Why MEV is the most important problem in crypto

Transcript

[Music] [Music] All right. Hi everybody. I'm Shay from Flashbots and today I want to talk about what I think is the most important problem in crypto. I think that if you kind of look under the hood, you'll realize that a lot of the challenges we're talking about in crypto today are actually all very related. Um, everything from execution quality to scaling to censorship resistance is all fundamentally downstream of one thing, which is how we approach something called MEV. And I'll say a little bit more about what that is

in a minute. But uh MEV in my view is probably the most important factor to advancing the plot in crypto today and and possibly in traditional markets as well. So I want to show you why your biggest problems are actually me problems and what this means we should actually be focusing our time on. I'm going to look at three things. creating new market structures, scaling them and making them global, resilient and fair. I think that these are some of the biggest challenges and opportunities for crypto and under the hood all of them are about me.

Okay, so let's start with markets. Uh I think probably one of the core values of public blockchains is that they give us a way to experiment with new market structures that drive better outcomes for people. So, think about decentralized exchanges that reduce the friction of listing new assets. Takes 12 seconds, a few lines of code to launch a new unisoft pool compared to, you know, weeks of paperwork to list an asset on the New York Stock Exchange. Um, or what about permissionless lending and trading protocols where, you know, a whole market can compete to liquidate bad loans or give you a great price instead of just relying on a few wholesalers.

uh in many ways I think this is one tlddr of defi you know permissionlessness drives more innovation and competition which results in better outcomes for people but how do you actually get executed in crypto you know what do these markets look like I want I want to look under the hood for just a minute uh one of the most popular things to do in crypto is to click this button the swap button uh exchanging one asset for another looks very simple uh but you know what actually happens When you do this, what goes into getting a good price?

Under the hood, it actually looks like this. So, under the hood of DeFi is an economy of bots, tons of different actors that help with routing and merging and landing your trade on chain. And there's also a lot of money changing hands here. There are incentives that apps and users pay these actors to perform, you know, useful tasks um maybe even without your knowledge. uh incentives that we call maximum extractable value or MEV. Uh and there's also money flowing upstream in the form of things like better prices that this market produces for your trades. Um and you probably have no idea this is happening, but it is. You know, your favorite trading apps, Cow Swap, Uniswap

X, run auctions to outsource the execution of your trade to sophisticated parties like solvers or fillers who can solve the hard problem at finding you the best route. and they compete in an ME auction to give you the best price. Uh your favorite lending protocols like Ave and Morpho offer permissionless bribes aka ME to incentive bots who come and remove bad debt from the protocol. These me bots again participate in an off-chain auction to land their liquidation in the next block and capture those prizes.

Uh your favorite wallets are actually privately auctioning off your order flow to improve your executions. For example, if you make uh a big trade, you will often create an arbitrage between the pool that you traded on and the price on other domains. So, most Ethereum wallets today, a vast majority, actually send their users trades to what we call an orderflow auction, uh which lets a whole market of bots compete to backrun you, capture, for example, that arbitrage and then split the profits with you. And, uh, again, the majority of retail flow on Ethereum today is executed through an orderflow auction.

TLDDR, you know, DeFi depends on me. Uh, at their core, all of these new protocols and market structures that we're experimenting with to give people better outcomes in crypto use MEV incentives to get very important jobs done. Uh, and also use permissionless auctions to harness all that competition for the MEV and turn it into things like better prices. Um, those MEV auctions affect everything from how your crypto wallets are executing your order flow, uh, to the price you get on trades to how healthy our lending markets are. And, uh, this actually could be relevant for more than crypto natives, too. A few years ago, the SEC propu proposed a new rule. I think it's rule 615, uh, which would require retail trades in

traditional markets to be routed through an auction. Uh, and for context here, in case you're not familiar, most retail trades today are executed privately by wholesalers like Citadel, who buy your trades from frontends like Robin Hood in payment for orderflow deals. And these POF deals can actually give users sort of better prices than the national average, the MBBO, but they aren't particularly competitive. Uh, so while users are getting a better price, they're not usually getting the best possible price.

Uh the SEC actually found a couple years ago that users were missing out on about one basis point per trade. So I think it comes to like a billion and a half dollars a year because of the lack of competition in peop markets. And the orderful auctions that we've been running for now years in crypto could actually be a pretty cool tool to help improve execution not just for memecoin traders or crypto natives but for retail participants in traditional markets as well. Uh but all of these new market structures again depend on me. MEV I think is at the core of making DeFi work today and could be at the core of improving tradi tomorrow.

Okay. Now I want to look at a second challenge which is how do we scale these great new markets? How do we on board the world to crypto? Uh scaling is obviously a huge topic these days especially for Ethereum. you know, L1 and L2 scaling are at the top of the new EF road map. Um, obviously these things are also very important to teams and rollups building on Ethereum. And I think that these priorities are right. But I think that many people are misdiagnosing the root cause of our scaling problems today. And because of this, we're I think at risk of overlooking a key part of the solution, uh, which is ME. My golden rule of MEV

is if you put free money on the internet, people are going to spam you for it. So if you offer permissionless bribes to bots, a lot of people are going to send you a lot of requests trying to win those prizes. And it can be worth it for me bots to pay for hundreds of failed transactions if they land just one that captures a juicy opportunity. And unfortunately, if you try to handle all of this competition on chain, those failed transactions eat up tons of resources on your blockchain. Uh, this was a major problem on Ethereum L1 several years ago. So, MEV bots were cluttering the chain to try to compete for opportunities in DeFi, which increase the gas price for regular people, raise the load on things like

the P2P layer, and clogged up block space. So, there's tons of failed transactions that now every node has to just re-execute in perpetuity. Um and the spam issue actually fun fact was one of the inspirations for uh founding flash bots. So the first product we locked we launched back in I think 2020 like five years ago uh was an auction called MevG that moved all this competition for me offchain. Uh but as Ethereum scales as more economic activity moves onchain as rollup comes online uh this actually just creates more opportunities for me and so our spam problem grows. uh today sort of five years later history is actually repeating itself. So uh we took a look at the data here and spam

transactions from MEV bots account for half of all the gas used on top Ethereum roll-ups today. There was I think one day where MEV bots were using something like 90% of the gas on UniChain. Fun fact. Um, and this this roll-up case is particularly notable because some of the unique properties of roll-ups. Uh, the spam is more intense. It has a kind of compounding effect on the rest of the ecosystem and I think directly conflicts with some of the unique goals that we actually have for layer 2.

Okay. Why is the spam worse? Well, this is what a spam transaction looks like on L2. Uh, and if you are an Etherscan sleuth, you you might notice something that seems kind of off, which is that this isn't doing anything. Uh, this transaction is quite literally not making any state changes on chain. It's just sort of paying gas to read some state. And we we don't see this super often on L1 because gas fees are expensive enough uh that it does not make sense financially to run your bot on chain. uh but it does on L2s for a number of reasons. One big one being that because the fees are so low uh because gas is so cheap, searchers can

basically afford to hammer every block with a bunch of speculative transactions. Uh we've seen bots I think use something like 650 times more gas than they actually need to execute just one unis swap v3 arb and creating more gas does not solve the problem. Uh this winter base I think doubled their gas throughput. Pretty cool. It was like 15 to 30 uh megs per second. But the problem is that the effective gas throughput for users stayed the same because almost all the new uh throughput was captured by spam. Uh as we make gas cheaper, searchers just use more of it to express more complicated logic and run even more complicated strategies.

uh I think you can't engineer your way out of this. Uh the limiting factor in my view to scaling L2s today is not throughput, it's economics, it's me. And the other thing is that this sort of me problem on L2s, all this spam, well it impacts the rest of the ecosystem, too. All of these failed spam transactions, they take up blob space as roll-ups post data L1. And so we have to kind of pay for this not once but twice. Um, and I think this is just important to call out because everyone's talking about raising the blob limit. Well, L2s also need to invest in mitigating the spam on their chains. And I I say the whole ecosystem because one rollup spam problem can impact not just the L1 uh

but other roll-ups too. So when there are huge spikes in demand for blobs, uh the blob fee markets go haywire and create all of these headaches for people. So, uh, there have been some very funny episodes where airdrops on L2s caused these huge surges of demand for blobs on L1. And actually, I don't know if you can read this, but, uh, they made the blobs more expensive than call data, which is what the blobs were supposed to replace. Um, and all this congestion can just take hours to clear. Okay. Finally, last last thing here is that spam really limits the kinds of applications we can build in crypto. Uh I think one of the promises of roll-ups is to create so much capacity that fees approach zero for regular users. Um and

you know the idea is in a world where users are mostly price insensitive you can build many many more novel use cases I don't know microp payments agentic flows onchain social networks right uh but the spam from me searchers creates this persistent elevated base fee not just for the bots but for users too and we really really need to solve this problem if we want to onboard more kinds of applications to crypto okay TLDDR are scaling is an MEV problem. Scaling L2s, scaling L1's, scaling the kinds of applications and markets that we can create.

Uh, fortunately, I actually do think there's a path forward here. We have we've run into this exact problem before. you know, five years ago when we introduced off-chain ME auctions to mitigate the externalities of spam on L1 and since then more and more DeFi protocols have also started using off-chain ME auctions to improve user execution as we saw earlier. So I think we can and it would be a very good idea if we kind of applied that same primitive the the off-chain me auction to layer 2.

uh which brings me to the final challenge I want to talk about which has to do with the core properties of public blockchains themselves. One I think theme you maybe have noticed or maybe picking up on throughout this talk is that we have actually been talking very little about what happens in the protocol because in an ideal world me happens offchain. Um, there's an obvious scaling reason for this. We were just talking about if we want to prevent spam, we need to move the competition for MEV to off-chain auctions. And there's also an executionbased reason for this. So if somebody knows that you're making a trade, they can frontr run you to make a

profit and give you a worse price. So if we want so we really want to keep ME bearing transactions private before they get executed on chain. Uh, but famously public blockchains not private. So if we want our MEV auctions to be privacy preserving, they need to be run on sort of different confidential infrastructure offchain. Which means that for most people, a huge part of the experience of using public blockchains does not happen on public blockchains. uh if your transaction is MEV bearing you know trades liquidations any really DeFi activity uh most of your execution happens or is determined

before your transaction reaches a validator uh and the majority as you can see of of transactions on Ethereum today and actually for many years are me bearing and go through these private channels. So I think that this means when we talk about the properties that we want Ethereum to have, you know, properties like censorship resistance or neutrality, fault tolerance, permissionlessness, uh you can't actually just stop at the protocol. You really need to embed these properties in our off-chain me auctions. And there's actually probably a whole talk on each of those properties. But uh in the interest of time, I just want to talk about one that I think is really interesting. Um, arguably one of the most important values that Ethereum has

is censorship resistance. So, we're trying to build a platform that can credibly and neutrally resolve competition between untrusting parties, say trading firms that don't really trust each other or institutions in different countries that have different regulatory regimes and don't really have a shared system for resolving disputes. And a key part of making that work is making sure nobody can prevent somebody else from transacting. Uh but in my view most of the conversation about around censorship resistance today is focused on what happens in the protocol. Uh designs like inclusion lists and fossil while great focus more on how proposers can ensure that every transaction lands on chain.

Uh but that's not good enough actually for me. Uh MEV transactions need to remain private. So a mechanism like Fossil where you know you're focusing only on publicly known transactions doesn't actually work and the kinds of transactions that produce me sorry things like like trading are often very timesensitive. So you know actually most traders don't want to be included after a certain time stamp because it'll give them adverse deduction as their prices move. Um and on top of this there can be a massive incentive to censor me bearing transactions uh which I think is not really present in other cases. So uh imagine for for the sake of an example

imagine some me opportunity is worth 10 ETH. Uh that means actually there is a 10 E incentive for somebody to censor the competition. uh because if nobody else can bid for this opportunity, then you can basically pay almost nothing to win the auction by default and earn the 10 ETH. And this is really different, I think, from transactions that are doing something simple like, I don't know, transferring some ETH from tornado cache. Uh sure some nodes might have compliance policies that prevent them from processing those transactions but that's really different from proactively competing and stopping uh transaction from landing to go and capture me. Uh all of this means I think we not only

need a really different approach to censorship resistance for off-chain markets uh which are you know more private and fast by nature but actually we need a stronger form of censorship resistance that can handle the sort of adversarial behavior that specifically exists in ME. Okay. So these are some of the properties that I personally think we should be talking about more and I want to nerd snipe you to start thinking about them. Um, first of all, how do we make uh not just the settlement and execution layer, but the networking layer censorship resistant? So, can we do things like anonymize traffic so it's impossible to selectively censor bids in an auction? Uh, second, how do we raise

the economic cost of censorship? um what would it look like to maximize the number of auctioneers and architecturally decentralize our me markets so that the cost of censoring bids in the auction is much higher uh because now you have to you know bribe many many more people to ignore your competitors uh and third how do we distribute these auctions around the world so that faults or preferences in one region you know don't create an advantage or a risk for the rest of the system um and and is it actually possible fun question is it possible possible to do this to geodistribute our infrastructure for MEV in a way that also meets the latency requirements of MBV use cases. Um I think these are all

really interesting questions. I would argue that an Ethereum without these things an Ethereum that only works for you know slow value public transactions is not actually living up to its goals. Um but on the flip side, you know, if we can create these fast, confidential, uncensorable global markets for me, that's a really really powerful tool to help uh kind of advance the vision I think we have for Ethereum. So the TLDDR though is we have to solve for these things in ME in our off-chain markets if we want the majority of crypto users to actually benefit from them. Uh because under the hood, MEV markets are what make DeFi work, what make it possible to scale public blockchains and where I personally think that the strongest

forms of censorship resistance and other properties that we care about will actually originate from. Uh those are obviously very hard problems. I'm I'm pretty optimistic personally that we can solve them. Uh and I hope that you will join our collective of researchers and developers working on these problems today. Thanks. [Applause]