The state of Ethereum's supply network

Transcript

[Music] no no no no no no no no no no no [Music] no no no no no no no no no no no no no no no no no no no no [Music] thank you

hello yes okay hi everyone Thanks for thanks for coming for to this talk I'm tomasiri I'm a researcher that works at the robust incentives Group which is a group at the ethereum foundation that's focused on mechanism designed and economic incentives around the ethereum protocol and today my talk will focus on the state of the ethereum supply network okay so first for those who haven't been to other iMovie talks I'll just start with some key Concepts and definitions first one everyone has their own definition

but Mev or maximal extractable value can be just considered as the excess value that can be extracted by adjusting the execution of users transactions and by adjusting I mean reordering or adding or removing transactions and this is usually done to this reordering of transactions is usually done to extract some value that is often kind of generated by market inefficiencies so an example is an Arbitrage and this kind of power is usually given to validators um but we'll see that there are some subtleties that were introduced with PBS

and we with other actors coming in so PBS or proposal Builder separation was introduced with the merge as kind of a concept but also like an actual novel protocol feature and it's designed to preserve decentralization amongst the ethereum validators so what it does is basically ethically effectively distinguishes the role of block production that is now dedicated to block Builders and the role of a block proposal that is now handled by block proposals and block proposals are just validator that are randomly selected if

we start is every slot to propose a block to the rest of the pr2br network and so yeah this challenge kind of shifts the burden of um of uh out of the proposal's hands because now like all the kind of computationally intensive Mev extraction strategies can just be done by the builders so validators just have to basically get there and pick a bid and propose the block to the rest of the network and that's pretty um that's not the very computationally expensive um but the thing is that proposals are still the ones um with the ability to reorder and incorporate like additional transactions

after they receive blood flow for from Builders so you just if we just keep it that way it doesn't really work so there was a need to introduce a new piece um and so basically to protect Builders from Mev theft or like strategy appropriation um the ethereum and flashbot researcher introduced math boost and map boost is like a Naruto protocol software that kind of basically allows block proposers to access a Marketplace of blocks from a builder Marketplace here and that's done through trusted intermediaries or trusted third parties that we call relays um and so here it basically how it works you have the builders

there are participating in an auction we call that a med boost auction and they submit evm valid blogs and their Associated bids and the bids our value are based on the priority fees and the Mev that was extracting from the user's transactions and and so they submitted over the course of a block you can see that it's usually increasing and the relay just kind of well it used to validate blocks but now it really doesn't fall for some but basically what it does is like it makes those available to proposals and proposals at the end of the slot so around 12 seconds we'll just

like choose the highest bid and commit to it by signing it and that fact that the design it is cool because that means like they can't like build another block otherwise they will be equivalating and that's slashable so once it's signed basically the block payload is released to the rest of the network and it's propagating and so these kind of mechanism with the relay kind of ensures that proposals can't steal the content of a black Builder and um but it does rely like on trusted third parties that are relays so we'll see that some research is like kind of targeted at uh trying uh at like looking

into ways to get rid of relays in the future and in shrining that's what we mean by enshrining PBS in protocol okay but this is only part of the story um and there are like many other parties and mechanisms involved in like whether like consuming or producing resources in the blogspace market but also like generating an extracted extracting Mev and that goes basically from user to proposal so in a post I was proposed that was written by Stefan Gosling last year um he proposed like a mental model to understand the chain of event of events for the whole kind of blog space Market from users to proposals and so it's it's stuff like this

um I was almost afraid to use the word intent there but the users start with an intent like I want to swap it for usdc and then they connect to adapt like uni swap via their wallets and basically then they sign the transaction to execute the Swap and this transaction is usually made available in a global topic that's called the mempool and it's pending and it's basically waiting to be included in a blog so what Searchers do is they look at these pending transactions and they reorder them so they can extract Mev from them and that material materializes into bundles so that are like set of atomic transactions

that uh so what they do is like they make bundles they take a cut because they have to get paid and then they forward the vendors to the black builders and in principle the black Builders just has to take those bundles and pack them together in addition to all the transactions in the mempool that weren't necessarily generating Mev and so builders construct a full blocks and then they participate in the map instruction that I just mentioned before so the whole interplay between relays proposals and builders so hopefully I've conveyed how this mental model of the supply chain is helpful to understand how things work but

um since VBS was introduced things have changed quite quickly and like there was kind of an explosion of Innovations and and protocols and experiments that have popped up around blog space production and allocation and named extraction and and this in in all opinion motivates like a shift of mental model from the Mev supply chain that I just talked about to an MVD supply network and yeah when we were discussing this with other like ring members we looked up the definition of supply network on Wikipedia and just with changing a few words it kind of made sense uh so an ethereum supply network may be

defined as a pattern of temporal and spatial processes that are carried out at facility nodes and over distribution links which adds value for users through the manufacturing and delivery of blogspace okay so maybe an updated picture for the state of the ethereum supply network as it is looks like this now um this is a lot but it's not exhaustive at all there are like really tons of things happening in in that space because it's all out of protocol so everyone can just be in innovate um and they do so I thought it would just be interesting to add some research

topics and resources that are associated with the new boxes I've added to the figure so the first one is ofas or other flow auctions so here it's actually cool it's like basically Wi-Fi's allow users to directly send their transactions to Dedicated main pools via like a custom RPC um and in that dedicated mempool you have solvers and virtuals that compete for the right to extract immediate opportunities but you can Implement new features like programmable privacy so for example users can choose to share or not some elements of their transactions and you can also like another cool

feature is like choosing how you want to redistribute Mev so there was this big big thing about Mev rebates we in which some some of the value that was generated so the Mev is actually rebated back to the user for for whose transaction was actually generating Mev uh another interesting thing to look at in the future I haven't seen too much research on this but it's a account abstraction and well no account abstraction is there is some research but like the role it plays in Mev a bit less um so account abstraction is um and it's implementation via like this

new standard of erc4237 um is worth looking at in terms of what a role will play in in this Mev Supply deck Mark and I'm kind of interested on how mempool fragmentation so the way basically you'll have like a lot of alternative mempool that will pop up how this will have a role and maybe we can look into this then I added a swap box um so it's not live yet but there were some recent announcements by flash votes and I expect there will be a bit more maybe this week um and like the idea is to propose like an end-to-end uh so end-to-end is user to proposal

chain that's like dedicated on a 2mev and that's trying to destroy decentralize the whole block building process um yeah so that's the idea uh I've also added eigen layer so this uh so again a layer I guess is live on mainnet but like the applications are not uh but basically it enables validators to restrict their if and perform extra duties or offer extract services and we can expect some of these services will be directly targeted to Mev extraction or block space prediction and this is an exchange for additional protocol fees and Rewards and they have a proposal I think

specifically on that on like partial blog math boost building through eigen layer if you want to check it out and then you have all the Mev aware adapts you have the existing course web but also like it might be interesting to see in the future what people can do with uni swipe hooks regarding mvv um and so yeah my last part is maybe what closer to what we've been focusing on at the rig um and uh yeah at the rig we've focused mostly mostly on this part I guess it's also the part that's closer to proposal so closer to to the protocol uh and I thought it would be a good opportunity to mention a few examples of well we are

working on so the first example is research led by Caspar here on timing games we got an article that got accepted at SBC so I've linked it here all the blue underlined caption now are actually linked so if you want to check it out after the presentation that's a it's a way to do that uh and yeah the this line of research focuses on basically players proposals that play a timing game which means before um you know I said around 12 seconds proposers can just choose a block and propose it to the rest of their Network and they can play the game of just waiting a bit longer so they can accumulate more Mev and earn more

rewards because muv is increasing linearly with time um and so it's interesting in itself but also it introduces a a way faster like a larger or high level questions of um whether consensus protocols and designs in addition to the usual um liveness and and finality in concerns also need to take into account economic fairness so that's again research that that I think is super interesting another very important line of research is epws design so I've mentioned there is ways of trying to enshrine PBS and Maple

suctions and getting rid of relays to have this uh in the ethereum protocol protocol and this is research Barnaby with Pepsi but also Mike Justin and Francesco have been very active on and you have a very kind of implementation security Focus roadmap um I'm trying to speed that up because I'm very late uh but uh and then yeah something I've been very interested in is on the more I guess modeling simulations part of things to try to simulate map boost options are there now and to try to tweak the parameters and to look at Builders strategies and what

what they are using as building strategies so we can try to design auctions or suited for what we need which means avoid centralization minimize the risk of people using side channels because our mechanism is actually not good enough and everything so that's uh that's a kind of simulations empirical modeling approach and there is another formal more formal approach that Julian has been leading but also with adavity and Barnaby that is looking as more like theoretical and formal models of PBS um and I think there are like some great insights to to learn from that too and

and I I don't have time to get too into it but a Pepsi is a way more General uh kind of uh way of uh I'm getting signs of like hardcore signs okay check it check out Pepsi it's very cool the takeaways I'm going to be very quick here but basically I think thinking of the Ethereal like this ethereum supply network is is important because it helps like understanding all the interconnected parties that are interacting together um it can probably also be an actual applied path to analyzing the network as a network like as a complex system maybe using like metrics that are suited for

it so of web theoretical analysis and so on and also I just wanted to like Shield research on the Mev supply network because it's super broad it spans like very various research domains and so we have rig open problems which are specifically dedicated to research chairs that want to contribute and collaborate with us so feel free to to check it out and come to us afterwards and the review home page where you can find all our posts papers and and talks thank you [Applause]