The maximal extractable value (MEV) problem, if left unchecked, could represent an existential threat to Ethereum’s core values. One might argue that it is a natural result of solving market inefficiencies, but it entices to centralisation and collusion in order to extract the highest amount of value. Additionally, through front-running and back-running, it creates unnecessary overhead, which caused up to 2.4% of the total network congestion back in DeFi summer of 2020 (according to Flashbots’ research). As current infrastructure cannot eliminate MEV altogether, Flashbots and other research groups have tried to come up with solutions to address its aftermath.

We were joined by Stephane Gosselin, co-founder of Flashbots and Frontier Research, to discuss the MEV landscape, what research breakthroughs have been recorded and how they impact DevEx and UX.

Topics discussed in this episode:
0:00 - Introduction
0:43 - Reflecting on ‘Frontrunning the MEV crisis’
9:20 - Decentralising the MEV ‘supply chain’
11:40 - The vision behind Frontier Research
14:58 - Request-for-Quotation (RFQ) explained
20:18 - Block building
26:09 - Uniswap X and how it might impact LPs & block building
29:36 - Decentralising block building
37:40 - Proposer-builder separation (PBS)
45:10 - MEV burn
51:43 - How MEV influenced different applications’ UX
55:14 - Intents

Links mentioned in this episode: 
- Stephane Gosselin on Twitter: bit.ly/3PjzDoy
- Frontier Research: bit.ly/3PlJZV3
- Frontier Research on Twitter: bit.ly/3Z2OAPl
- Flashbots: bit.ly/3EmdIqz

This episode is hosted by Brian Fabian Crain & Felix Lutsch. Show notes and listening options: https://epicenter.tv/511
#MEV #flashbots #ethereum

Welcome to AppCenter, the show which talks about the technologies, projects, and people driving decentralization and the blockchain revolution.

I'm Brian Frane and I'm here with Felix Lutsch.

And today we're going to speak with Stefan Gosselin.

Frontier is a sort of new MEV focused company, but he was also previously 1 of the co-founders of Flashbots and I think sort of the lead architect, I think if that's correct, or product designer or something like that at Flashbots.

So thanks so much for joining us, Stefan.

It's always a pleasure to get the chance to come back on this podcast and chat with you guys.

Actually, maybe you can just get started in there.

So 2020, you wrote a blog post on ETH Research titled Flashbots Front-running the MEV crisis.

And I'm curious just if you can like reflect a little bit sort of looking back, I mean, you called it back then like MEV crisis.

I'm curious, what was your understanding of the MEV crisis and what does it look like today?

I was forced to like think back on this actually earlier today.

Someone asked like, do we actually have a counterfactual to PBS?

Like what would have happened if we hadn't launched MF boost?

Would we actually have a significantly different distribution of concentration on the validator side?

The MEV crisis was basically the thesis that said, if there wasn't action taken by the community and early Flashbots members who sort of felt responsible to take this action, there would be an increasing concentration of hashing power because the level of activity on Ethereum was increasing to the point where MEV became meaningful.

If you remember during summer 2020, DeFi summer, we had all of the start of the yield farming and like really Uniswap getting traction for the first time with all the liquidity mining programs.

It was at YAMM finance and all this good stuff.

And so it was clear that, you know, whether it's token sniping, arbitrage, whatever else, there was an advantage of operating this at the validator level.

And validators weren't currently doing it.

The activity was happening through the transaction pool, through gaming the algorithm that the miners run for packing blocks.

So the crisis was, well, it's very clear that miners have the advantage in doing these games.

They are likely to start to invest into running things themselves or run custom clients, make partnerships with trading firms to be able to extract some of that value.

That has the potential to lead towards more concentration because there's increasing returns to being the ones that play this role.

Maybe there is a better mechanism that we can develop and launch that enables us to maintain transparency over what is happening in the system and maintain open access to it.

So make it so it doesn't require a special deal with validators, but you can freely connect and participate.

And finally, have some way to redistribute the value to the actors in a way that's like welfare maximizing or whatever definition that you can have.

You know, you sort of mentioned like, hey, you wrote this post, you made it on E3 search.

And I'm like thinking back to like the last 3 years of my life, and I feel like all I did was write 2 posts And like everything else was like fluff, right?

Like I wrote this front running the MEV crisis, which was like a spec for Mavgeth, which was like the client that we developed and then ended up getting adopted by miners.

And the second 1 that I wrote was introducing MevBoost as a spec.

And then maybe like I should have just written those things and just like, you know, went to the beach and did nothing else the rest of the time.

Because those specs alone, I feel like moved the needle into what the infrastructure looks like and were the most impactful things that I really did.

So yeah, there's this really fun progression from there not being any infrastructure to now having gone through 2 major cycles of the MEV infrastructure changing on Ethereum.

And do you feel like these, both of these things had the impact that you hoped it would?

I mean, it's like it achieved the like the outcome in terms of like, it's a system that got adopted and like everyone got excited about it and, you know, introduce a new industry and ecosystem with a lot of people that are having open discussions about these things and implementing alternative solutions to them.

I mean, 1 can debate whether it's a crisis at all, if it ever was actually there, or if it's just like a natural way that an industry evolves and whether it was MEV-GETH, MEV-BOOST or something else would have happened, the end results would have been the same and we would have ended up in the same location.

At the end of the day, I don't know that the amount of concentration that we have at the stake level today is meaningfully different than what we would have if we didn't have PBS or any sort of MEB infrastructure.

Maybe we have more transparency in how the system works because we have this open pricing mechanism where the information is being routed through.

We have a dollar value for the price of blocks and we have some nice dashboards that allow us to show it.

Maybe there was some path dependency there that's different.

In terms of like how things are evolving, it seems like pretty inherent to the way that blockchains are designed that the end result will end up being very similar.

Maybe just commenting here 1 thing, at least I do feel like reasonably confident that MEV did not increase validator centralization in a meaningful, like was not a meaningful driver for validator centralization on Ethereum.

So I think from that perspective, if you take like, you know, concentration on the validator level, then I think if that's the kind of criteria, then I would say like, yeah, no, I think it worked out right.

Or it happened to some extent for other reasons that were unrelated to MUE, I think.

if it was necessary, but we know that it didn't fail.

I think the thing that I'm the most excited about in the end and in particular with the introduction of MevBoost, like so MevGeth was like a really simple solution.

It was like get FlashBots to run a server and then route all the MEVs through this server and then multiplex it to the miners.

And it would just introduce Flashbots as a server that runs the MEV market.

MevBoost is a much more open system where you have a lot of different layers to it and a lot of room for individual entities to compete and offering additional services.

So just the introduction of the block builder role and the relay role has sort of meant that this thing of operating an MEV market that Flashbots was doing now is being done by like, I don't know, 678 companies.

And there's quite a bit of turnover, I think, in the dominance of these parties.

They're trying to figure out how to make it sustainable, how to have a revenue model, and competing for features that they develop.

I think that's all very positive because it brings a lot more diversity to the way that these systems are being iterated upon.

And just the number of different perspectives that are being reflected into the architectures that they get adopted.

If nothing else, just having more different stakeholders that are involved in developing and competing against each other for developing this infrastructure, providing services on top of it, that seems quite a decentralized way to approach infrastructure development.

So it's almost like the third thing you did, right?

I guess formalizing this MEV supply chain with the different actors and how you can break them apart and how you can separate the roles.

Did anything change in that perspective from how you looked at the MEV supply chain during MEV boost and nowadays?

Or is there anything you haven't foreseen or something?

I don't think there's anything that I haven't foreseen.

I think what I've been trying to introduce as a reframing of the MEV supply chain is to think of it less linearly and think of it more as a chaotic set of games.

I presented a little while ago and we wrote this blog post with Frontier about infinite games and introducing this idea of a transaction supply network.

So instead of thinking of MEV supply chain as being like users originate MEV and then it sort of gets sort of funneled through this system of extraction all the way to the validator.

Inside do you think about a cluster of a bunch of different infrastructure component that yeah it originates with users but depending on, I can say they're intense, like whatever they're trying to achieve in interacting with the chain, they get routed through different sets of specialized infrastructure that leverage tools like auctions, that leverage tools like privacy to provide those services and capture that those systems, reflect that value back to the user and provide the utility that the user is looking for.

It's kind of like this thing to say like, okay, MEV is like a lens through which you can see the world, right?

And once you start to look at like bridges, you look at exchanges, you look at all these different things from the MEV lens, it's like a toolkit that allows you to analyze them and understand how do the dynamics play out if you set this up in a decentralized setting.

I think it's useful to have this lens, but still look at all these components individually, all these games individually, as opposed to say, like, there's only 1 game here and it's like the MEV supply chain and like, that's the only thing that matters.

I'd rather look at individual components.

Can you talk a bit about your vision for frontier research?

Like what do you hope to build as an, like what kind of organization you hope to build and what kind of impact do you hope to have on you know the MEV landscape or crypto more generally?

And I took some time off and decided, like, is this something I want to continue working on?

I felt like there was still some contributions I had to make to the space.

And so I started sort of a research organization around this that helps working with various different teams that are participating across the entire ecosystem of MEV infrastructure, help them sort of uplevel their understanding and mental models around MEV, help do some analysis over the system that they produce and advise on what market and whatnot.

So we've been doing that for quite a few months now, since the beginning of this year.

Now we've started looking more actively at where are the opportunities for developing products.

We've started incubating a block builder called called faith builder That participates in this game.

I sort of see very abstractly the set of games as being an abstract set of 2 2 roles You either have a message passing system with some rules over how these messages are handled and aggregated, and you have individual agents that can be called like solvers or searchers or block builders or whatever else that aggregate these messages and then solve them, optimize them according to some objective function.

And so the block builder that we're running we see as being sort of fundamental to enabling the development of all these different games across the ecosystem and participating in them and helping them bootstrap to provide these better user services.

So it's a generalized solver that aims to participate in all of these different blockchain games.

So yeah, that's sort of a venture that we're incubating from the Frontier team.

So this means it's a builder in the FlashBuzz MUV boost ecosystem, but it could also be a solver maybe for Cowswap or other products like that.

They're very similar games at the end of the day.

They achieve similar objectives and they are architected in ways that benefit from participation in multiple of these games.

The move of activity and the trend, this is 1 trend that's interesting to discuss, is how much RFQs are going to gain in momentum and adoption over the course of the next 12 months.

Yeah, I can give like, maybe I can give like an MEV explanation to it.

And then I'll give the like the maybe I'll give the user explanation and then the MEV explanation as well.

So The user explanation is you're used to doing trades on top of DeFi by going to like uniswap.com and like requesting a swap.

What happens then is you're tapping onto AMM liquidity that's on chain, but you have a lot of inefficiency in the prices that you obtained.

So the liquidity that's on chain is stale.

It isn't by professional counterparties and you're susceptible to sandwiching attacks and like exploitation of your orders if you routed through there.

So it doesn't give you very good pricing essentially to interact with Uniswap relative to the pricing that professional traders would be able to get by interacting with all the different liquidity venues.

RFQ just allows you to outsource the work of routing through a bunch of different liquidity venues to some solving system.

It says, hey, we're going to introduce a marketplace of service providers that are going to compete to give you the best price.

And they can do the complex work of figuring out how to access liquidity wherever it is.

You have the same experience of doing a swap and getting the tokens that you want, but you can also expect to get better pricing on the trades that you do.

The MEV lens is RFQs like recapture value that's being leaked sort of down the supply chain.

So If you look at the distribution of validator revenues today, over

of it comes from basically arbitrage against Uniswap LPs.

And so Uniswap LPs put a bunch of capital into their AMM pools, and they're relying on CFI, DeFi arbitrageurs to update the prices that they offer relative to the prices on other liquidity venues like Binance and whatnot.

And arbitrageurs are capturing a significant amount of value.

The AMM LPs are losing that value, and this is called LVR, loss versus rebalancing.

There's a service that these arbitrages are offering, which is to maintain prices on-chain.

It comes at a significant cost for the LPs and a lot of that value ends up being captured by validators.

What RFQ does is it says, okay, thank you very much for the system that was generating all the value for validators.

We as the originators of this order flow would rather recapture this value either for ourselves or for our users.

And so we're going to create the system that enables us to internalize essentially the liquidity and provide that value back to our users.

So the sort of trend on this RFQ stuff is for more and more of the liquidity to move off chain, to move away from AMM contracts and be sort of accessible by these professional market makers, these professional solvers that can get better price improvements to the users and better quotes to the users and expose less of that sort of down towards others.

And then in this RFQ example, so let's say I'm on Ethereum, you know, I want to trade some ETH for USDC, for example, then I'm making this request for quotation or something.

And then it was like, Hey, I want, I have ETH, I want USDC.

This also means that somebody could come and say, Hey, I'm a trading company And I can make that trade on Binance and here's how many, you know, USDC I'll give you for this ETH and you know, nothing ever settled.

I mean, of course, they're going to send me the USDC on Ethereum, But the trade actually gets executed on Binance.

That's like, yeah, that's mostly how RFQs work.

Traders don't necessarily need to make the trade on Binance.

At the end of the day, they hold a bunch of tokens on chain and they hedge their exposure on other venues.

So basically it's like they have inventory on Ethereum that they are providing within their market making.

And their goal is to keep this inventory within some range of parameters that's in line with their hedging on the centralized exchange.

And yeah, whenever this inventory goes too far in 1 direction or another, then they lose money, but so long as it stays within a certain range, then they make money.

I feel like block building is something that's very poorly understood, I think, by most people, even probably even Ethereum users, they have like no idea, right, that like Val is not actually building their blocks and this happens in some other way and how that's happening.

So can you just explain what is the, What's block building and what is the block building landscape in Ethereum at the moment?

Yeah, so block building is basically validators outsourcing the job of putting transactions into a block to a marketplace.

Not everyone can be a good block builder because block builder is very dependent on having access to order flow.

So there's a trend here, which is increasing amounts of private order flow that's being routed to block builders.

Interestingly enough, a lot of the discussion there was initially like, oh no, block builders are going to get exclusive order flow deals and become super dominant and it'll be centralizing towards the chain because you'll have just like 1 party that produces all the blocks.

Realistically I think there's a lot of economic dynamics that mean that that's not going to be the case.

What is happening is there's an increasing adoption of order flow aggregation systems.

So more and more of the transaction flow is being routed through systems that say, we will directly send to block builders, we'll send to all the block builders, but we'll directly send to the block builders and never touch the public transaction pool.

And so essentially, if you're a block builder, you have to connect all these different private sources of order flow to be able to build competitive blocks.

You compete on this access of the order flow, compete on the efficiency of the merging, and then submit blocks to validators for inclusion.

There's sort of 2 types of block builders.

There's block builders that generate their own order flow themselves, which you could call integrated searcher builders.

A lot of the dominant block builders to date have been of this kind.

They don't really monetize from playing the role of the block builder.

The block builder is a way for them to execute trades that they would want to execute anyways and have some edge in providing this execution.

And then the second type of block builders are neutral block builders and they're the ones that don't generate order for themselves.

They just aggregate the trading as done by a bunch of other searchers, traders and provide sort of fair execution across all that flow.

And that's like 1 of the big questions right now on the block building side is, is it inevitable for block builders to be traitors?

Or can neutral block builders provide services and be competitive?

So if you think about a business model of the builder-searcher, I guess they make the money through their own searching and the neutral block builder has to come up with some way to make money from the private order flow?

Or how do you kind of attract the order flow just by being a neutral party?

Or, you know, I guess whatever you can share.

Or how do they compete with this like integrated builder?

mean, either you're like doing the trading yourself or you're providing some service to someone that wants to sell transactions on chain.

Let's say that you're a trader, you're considering running a block builder.

Not everyone can get access to all the private order flow that's required.

You have some potential revenue versus cost calculation that says, are we going to do trading or not, or do block building or not?

So they're willing to pay for someone else to provide that service if they don't have to sort of spin it up themselves.

I think this exists with several other systems as well, where they would be willing to submit flow to a block builder if they can get certain guarantees or a partnership on how it works.

I genuinely don't think block building alone is a very good business to run.

It's a very competitive marketplace and unless you're able to get some very special deals basically the profits get competed away.

So the model is not necessarily super attractive.

So what is going to be the future of block building?

So you think there's other products that are going to be built by block builders?

And what do you think are the most promising ones?

I think there's 2 trends that are worth keeping an eye on.

1 is a transition of order flow from public venues to private venues.

In particular, the transition of Uniswap Order Flow from the mempool to Uniswap X, that will influence a lot of different things.

I think that will mean that searcher builders become less competitive and less important, and neutral builders will become more prominent.

And maybe just also explain what is Uniswap X, because I think many people will not be familiar with that.

Uniswap X is like the RFQ platform, basically, that Uniswap launched.

Works similarly to the 1 inch fusion, similarly to to CalSwap in some ways as well.

It aims to use all sources of liquidity to be able to provide price improvements through running an auction with a bunch of professional private market makers for the right to settle a swap.

So what this does is it transitions a lot of activity away from the public transaction pool.

So we can say sort of goodbye to all the Sano Chang, which is great, great for users, but also reduces the opportunity space for those trades to take place.

It makes providing liquidity as a passive LP and AMMs a lot more difficult because you have active market makers that have first look on all the retail orders and the chance to settle those.

And so as a liquidity provider in an EMM, you sort of just get all of the leftovers that for whatever reason, the professional market makers aren't selling.

So you have to ask yourself, why aren't the professional market makers selling this?

And why am I as a retail, do I believe that I'm making money off of this trade.

And so I expect there to be just a decrease in the amount of liquidity that's available on layer 1 AMMs and more and more of that trading to sort of settle on this either private liquidity or layer 2.

This in turn sort of reduces the arbitrage opportunity between the layer 1 and other trading venues.

If you have less liquidity that's available to be targeted, you will expose less value.

Separately, there's Uniswap V4 and hooks and other mitigations are being developed to be able to reduce the amount of LVR that are exposed by passive LPs.

Essentially, all these microstructure changes that Uniswap is introducing reduce the amount of MEV that gets exposed in the system and reduce the opportunity for integrated searcher builders to have a significant portion of the value of the blocks.

So more and more of the value of the blocks is going to be distributed towards other use cases relative to just arbitraging Uniswap.

That will cause the shift away from searchable builders towards more neutral building.

And then how do neutral builders differentiate?

I guess maybe we went into a bit, but is it mostly on this like, so offers they can make for private order flow or what other like dimensions are there to differentiate for a neutral builder?

And what are your thoughts on decentralizing the block builder?

Is this, do you think this is, is it possible to build a decentralized block builder that would be competitive?

I'd say it's, I mean, I don't know that much, I will say, about what decentralized block building actually looks like.

And the designs that I've seen or the attempts at it have been unconvincing.

There's this bigger question mark, which is like, What does decentralization actually mean that needs to be answered when you're even trying to tackle these questions?

The answers I've seen haven't necessarily made much sense to me.

This is a very deep rabbit hole to dive into.

I see all of these things as games that you're developing.

The main metric to me about how decentralized it is, is what are the barriers to entry and the advantages to scale or to network effects that the system involves.

And I think the definitions and proposals for decentralized block buildings to me are like trading off what is like a competitive market with individual agents for like a network that has network effects.

And to me that doesn't necessarily feel more decentralized.

Maybe there's something I'm missing, but it is susceptible to same centralization pressures at different nodes of the network as any participant in a latency system have.

And So while it may have properties and look like a network, it doesn't necessarily seem very decentralized to me.

I don't know where decentralized block building is headed.

I think there'll be a lot of experiments and attempts at developing it.

I would tend to believe that there's something promising to be said for adding more constraints to the way that block building works and essentially giving validators more control over the policies of what they accept as blocks.

And maybe you could say that that sort of decentralizes the role of block building a little bit by constraining them to some set of rules that are enforced at the validator level.

And so you are leveraging sort of more distributed validator set and the requests that those validators may have while still having the block builder role to actually enact those requests.

That might be like a direction for more trustless block building.

Is there like a specific approach to decentralized block building that you've looked at?

I mean, it's, I guess there's a few reasons why this, I have not looked at this in any depth, but there's some things that sort of like triggered this a little bit.

I mean, 1 thing is that, you know, there was like so much effort and thinking that kind of went into, oh, let's try to make Ethereum, you know, as decentralized as we can, that there's lots of different validators.

Now, did that work or not versus like other proof of stake chains?

There's some aspects in which Ethereum is maybe more decentralized, some aspects maybe it's like similar to other networks, but you know, it's still pretty centralized, right?

See all Epicenter Podcast transcripts on Youtube

Stephane Gosselin: Frontier Research - Solving Ethereum's MEV Problem. Ep. 511