmation times that people understand and yeah I think the thing about layer twos is the different security assumptions mean that there's extra there's many different latencies you have to worry about and it's actually the proliferation of the number of different latencies that seems to be a bigger risk in some ways so for instance um you know in a optimistic Roll-Up there's the latency for your transaction to get confirmed uh on by the sequencer there's a latency for your transaction blob at least just the data to get confirmed by the L1

then there's a latency for you know if you have a frog proof for submitting the fraud proof and then there's the latency for the fraud proof confirmation um the the I separate the last two because you as the user if you want to generate the fraud proof it takes you some time and that time might be significantly larger than the time it took you to generate the transaction depending on the architecture of the fraud proof um and so it's that's meaningful time to you now for all of those buckets you have to assign some probability of like how much you care about that latency right and now the users have to have sort of utilities at different time

scales and for each of these different sort of types of events like how much of a loss do I feel and I have to submit a fraud proof um this type of stuff certainly changes Market structure and I know Justin's definitely written about this of like if all the Mev goes to the L2 sequencer or maybe if you know maybe there's some World in which they're decentralized um you know how how does the L1 kind of get some of the redistrative benefits of Mev and of course I think some of these designs for Roll-Ups that that sort of share that at that boundary are sort of in in that view um

but I think that there's always going to be some latency game right we're dealing with like physics at some level um and all I have to say is blockchains are dealing with latency very far from where you know you're actually at anywhere near the limits of physics because I mean let's be honest like I can't even evaluate a BLS pairing in like less than you know 500 microseconds or something so it's like you're pretty far from from like the physical limits but I do think that uh you know there's always going to be the more the supply chain gets complicated the way more there's many different

latencies and now your optimization problem is not just like hey am I fastest it's like hey am I fastest in this particular latency Dimension like my fastest at intercepting fraud proofs and putting transactions right after them because I know that it will cause some State change that I make money on um upon upon the fraud proof executing correctly like there's stuff like that that I think you know doesn't exist or barely exists now but the introduction of these like multi-time scale systems will will basically create a ton of new Mev of some form um now the question to me I guess is like will

verifiable computation so ZK proofs I'm not I I know I know some some people on this call maybe like tease personally I'm very I just find T is to be trusted execution environments to be uh just historically fraught with a lot of things I'm not I'm not super big fan of them personally but I I do think like once we have better verifiable computation uh whether it's CK whether it's you know IVC fhe whatever um you know how much can we move away from adding more latency layers right you have this like latency onion you could

think of like two competing forces one is like improving user experience means like adding that increasing the complexity of this latency on the end of like each different layer of like interaction that you have to deal with and you can think of like the new cryptography is like reducing the latency onion and the question is like what's the you know what's the trade-off where's the equilibrium and a lot of that depends on how fast we I think get better at some of the new cryptography um you know I think this is the first time probably since 2018 where I I personally had the very strong opinion that new cryptography actually will change the economic microstructure much more than the other

way around really new economics incentivizes people to use kind of you know new cryptography I actually think cryptography will drive some of the microstructure changes a lot more in the next three years than it did the last three years it's very interesting jumping off point maybe to get to our last topic which is privacy um why maybe to stay with tarunya for a second and then Circle back to Justin tune why why do we want privacy in the Mev supply chain yeah I mean I guess there's a couple things one is you know as I said earlier in this like idealized theoretical world an auction is supposed to be able to charge these negative externalities

but if people don't have privacy at a very simple level if I can see everyone else's behaviors I can make that I can I can basically you know distort the market somehow to make sure that I'm I'm paying less of the externality cost so that's sort of like the like highest level economic reason for why privacy can be important in that you know uh you know in theory you know seal bit auctions and like ascending open auctions should have the same Revenue but in practice they don't because if I can see other people's bids I start changing what I do if we if if some of

the value of the object comes from being a common value it depends on what I think everyone else's value is then it becomes much more manipulative I can manipulate things and and the only real way to kind of prevent those types of games is is through some form of privacy now the problem with privacy is that I think people in cryptography and people in machine learning have two very different definitions of privacy but I think the existence of D5 applications suggests that you kind of need to have both at the same time and what I mean by that is you know there's this notion of of things in finance can't be a hundred percent private

there always needs to be some public information right like you're not going to trade against the decks if it doesn't tell your price you're not going to take a loan if like you it's a black box and you have no clue what the interest rate is you're not going to bid in an nft auction if you have no clue what the sort of minimum price is or some properties of the thing you're bidding on so there's some sense in which Finance unlike a lot of other things explicitly needs to have this this notion of public information somewhere and this gets back to our the overall point about auditability and verifiability that the reason you care about that is because you want to be able to verify that you didn't get

over and and or like many people want to verify that and so the the sort of this sort of idea that there's some public information that's can be used to audit is also a form of loss of privacy because you know when you observe a change to a price if I can traips back through the logs and say oh okay like yeah this person did a private you know Dex transaction but hey guess what they caused the price to move this much so now I can infer how much they traded there's a sense in which you that's a different type of privacy than just like hey my state is shielded and and there's sort of always this tension

and finances how much information is public so that there's some efficiency in the system and verifiability versus how much is private but like you know they they're connected to each other right like there's always some impact thing and people in machine learning only care about this form of privacy in fact arguably they they mess up the state form of privacy completely and and there have been a ton of these kind of like attacks on machine learning models that demonstrate this right of like tricking the you know facial detection thing to to to to incorrectly do stuff or people with language models like tricking them into to getting into an intention Loop and

then not being able to get out of it but my point is that I think crypto is actually you know because there's capital F Finance involved you kind of have to have both the state level privacy and state level Behavior as well as this public data level privacy and somehow melding those two pots together is going to be extremely important that's that's my my bully pulpit on this thank you thank you um I think we we have a last question um and before we end and uh yeah I would like to kick this to Justin because uh I know you have written and uh yeah published about this at length but

what what privacy approaches um do you see in order to encrypt public blockchains and maybe what a very high level the trade-offs of these what can we expect over let's say the next five years right um so just to give a little bit of context of how I think about privacy I think as privacy as a tool for for sovereignty um like imagine for example if um you didn't have privacy in your own mind in your brain like someone could read your your mind then any action that you would want to take could be front run for example and like the two things that privacy helped for in the context

of blockchains is one front running and and two censorship and I think by far like the most important form of privacy that we that we want is ephemeral privacy is that this this various short amount of privacy between when you have the intent that you want to make a transaction and uh the the the transaction get gets settled and that could be you know just a few hundred milliseconds of of of a privacy now in terms of how do we provide these few hundred milliseconds of of privacy we want to basically build this notion of encrypted mempool what is an encrypted man pool it's a place where you can go broadcast your transactions

and retain uh privacy for a few hundred milliseconds until your transaction gets settled and the way that it works is that you broadcast your transaction encrypted the encrypted ciphertext gets included on chain and only after you have inclusion on chain do you do the decryption and the settlement and um you know the the act of including and decrypting and settling all of that can happen simultaneously the important thing is just that in in transit in Flight you you have a ciphertext instead of the the plain text now in terms of the the the very important property that you want of the encryption is what's called

guaranteed decryption you want to be able to guarantee that the ciphertext can be decrypted and will be decrypted so that you don't give the user the option to not you know give the decryption key so basically the decryption key should be kind of outside of the user's control and there's basically a few flavors at play here like the the first one is a what's you know time-based encryption specifically what's called delay encryption the idea of delay encryption is that you provide the ciphertext and then automatically some amount of time later it could be for example one second later

um the ciphertext just decrypts itself and you can see the plain text and this is tightly a link to this notion of verifiable delay function where you're using sequential computation to mimic the passage of time right because it takes some amount of nanoseconds to do one cycle of the sequential computation and you just add up you know sufficiently many nanoseconds to get the amount of delay that you want the other flavor for for encrypted mempools is to use basically a committee with a threshold assumption so you have um you know something like an NPC where you have multiple parties and you assume that sufficiently many of them are

honest and online and um those parties can work together um and and decrypt uh your your your ciphertext so these are like the the two flavors and then there's this final really awesome flavor which generalizes all the forms of of of uh decryption with guaranteed outputs which is called witness encryption now here we're nodding out a little bit because with this encryption is is not practical but the the definition of witness encryption is that um you can specify any statement that you want I as your decryption key so if you want to have delay encryption

your statement could be I know the output of a vdf um and and here's a proof that I I know it if you want to do um you know threshold baits decryption you could ask okay I know uh you know M of these n signatures and I can give them to you and that that is the basis of the decryption but you can do all sorts of other fancy and wacky stuff so you could do for example only decrypt once the blockchain has finalized so you include the ciphertext on on the Chain you wait until you get finality and then that leads to decryption another thing you could do is a hybrid approach whereby you have threshold decryption but if it

so happens that you can't decrypt because some of the um the members of the committee go offline then you can fall back to a delay based encryption so it's kind of an or statement either you decrypt it with the committee and or you used uh the the delay encryption and the reason why this all statement could be interesting you know kind of brings us to some of the trade-offs here at play so with time-based decryption the main trade-off is that there's a little bit of additional latency from for the user from a from a user experience standpoint it might not be very large you know it might be on the order of hundreds of

milliseconds or seconds but it's still kind of noticeable in the context of a of a web 3 user experience and it might be on undesirable whereas a threshold committee can really very very quickly do the decryption the downside of the threshold decryption is that there's much more trust involved instead of trusting physics and only physics and Mathematics you're actually trusting humans and here the um there's two two things that could happen that are bad like one is that you get a colluding committee that prematurely does the decryption and kind of rugs You by front grinding you or censoring you because they can't see the plain text before everyone else

um and then the other downside which I already mentioned is that you know the committee might be offline and so you never get the decryption and so that that that's bad for for for liveness and it can also be um basically used to extract toxic Mev through essentially censorship and basically the the committee as a whole has the option to not uh go ahead and and do the decryption and that gives them uh leverage now actually I do want to highlight a third form of guaranteed decryption which I haven't mentioned which is Enclave based so you could use Hardware to kind of guarantee that a trusted Hardware to guarantee that your ciphertext will get decrypted or you know if you want to be like very very

kind of naive and and and blunt is you can use the trusted third party and actually that's what is often used today for example of flashback protect you can think of that as an encrypted mempool which is operated by a centralized operator yeah I thank you I think that that gives a nice kind of road map from so today we are basically we need privacy and Mev supply chain because without it it just doesn't work uh for the reasons that tune laid out and so today we defer to a trusted third party but we are actively looking into various designs to decentralize that I'm sure we could go on for hours talking about this but this is probably as good a place as I need to to end it uh so Tarin and Justin thank you so much this

is uh you've given also a lot to think about now if if either you want to do a little Shameless Shameless shell for that for the work that you do to ruin maybe just uh tell people where to follow you or find out more about the work that you do at Gauntlet and Justin scene for you and um whatever you want to talk about ethereum Foundation or ultrasound money or or whatever yeah I think the easiest place is find me on Twitter uh which is my name uh and uh I think the only thing I will shill is that I'm still working on making sure that people realize Fair ordering was a scam well sure you made a meme about this so we'll we'll share that actually in the in the video as well but uh appreciate that

it's a good meme so on my side it's it's very easy to contact me I have open DMS on Twitter but I'm also reachable on on on Discord and email and various ever um mediums in terms of Shilling um a different Foundation I I think we're at a place where we're very comfortable with the the research um and so what we need now our main bottleneck is coordinators actually we recently made a higher we hired Mike newder to be the enshrined PBS uh coordinator and I think it would be great if we had someone who could be a champion a leader for inclusion lists

and similar thing for encrypted mempools and Med burn and the reason why we need these Champions and coordinators is because the way that the development happens at the firm consensus layer is extremely decentralized we have you know five consensus layer clients we have many execution layer clients you need to interface with academic mix with Auditors you need to educate the general public you need to go on podcasts like this it you need to attend uh you know the governance calls like the Oracle Dev calls you need to write eaps it's a ton of coordination work and that is the main bottleneck and if that's the kind of work that you're interested in definitely do contact me on the ultrasound

uh project side of things basically what we're building is public goods that um you know no one else is is is is comfortable doing um so neither different Foundation is comfortable doing because the remit of different Foundation is the the uh the the consensus layer and and none of the kind of token based kind of for-profit commercial projects are also comfortable doing because it is it we we don't really have such a a commercial mindset and so we're we're providing infrastructure like the relay which is just like pure public goods and we have all sorts of other ideas of of core

infrastructure for ethereum so one idea that we had recently is what if we build an ultrasound Oracle so if you want to build like a truly decentralized um stable coin while you need a truly decentralized Oracle if you SD Oracle and right now you know maybe the best thing that you can find is chain link which is the three out of five multi-sig uh to my understanding and you know maybe we can do significantly better and and build this Oracle uh and we have all sorts of other ideas for building cool public goods so if that's of interest to you do contact me as well awesome guys so if you're listening to this and that sounds uh that sounds interesting definitely reach out to Justin

um so Justin and true and thank you guys so much for for helping us when I kick off this season and uh yeah uh giving us a lot to think about so I'll have to do it again sometime soon that was great Justin tarun thank you thank you guys cheers thanks all right hatsu that was a that was a great episode uh certainly we covered kind of the breadth of topics that I think you and I were were hoping to to cover here from this episode and dug into a lot of the the big questions that we sort of teed up in in episode zero right the kickoff so what um you know I'd kick it over to you first like what did you find interesting or what do you want to kind of dig into oh my I mean I'm really interesting to hear

interested to hear from you as well um for me I think the highlight of the discussion was to ruin dropping some knowledge about auctions I I'm just a sucker for option Theory um when he drew this he teased out that Mev auctions are so different from any other form of auction that you see um in other markets including the hft um because the good setup being sold are so contingent on each other right so whether the the item that's being sold at slot zero in the block is uh it influences the item that's being sold at slot one two three four and vice versa and and everything's

interconnected and yet happening at the same time and designing such a system is so much more nuanced um and difficult to get right that I yeah I think it's one of the most uh sort of intellectually stimulating um topics I'm in agreement with you there I'm in agreement I actually really liked the uh not to not to guess you up here but I like your definition that you that you gave it at the beginning which is in any sort of economic system there are actors that have certain advantage or preference and start to see me be everywhere and I I did I did think and it was interesting that he also

brought up the Google ad Network and that was an example that you used of real life MVP as well Justin actually gave this definition right that that you referenced uh that Mev is the value that the system can extract from the activity that the system is providing and originally it didn't quite sit right with me but then I let it sink in through the course of the episode and I thought um well this is actually kind of true right because it um it resonates with the with this idea that Mev cannot be minimized so it cannot be removed right it can only be minimized and then the rest has to be distributed According to some uh precedent like he called it Mev precedence some ranking right and um and

and um I think where he's very right is that these agents are all part of the same protocol they're all part of the same system and we need to make sure the system is designed in a way that these agents can design can extract less and where we do where we can't keep them from extracting we need to have a good ranking a good mechanism for Distributing these things including back to the user yeah I agree I I also found myself thinking about his definition and uh maybe trying to poke a little bit at this analogy of Mev as a supply chain whereas you know Supply chains are set up to produce a good at the end of that

supply chain whereas Mev is almost something that already exists and the supply chain or supply network of participants that you and I have been referencing are actually more of an allocation mechanism of where should that Mev end up going um so it's a little bit of I'm not sure how consequential that is but I found myself in my head thinking about that distinction as we were going through the episode I think it remains primarily a supply chain but I thought it was interesting that he used it as a jumping off point to create a ranking as well yeah I agree um now one of the and this might this might get a little bit beyond you and I were just kind of briefly discussing this but

um you know it's a little bit beyond strictly the the bounds or the parameters of of Mev but I'm sure we're going to talk about proposer Builder separation um you know you give a great talk about PBS on rollups uh and how that's going to move into the four but I I do think you know it's worth kind of continuously posing this question um of you know how protocol decisions end up getting made and what Solutions are enshrined versus which ones are kept off chain and the trade-offs there yeah we we would have loved to ask them the guys this but the episode was already a bit on the long side and um I think it's such a fascinating question

um so I think that's generally kind of three ways that the um Folks at over at the ethereum foundation are thinking about this the the first is um and Shining something uh has a cost right and so uh because it it increases basically the complexity of the protocol it it makes it harder to to reason about um it um yeah basically creates some form of bloat in in the protocol introduces more risk and higher cost of Maintenance and so you want this cost to be balanced with uh the benefit that you get so why would you want to enshine Something in the protocol in the case of

math boost for example you would like to get rid of the this role of the private relay and somehow bring it into the protocol where you give these this job to maybe the committee of the testers or the validators or some role that's that's basically part of the core protocol and if the benefit that you get here in terms of removing a trust assumption is very small then very often would not be worth the cost then the the second um I think the second thing that that they are thinking about is how much is the market structure for this particular protocol is still changing and a protocol kind of it it negotiates or mediates kind of coordination between different parties

um but the protocol must reflect the economic reality and the technological reality that have that that happens between these parties and so in order to stay adopted protocols have to evolve if the market device the protocol has to evolve right same as so um I I recently saw a great example um and we can link to it but so one of the reasons why the pandemic was such a big uh it had such a big like mental toil on people was that many of the protocols of daily life actually got disrupted by this such as um you know being like physically close to people shaking their hand and and stuff like that right these are the the

protocols of human interaction at the most basic level and if an event happens that disrupts you know this basic level then it's it's it's much more taxing for people maybe um kind of Beyond everything else and um and yeah so if you if you still have to change the protocol uh much then also you don't want it to be an inside inside you don't want it to be enshrined basically because everything's it's much harder there yeah and um oh so we already touched on this I guess but you know the the third point is is just this idea that you want the protocol to be as lean as possible because you want it to be very easy to to reason about right you want it to be very easy to to secure and and so on

yeah that's um maybe leading into the the other the other part of the discussion that I thought was was interesting and there was kind of a you know partly this question that you posed was answered by tarun and then partly by Justin as as well but just sort of uh you know what parts of our kind of ethereum is the base protocol level responsible for and then also like application developers um on top of on top of ethereum and there's kind of two questions there right which is it's almost like this question of who is Meb is it anyway because one way you could look at it is there's going to be economic advantage to generating Mev and that will probably tend towards the execution layer the the sort of app layer but then also you need

to protect against some of the more harmful forms of Meb as well and mitigate some of those negative externalities and should that also be the responsibility of the applications or you know should that be built into ethereum and those are just those are interesting questions that is such a yeah yeah I love this I love thinking about this question and I think there's no easy answer but um I I think that the way that Tyrone framed it as should ethereum be the fed or the FDIC of you know of Mev in the sense that should that develop uh mechanisms that basically protect the applications that build on top of it and they use us or should should these application developers get better at making apps that expose less Mev and I think I mean the answer is a little bit

of is a little bit of both I really do see a risk that uh protocols put Mev protection at the base layer and this typically comes without it doesn't come without trade-off right you do take on more um just assumptions for example uh worst user experience and so on and you kind of bail out the application developer on top and uh and in return no application developer cares about minimizing Mev anymore and so then we beat this to there it just it Stacks up and it gets somehow either protected against the protocol layer or redistributed According to some formula and I I wonder

if that doesn't introduce more instability potentially um maybe of the invisible kind that isn't the problem until one day it is I think you might be right I also think just one more observation before we move on to latency because I thought that was that was a fascinating part of the discussion as well especially Justin's observations but you know one thing that I feel like people sort of mischaracterize sometimes crypto is often called the wild west where it's really you know the way these networks are designed is they're trying to implement rules and I think there's like a couple of different ways that you could look at that and I think have you ever heard this expression that uh politics you know the country should be run more like a business and that seems

to appeal to people and I that's a real pet peeve of mine that people say that like designing political systems are very different than designing economic systems although obviously those two need to interact and in a business you know the way we've got capitalism kind of set up is you're trying to maximize efficiency you're not as concerned with fairness whereas in political systems you're maximally concerned with fairness and I do you know I think you can I think it's parts of crypto are applicable to both but I do think like looking at it from a government or a legislative or a public policy sort of perspective especially like when you get lower on the like especially for the L1 layer of crypto I think that's the appropriate lens to look at things oh

this is such a good analogy that you just devote I love this uh I think that the question of how to approach Mev is so similar to um how government governments think about kind of economic policy and especially social policy because very often I mean you are actually in the episode itself you gave an example where it's kind of policy can really backfire right policy that is well man that's that's meant to make things more fair at least feel more fair kind of severe uh economic repercussions and but you can also get it up the other way around so there's a web and economic policy that then creates kind of very unintended social consequences and um the Mev field is is a Minefield in my

opinion with like problems like that because all of the solutions that seem you know the most easy the most attractive um they all get the systemic effect that I'm very glad that we touched on this in our first episode where we uh to introduce some of these themes right but I think some of the most uh yeah simple approaches are also very reductionist in the sense that they they do their kind of Band-Aids that that make it seem like the problem of Mev goes away but really um you're shifting this risk to the system level and you introduce a lot of centralization and you really risk enshrining certain Market structures for example around latency around Geographic

centralization that then ultimately um end up killing your system in the long term yeah I would completely agree agree with you and you know the backdrop of of this is yeah I spent a lot of time thinking about this you know just about like he's zooming out and looking at the economy and things like inflation and some people view inflation as a result of something going on in the economy I think you could very compellingly look at isn't inflation as a political decision as well um so it's it's just it's interesting that construction of these systems um before we go too high level I want to dive into the discussion of latency that you had especially with with Justin so I know this is something you're super

interested in going into the season so you kind of yeah what did you think of that discussion I thought that was really interesting I mean when Justin talked about all of the optimizations that they're doing at the ultrasound relay I I knew about some but not nearly all of that um that was very interesting um the way that you you can optimize this I'm sure if we had so when we when we have someone from Solana on in that episode I think or also in the search episodes I think we can tease out some more we can pull that string a little more um about how these actors optimize their their latency um and um yeah it's very clear that it's having a big effect on the supply chain and so I was appreciating that Justin

also had spent some thought about how um how they want to prevent this from becoming a big problem for ethereum so I think I think he said they they are looking into spinning up where for one they want to kind of Open Source all of their knowledge on how to optimize latency so kind of turn it from a competitive Advantage into more like a best practice that everybody can adopt um and then the second was they have this planned for uh spinning up more infrastructure around the world right so I think there's two approaches to this that uh both can equally work one is you just spin up a lot of different instances of infrastructure everywhere

so it's gonna code everybody can be close and then the alternative is to spin up like one piece of infrastructure that is equally far away from everyone right somewhere like so far away and like a Mongolia or whatever that or or in space right um that it's equally far away from everybody and so the relative distances don't matter so much anymore and I think um maybe we could have like pulled that string a little more but I thought it was nonetheless a very interesting discussion I thought it was too um and just just to give folks a sense of how competitive this gets I mean I'm not sure how many how many of you have read sort of Flash Boys the the Michael Lewis book but that kind of gets

into the lengths that people will go to for even very very small you know amounts of latency you know to the point of tunneling you know secretly tunneling up Broadband wires right across mountain ranges or you know there are there are trading firms in crypto today that operate their own microwave Towers this is already here right and um um these parties are monopolizing a big share of the uh The cross-centralized Exchange Arbitrage but also they control kind of the these parts of the Cross domain and maybe between let's say ethereum and and binance uh or or human coinbase and so on right um and um just to date it I mean if I

don't know if you saw the cftc uh what was it uh complaint against Finance yeah um a long a long list of allegations and and one of them was um that that binance is actually secretly selling co-location rights um to trading firms that give them as significant benefits so you really have well latency playing a huge role at all layers um of the stack and really introducing a lot of unfairness um into this system and yeah I think we got to be really careful in how we design our blockchain systems so we don't just end up replicating the exact same Market structure

I agree with that for I think and I'm not sure this was spelled out as explicitly in this episode but just for the the geographic concentration and risk there right so if you concentrate too much of the the critical guts of what what powers let's say ethereum in one jurisdiction than a hostile government or regulator could severely compromise the Integrity of the entire economic system which is not what you want yeah now the the uh the just digging into one more question about that the latency aspect there just to make sure I'm 100 understanding this correctly right when the the reason sex to Dex Arbitrage is so big is because obviously centralized exchanges like a coinbase or binance you're you're updating price

within sort of microseconds right so you can almost imagine this like moving price range that binance has and then these arbitrageurs right are are like the you know the Dex price because of let's say a 12 second block time lags necessarily and what you don't want to do you have this great description of statistical Arbitrage it's not Atomic where you're not taking balance sheet risk in statistical Arbitrage you're taking a small amount of balance sheet risk but you want to minimize the amount of balance sheet risk that you're taking so if you almost you're trying to get this moving little needle here but you want to get it right exactly before the end of the ethereum blockchain which is why that latency is so important is that correct yes I that is one part of it um the other reason is where price

Discovery happens so imagine for example that you have uh you have finance and you have ethereum and you have two decentralized exchanges on ethereum so if price Discovery happens on binance and let's say that the price moves up on binance but it doesn't move up on any so on the two ethereum exchanges then you could only Arbitrage that Difference by taking a balance sheet risk but if um if you're on binance and on ethereum at the same time then you can actually sell the asset on binance and buy it on ethereum at the same time and so that's why you need um you also need preferential access to binance in order to to make that trade really efficient all right let's let's kind of you know conclude here just

talking about privacy and this is something I know that you're extremely focused on in flashbots and also it's just sort of a personal interest of yours Justin was getting in right at the end about this idea of sort of encrypted mempools which I'm sure will come up in in later episodes this this season but what did you think about that the Privacy talk I mean Justin has some super interesting presentations on encrypted mempools that uh yeah we I think I I'd recommend everybody to check out it it's a it's it's not super easy to follow but I think the uh the ideas are really interesting and there would have been so much more to dig into there especially around um kind of the role of and tarun was was

actually teasing that which I thought was great which is the question of what information to reveal versus not so this side this is the idea of like metadata from a transaction for example and so he was saying that you need some of the information to be public for markets to even function and so this is also one of the core ideas that we have at flashbots that users will want to reveal some of their information but not others and they want to do this in a way that kind of gives them the power to decide what they want to reveal when and um uh yeah I I think um this this is definitely a super interesting field also because encrypted

mempool is so encrypted man puts I think would be one of the ways that we achieve uh order flow parity between builders so the this this idea that order flow doesn't have to be exclusive between Builders but as a user you can encrypt your order flow uh um and then um it gets decrypted only when like anybody can use it to do computation on that and like to background it to include in their blog but they can only decrypt it if all of the user's conditions are satisfied and this is one of the core ideas in Suave and it's also one of the ideas behind encrypted mempools maybe in like a bit less sophisticated way um and the real challenge here is how do

you enable uh computation to happen on top of that data before decrypting it and that's why we are interested in uh in using Tes because that's how you can do this inside an environment where you can do computation but the data cannot get outside um there are other techniques that that we're looking into but I think I mean I think um this is the future of Finance in my opinion right this idea that user user data is private but it can still be used in in computation and I think if we can figure out this Paradigm in blockchains then it's only a matter of time before blockchains start taking over the entire Financial system and large parts of of

kind of you know this type of economic activity anywhere I agree with you yeah I agree but you know you know what's cool about that is uh it's always worth and because there are so many ways that it's very easy to poke fun at blockchains about how they're inferior to more centralized systems that we have today and it's very cool to look at areas of blockchains where it's like hey wow this actually is a an order of magnitude improvement over the way that it works today and even it can it can even be in one small sort of an area that seems very small today but will eventually become the norm so I would agree I think kind of the design space around privacy is yeah certainly super interesting and it's going to continue to pop up at the course of the Season yeah it just already thinks that

blockchains are like a thousand terms better at than than their comparative systems but the thing is they are also like in some cases they are a thousand times worse right and so it's a matter of I think leg right keep reinvesting in our strengths that like I think that blockchain has a much greater shot at solving the weaknesses in our systems then it is for traditional systems to kind of catch up to their weaknesses that blockchain is currently ahead and so I think it's not as it's not it won't replace everything like not all activity will happen on the blockchain or something but but I think it's when used in the right way blockchains just can massively massively

lower the cost of trust and um and and trust is an essential ingredient in in any human activity that requires coordination and so uh I think the more we want to coordinate them the bigger the role that blockchains are going to play in the future I agree yeah I trust it's a foundational primitive for Humanity and the way we've constructed Society so um 100 agreed on that now this this next episode just to give audiences a preview a little bit we're going to be talking to Matt Cutler who's the CEO at at block native and specifically we're going to be getting a little bit more into the economics of Mev and in particular um this idea of we really want to pick

his brain a little bit on where we see value kind of accruing in this supply chain sort of analogy right so you know historically in at least in ethereum an enormous amount of Mev has gone to the proposers right but we're starting to talk about you know a lot of you know just piggybacking on the discussion around privacy is uh you know a lot getting some some leverage back to the user and redirecting some of that Mev up to the to the user and at the same time we need to find a way to incentivize Searchers and Builders and relays uh you know Give Them Enough profit so that they're not seeking these alternative off-chain uh kind of contracts so we really want to kind of dig in what what else are you kind of excited to talk to to Matt about I think where Justin today

gave us the philosophical view on how things should work I think Matt is going to give us really the state of how things work today just the empirics because block native is is kind of a way of stood up server of the Mev supply chain okay and I'm very excited also to dig with him into these latency Dynamics and kind of the search I was that are going on because the thing is um these uh these latency races and so on they don't leave a footprint on the Chain itself because only the winning transaction tends to get mined right and and so you don't see maybe the 100 transactions that that didn't get mined and you don't see really the time difference between between those and but

block native does keep track of such things I know they have very rich data internally and so it would be very interesting to see if we can get met to uh yeah to to reveal some of the what's going on at their block Builder and at their relay and um yeah how he thinks this uh this space will evolve completely agreed that very well said all right guys uh obviously this has been a fun one see you next week thank you