# L2 Specific MEV Mitigation Strategies by Joseph Poon | Devcon SEA

- Speakers: [Joseph Poon](https://streameth.org/speakers/joseph-poon)
- Channel: [Devcon](https://streameth.org/devcon)
- Date: 2025-10-09
- Duration: 24:50
- Topics: Science & Technology
- Watch: https://streameth.org/watch/yt-IXV3yFXHlfo
- YouTube: https://www.youtube.com/watch?v=IXV3yFXHlfo

## Description

MEV mitigation and prevention has primarily been researched in the base L1 Ethereum layer. This talk explores L2 specific strategies, including the future in the event of decentralized sequencing. We explore emerging EIP proposals and drafts (EIP-7640), the use of intents in L2s and other new constructions.

Speaker(s): Joseph Poon
Skill level: Intermediate
Track: Layer 2
Keywords: Layer 2s, Rollups, MEV, defi

Follow us: https://twitter.com/efdevcon, https://twitter.com/ethereum, https://warpcast.com/devcon
Learn more about devcon: https://www.devcon.org/
Learn more about ethereum: https://ethereum.org/ 

Visit the https://archive.devcon.org/ to gain access to the entire library of Devcon talks with the ease of filtering, playlists, personalized suggestions, decentralized access on Swarm, IPFS and more.

Devcon is the Ethereum conference for developers, researchers, thinkers, and makers. 
Devcon SEA was held in Bangkok, Thailand on Nov 12 - Nov 15, 2024.
Devcon is organized and presented by the Ethereum Foundation. To find out more, please visit https://ethereum.foundation/

## Transcript

[Music] it as a community and why ethereum is well positioned to do that Beyond other efforts L1 chains Etc um the general approach is fairly simple um you know we're talking about you know perhaps below 50 lines of code encompassing you know everything we're talking about here and I genuinely believe that simple designs are what are effective and often times are you know the nut of the problem right it's very fractal you want to keep things simple initially and then some complexity comes out later I think we've over stacked the complexity too hard today initially and I think we should take a look at everything from fresh eyes ultimately what is this about this is ultimately about building the world Ledger right what is the world computer today it's the Global Financial system and you know that basically determines how much wheat me make next year you know when when you know the the satellite feeds start showing you know a lot of brown Fields you know the the prices start going up for September delivery and then Farmers start planting more that's effectively a global computer and it's driven by a ledger and the opportunity is for to you know drive that through crypto blockchain Etc and we ultimately need to be thinking about the opportunity of growth above any type of s short-term uh yield generation or value maximization because that's really the promise you know in terms of serving Society I believe most MAV is pathological you know it's a chronic issue that we're facing um if we believe that we want Grandma to be using you know these types of technologies that we're all building um Grandma's not going to have fun getting sandwich and she's probably going to hate it more than you do um these types of activities are not permitted in trafi um we need to be putting effort into fixing this algorithmically and Via incentives rather than you know through legal means obviously and that requires some effort in terms of our community to try to fix the problem as opposed to live with the problem because they're not going to get on board without actually resolving these things um let's look at what's going on in The Wider ecosystem in terms of an L1 heavy approach right me is over half the transaction costs paid by users in aggregate in salana even ethereum is nowhere near that L1 U my thesis is that the users have a choice to go to l2s and then there's there's uh there's parallel choice in terms of that um and you know to properly compare transaction costs between these two chains requires including the total roundtrip costs including aggregate me token terminal recently came out with these stats don't really look at how high the bars are because the the um you know the axes are different um what I want you to really focus on is the green versus the the pink and purple right what we're what what what this is really showing is that you know the green is effectively me activity this is not an accurate representation just it's fairly loose because you know how you can you know properly measure it is somewhat subjective but what you're looking at on salana is that me is over 50% of the the value capture this is a very very very important point to look at um on eth you know it's somewhere between you know let's say 20 25% something like that right but what's interesting on the salana side is that it's growing now why is this really important it's that me cannot be efficiently tuned it's a very binary issue right either you have it or you don't it's not like transaction fees or gas fees which you can adjust right you all you can do is okay I'm going to raise the bo a gas cap I'm going to Raise You Know The Blob size I'm going to reduce it or increase it and and by increasing or tuning that by let's say 25% well you can see transaction and gas fees Go Down cmeasured based on capacity and demand M there's no knob for mvv like this so as you start scaling up l1s and l2s me is going to perhaps remain at minimum remain static but more likely capture a larger and larger share of the roundtrip cost for users and this is a big problem because chains with pathological me is going to be less competitive long term it's not something that you can effectively resolve unlike unlike things like you know how how how much how much throughput you want on your chain so that brings us the question is ethereum positioned to effectively eliminate me detectable me I think we've already solved the problem to some degree you know I we've been working for the past several years in ethereum um inadvertently or or intentionally depending on you know who you're talking to it to resolve me um you know I had a talk with like um with like a you know large VC Focus the ethereum focus VC and it was like do we really want to be talking about this because a lot of people in you know in in in the l1s are sort of misguided in terms of this approach but you know it's you know we're open source community so let's talk about it right again this is very very simple right it's very very very simple if you have no no l2s in an environment with no me it's straightforward if you have me it's it's pretty complicated you have no choice it's not it's not complicated you have no choice right when you have an environment with l2s like we do in ethereum if a rollup starts performing me and might I remind you effectively no rollups today are performing me a lot of people are talking about it EV like maybe half or more of the rollups are talking about it none have done it why because the moment you do it people are going to switch over like you need to be thinking in terms of game theoretic incentives and not complex algorithms what you need is parallel sequencing and that's what l2s and invert get you it's really that simple so I genuinely competitive parallel sequencing solves pathological Ms you know these types of sandwiching type of attacks um multiple l2s are actually multiple sequencers this brings about sequencer Choice um if one starts conducting MAV um users will Route Around damage there's a credible threat of routing around this damage um ethereum l2s are actually an me solution and it's not just for scaling if you want to be solving parallel sequencing you're going to be rein Reinventing l2s that's what's going to happen um I genuinely believe that projects such as salana and other emerging projects that have an L1 specific Focus um will basically have meev dominate their their total roundtrip transaction costs um and approach and and approaches to resolve this will basically reinvent the wheel right like so if you're like okay let's just have some type of you know parallel sequencing on the L1 and not do this L2 thing well then you have you have some type of conflicts that arise between transaction ordering with contracts that overlap right you're going to have disagreements of that um and then it's going to be like well let's SE separate these contracts into separate name spaces well you've just reinvented elt right with with less performance um the work that the ethereum community has put into l2s rollups Etc um requires years of investment across contracts use their behavior wallets infrastructure and it's an ongoing process but you know this has already been years in the making um you know other efforts such as tees don't solve this problem either um tees you know you might as well locate collocate all your validators Intel Intel's data center um the sequencers can still control the input you know it's not a credible solution forced inclusion basically forces collusion right um you know like if you look at how me proposers work today you're basically dealing with a handful of entities that collude with each other it's not a credible solution parallel sequencing solves this in a much more credible way I'm much I much prefer you know a choice between 10 different L2 sequencers rather than you know a cabal of let's say three or four proposers right what is this mean if you're working on an L2 you should publicly commit to refuse to do pathological e everyone is already refusing to do pathological me but it's a competitive Advantage for you to do so um and what do I mean by that why do I think this is important not just in terms of getting rid of this problem um but I think it's also important because if you don't acknowledge that this is the incentive and the state of a lot of l2s the if you look at where the where the sort of how this is going to play out um I really think um entities such as you know base and Inc and OK xchain okx chain are going to be dominating because their incentive is to play the long game their incentiv is to say hey we want to get users we want to get trafy board we want to get the entire world on board we don't care about this me thing because it's relatively you know dimin comp compared to the real opportunity um if you're operating in L2 you need to at least you know commit to be on the same level in the long term in terms of acting in your own users interests what else can you do um you need to reduce switching cost between l2s in order for this to be true choice you need to ensure that the choice is tangible effective and and and credible right so I think there's a shift in The Wider space right now from amms to rfqs intents and Order books this is the effort that I'm particularly interested in but it it's it's something that you're going to be seeing in the next year and what I like about this is this breaks the Monopoly and network effect power that amms have the amm has significant Network effects where you know the cost of sort of switching from one L2 to another is somewhat tangible I think over the next year as you see efforts move in these directions towards newer Technologies split liquidity is not going to really be an issue um but what's interesting is that in order for these new things to work we actually need revert protection which nearly no um no L2 supports right now revert protection is important I cannot emphasize this enough everyone actually wants revert protection um you need it for the new emerging systems and the me guys want it too right the meev guys want it because you know um they they they need it in order to do their effective auction system and I'll get into that in a second but let's look at rfqs in tents and Order books right currently RFQ spreads are wider on L2 um there's a greater risk in terms of using these types of intense constructions on l2s because transaction contention is a big problem you know if you have several transactions trying to get in only one of them is going to get in right and what in in order to do revert protection essentially all you need to do is the sequence Ser even the RPC server just says if this transaction reverts I'm not including in the L2 block um speed helps a lot rever protection helps so much more um in my opinion decentralized intents account abstraction are basically broken without revert protection um this is easy to do on many l2s they should be doing this immediately um there's been a paper recently from the me guys about you know um you know it's called quantifying the value of revert protection um they're effectively talking about how to maximize me um on these l2s I I'm I'm personally I I think their math is solid it seems reasonable I think the issue is more about eliminating meev rather than you know optimizing it um and I think you know hey everyone wants this if you're doing an L2 do revert protection ASAP it's very very very important um if you want to be doing revert protection on the L1 in a decentralized way um me and a couple friends have a proposal EIP 7640 it hasn't been merged poll request 8267 about how to do uh revert protection on bl1 um this is relevant because you know I'm mostly F I genuinely believe a lot of tracks move to L2 I think perfectly eliminating meev on l1s is borderline impossible um but there are ways to sort of mess with them there are other techniques such as um Shifting the state state tree uh proof to the next block and then you allow multiple um transaction routes um that'll also mess with the Meb on L1 there's no proposal for that but there's a lot of things you can mess with them on L1 But ultimately the real battle is going to happen on l2s um but in order to understand why revert protection is interesting and I cannot emphasize this enough is really how do you think about me right and this is going to be fairly obvious um you already know this but I'm talking about a mental model and how to look at the problem from a different angle right everyone talks about Searcher sequencers Builders proposers what is me right like there like that's sort of what everyone's looking at but the first step to understanding me is actually revert protection that's why everyone's like pushing for it right now even the me guys um you know you already know all this but you know let's say it's 2022 you want nft five people are trying to get it broadcast the transaction only guy gets it what happens to the other four guys right their stuff gets reverted we had this because we had a public M Pool and there was you know a Doss attack Vector but if you have a centralized sequencer this is not a problem this should not happen right um if you have decentralized sequencers there's ways to mitigate this as well um it starts looking a lot like the 7640 type stuff um but um but effectively you know if you don't include anything that gets reversed the one gets in the other people don't pay anything right this is key to actually how the me people think because they're focused on the fact that in an auction system there's one winner and four other losers right those four other losers trying to sandwich attack people need to not pay any money because if you're an auction and you pay anyway when you lose this is not a functional system this is the core of me design revert protection is the core nut of how everything works and this core nut is also necessary for things like intense rfqs and Order books in order to break this type of like Network effect strangle hold and basically the point is is that if you're doing some type of intent or any of these types of new things what happens is okay I'm feeling an intent right if I'm fing an intent and four other people are trying to F this intent I'm going to lose money if I'm not the winner right like a lot of the intense people like are not focused on this this is the main issue to look at right now this is the core thing this is the dialectic this is the conversation this is the root of it all but wait what if we move to decentralized sequencing what happens then right we need choice and sequencer in the sense that what you need to do similar to the prior example is just extend it down a choice within the L2 as well if we move to if we move to multiple sequencers within an L2 then all we need to do is say okay I'm going to have a contract within this L2 that says which sequencer my transaction can get included in if another sequencer includes it that I haven't authorized then that's an invalid block on the L2 right and then they get slashed or whatever it may be right it's that simple you don't need that complicated math all you need to do is give the users choice right so if one L2 this is the same L2 roll up if there's several validators if one of the validators starts acting pathologically then you say that that that that validator can't include my transaction right it's not complicated don't do this on the L1 this is going to be centralizing if you if you do this on the L1 because on on the l2s the set of participants tend to be smaller in in a lot of these you know if you do decentralized sequencing and also if you um you know the fact that like the entry and exits are going to be less less frequent um so I genuinely believe that meev is poised to dominate transaction costs from any l1s without l2s um ethereum is already very very well positioned for mitigation get itsme due to Prior and current hard work that the community has made um via you know building the l2s new market exchange mechanis mechanisms and wallet designs if you want to be solving me this is the effort that needs to be done ethereum has you know five six years of effort on this you know like if you want to be replicating this you need to be playing back this effort and that means that ethereum's L2 work is at the Forefront of the current crypto dialectic you know this is basically a conversation that's being had between people that want to extract value from the system and people that are defending against that for the users and this effort is happening on ethereum now I think that it's stacked maybe 10 to one in terms of like who's fighting for what right but I think that basically the you know this is going to play out in a way that you know you can't really game out against you know you can't create complex things against once you give the users choice and maximize user Choice which ethereum is already well on a trajectory to do so ethereum has a very very strong advantage in order to do that and I think it's very important work because you know we're all trying to Build a Better World on chain for everyone and I think that's the real opportunity thanks thanks for that great talk we have a lot of questions so we're going to try being quick to get to more of them we can combine the first two which is what is revert protection and does it open a do attack Vector on the sequence sir okay so um it's easier for me to go number two and then number one obviously um so revert protection is essentially if your transaction doesn't go through like like in the in the uh nft example right someone giving away an nft and all you need to do is just pay the money to buy it onchain um there's only one of these um one everyone broadcasts at the same time one guy wins the four guys loses um right now the current status quo if you just broadcast it on the public m poool is that you will be paying gas in order even though you don't get the the thing you wanted and this is an the reason why this was done in you know 2015 2016 right like this is in the beginning of ethereum is because there was a Doss of hack Vector in the M Pool when you when you propagate the transaction across different computers if you don't attach a cost to it then you can keep spamming things and then eventually you know like the system starts starts breaking down um so for the first question this is why it was done if you have a centralized sequencer this is not a problem if you have a decentralize and and I would say me Services by the way also think this is not a problem because me services are basically doing this revert protection themselves and if this was not a solvable problem no proposer would exist right definitionally the fact that proposers exist means that this is a solvable problem you can do things like flow control you know like saying like hey you can only do a certain amount if you have a certain locked up per address it doesn't matter the specific mechanism all I want you to understand is that Reverb protection exist on these me Services right they're not getting dossed out this is all very solvable and you there are various techniques um uni chain I don't know if that's a question um but maybe your thoughts on uni chain uh and its approach to solving me um so I guess I can combine the next two um I'm not particularly convinced that te is a credible approach um I think the unich chain they're doing a lot of great work right uni chain for example has proposed revert protection um they're the first ones to propose it um I think that's a lot of good work um I'm not convinced te is a credible thing like you might as well just run it at Intel's Data Center it's not decentralized it's effectively a version of PayPal the the interesting about thing about tees is that what are these things for right like these are basically like kind of like Hardware wallets these constructions when you when you put it on Prem on in the premises of a user that's adversarial to you they always get broken that's why DRM always gets broken right this has been like 30 years of History this stuff gets broken when you give people a box and and you don't trust them right these these these boxes are great when you don't trust yourself right why do you H why do people buy these like Hardware wallets because they is it because they don't trust someone it's because they're they're afraid of themselves screwing up that's what tees are great for it's not because you're afraid of some other someone else trying to screw you over it's when you're afraid that you're going to screw up yourself um how is rort protection relevant for sandwiching so essentially H this is such a okay um so so thinking about how to answer this um so when you have when you have this fight right over you know this this uh this nft example there's going to be four people four people losing and one person winning this Maps out also to something that someone can take advantage of someone can sandwich this one transaction and five people want to do it four people are losing out on it and ultimately the four people losing out on it will change their behavior if for example they're paying a cost to it you kind of don't want them to pay a cost to it uh is what the Meb people ultimately are thinking about um now revert protection is also needed for intents right and intents are necessary to you know remove these certain types of network effects and switching costs um but if you have this question we can we can talk a little bit later about it as well and last question uh how do we know the growing meev proportion on Solana is because of design rather than mcoin meta I mean obviously everything is a mix of it um you know the memec coin meta is Alive and Well on ethereum as well um I would argue that you know um ethereum memecoins are are fairly just almost as strong as the on salana when you view it as you know able to sustain you know a large market cap right um but rather than you know the the Zero Sum cycle um but uh yeah I mean that's a factor um but you know that that doesn't preclude the points of you know me uh meme coins rather are a large portion of Bas do you see me as part of base no right it's it's not a complicated argument thanks Joseph you can give it up for Joseph thank you so