Sybil-Proof Mechanisms by Christoph Schlegel | Devcon SEA

[Music] all right um here we go I wanted to pitch my favorite theorem of 2024 in mechanism design very subjective that join work with our brilliant intern and Flash Bots mingau and akaki over there and I want to also tell you a bit about how it relates to ethereum road map stuff like PBS and so on so here's a question have you ever wondered can we have a decentralized builder Market um under any any mechanism that we can use to to to select a proposer or Builder and how can we make this question more precise so how can we think about it from first principles and we wanted to turn it into a mechanism design questions so we wondering okay could we have an allocation mechanism to select a proposer that proposes blocks and we operate in a peral system so we worry a lot about cbls right and we want to be distributed or decentralized we don't always want to select the same proposal okay that's minimal requirement for decentralization question is is there any mechanism we're doing mechanism design so we asking not any specific mechanism can there be any mechanism that do that okay um Bitcoin has solved that problem in some sense right so you do proof proof of work is a mechanism problem with that one is if you have private value and in ethereum certainly there's private value of proposing blocks right we have even a secondary markets for pro proposing blocks called me boost if there's a secondary market for proposal rights it means people have different value otherwise there wouldn't be a secondary Market okay so and that value might be private information to people right because there are things like uh payments for order flow or there's private order flow that people use to produce blocks or they do searching the builders and so on right so they Ample Ample reasons for for private information private value mechanism design in in the end has one Insight if you want to learn anything about mechanism design that's that that thing it all depends on private information we want to elicit private information from uh participants in the mechanism right and so we turned ethereum roadmap discussion into a mechanism design problem okay and then we can ask questions about it in a formal way can there be any incentive compatible mechanism that selects proposer classic work from the 80s we know exactly what are incentive compatible mechanisms for the setup uh so-called monotonic mechanisms me has told us about them here are some examples we run an auction we run a lottery We Run The Lottery proportional to your value and so on Etc PP it's huge space of possible mechanisms that you can do but here's a here's a Twist if we add this requirement of permissionless of the C proofness then basically only one mechanism survives right and that's a that's a punch line of the paper there only one mechanism it's an auction so we have to allocate through an auction and it's super centralizing right that's what we know about auctions so in other words can we have a decentralized builder Market basically any Market structure with significant private value will lead to Builder centralization right so think about the illion proposals that are around there of restructuring the Builder Market or restructuring PBS or um doing a Testa proposal separation theorem and principle applies to that if there's significant private value probably we run in a pretty centralized Market okay let me add some Reflections on on this private value assumption and then let's open it for discussion or questions so why is a private value in Block proposal and how does it link to Market structure so that is that that is a question you might ask um so private value comes sort of from from private order flow and as I said Searcher Searcher Builder integration in ethereum so how would we deal uh so how would we remove it from the market structure if we cannot have a decentralized market structure with private value maybe we want to allocate rights in in advance so that's one one class of proposals so this execution tickets or uh execution auction or even p POS you can think of such a system or we could put constraints on the proposal so things like inclusion lists or like multiple concurrent proposal they trying to go that route okay first proposal um allocating rights in advance I think there is sort of many people have made this point that probably one downside of that is that we naturally have a secondary market so we replicate the market structure in the secondary Market which is super centralized that's basically the status quo so I think there's no way around that that's my opinion there to take and constraining the proposal I think some of the proposal like inclusion list I mean they make sort of the bad effects of centralization uh less bad but certainly they don't remove centralization and then we can think of other other proposals like mCP where in principle also our theorem applies so there we would also expect Market concentration unless maybe there's a very smart way of training the Builder and these proposals I'm running out of time opinion to conclusion ah centralization is the thing we have to deal with it and maybe one route out out of it that's the flashboards proposal I guess is modernizing decentralizing the Builder Builder role the Builder stack always think very very uh long and carefully about other other designs okay paper here scan it yeah sorry it's it's gone no no it's on archive okay well thank you very much so we have we have time for a few questions raise your okay not thanks for the talk um can you repeat the name of the paper it's called on C proof mechanisms thank you so much thank you for Shilling the paper any other question please raise your hand if you have any question over there what do you think of lock Building inside a te did we hire you to Shield these uh no only recently this wasn't planned I I'm very bullish on on block Building inside of a t obviously because that's one way of so here's here's the thing right so if you have this Market structure that will naturally to to build a centralization then we want to maybe constrain the Builder the powers of the Builder and one way of doing it is using cry cryptographic tools or programmable privacy tools and yeah definitely te is one of those Solutions one over there so uh if uh every mechanism eventually ends up with a centralized block Builder why when we look at other chains like salana do we not see just one big block Builder we see a few right is the answer so we see a few so that's that's that's I mean do you call uh Solana block building Market decentralized I uh I'm one asking the questions here so so is the idea that is the No but but but I'm I'm I'm questioning sort of your you you say theory is rejected I'm saying is it really rejected well I think I think uh no one would uh agree that sana's blocks are being built by a single entity I also didn't make the claim I just said we we have a centralized uh centralized Market outcome and when does a market become centralized what is a when does it qualify centralized um few few um few Builders building all of the blocks thank you and the punchline of the paper you know there's no clever mechanism that that can solve that you have to solve it so other ways okay well thank you very much I give a nice uh Round of Applause to Kristoff [Applause]