By @Stanley He
The emergency of based sequencing and preconfirmation (preconf) on Ethereum has seemingly been pushing the protocol further down the path that makes it less and less intuitive. In this article we want to show why such a process is necessary, and how everything is actually driven by the “frontend” - even the protocol itself and the validators.
Let’s start with MEV.
Historically, proposers (miners) are the biggest value capturer in blockchain economy. While searchers and builders fist fight each other in the dark forest, proposers are still passively getting a bigger share in MEV, although they are the last ones who see the transactions.
Those who first see transactions, namely wallets and dapps, have always tried to capture more value. In the name of the users usually. I wrote about it in 2022 and 2023. The idea is simple: whoever sees the transactions first (the “frontends”) get to alter them the most, thus in theory capture most of the value before any other players downstream. For example, a DEX can sell user transactions to searchers who pay the DEX, thus extracting MEV before transactions are even reachable by mempool searchers/builders/proposers. Atlas is building a customizable OFA (orderflow auction) solution to allow every dapp to do it.
To put it short, the state transition within a rollup is almost fully independent from the L1 state, leaving little room for L1 proposers to exert their influence. But a dapp — even a self-sequenced dapp — is still part of the L1 state, meaning it’s still within the purview of L1 proposers. All state transition, no matter how and who sequenced, must get the proposer’s sign-off. It is the proposers who ultimately control the inclusion of transactions.
Selling orderflow, however, isn’t enough. The problem is, only non-toxic orderflow is worth selling, but most of MEV is generated and extracted by toxic flow. From DEX’s perspective, OFA is just milking the naive users who are already exploited by everyone on the MEV pipeline. The biggest value drainer, however, is the CEX-DEX arbitrageurs who strive to become the first trader against the DEX in a block and profit from the price difference between the DEX and Binance. In order to capture value that previously goes to toxic flow, a DEX must sell a different thing: Instead of selling “the right to execute transaction A” like in OFA, it now has to sell “the right to TRANSACT first with the protocol”. And that means controlling transaction ordering, or sequencing.
It’s not a new idea for dapps to sequence (at least part of) their own transactions. MEV capturing AMM (McAMM) was proposed in 2022, where the AMM auctions off the right of the first trade per block, and all other transactions cannot go through before the winner executes his trade. In this way, the biggest pie of MEV generated by the AMM, which is the top-of-block CEX-DEX arbitrage, can be largely captured by the AMM itself.
Sorella Labs carries on the mission and is building an off-chain infrastructure for its own sequencing. Sorella not only auctions off the right of the first trade, but also gives the a uniform execution price for all the rest transactions through a batch auction. In this way, the Sorella off-chain consensus acts as the actual sequencer for the DEX, turning it into a self-sequenced dapp.
*The dynamics between Sorella-type of AMM and transaction ordering*
MEV stems from the transaction ordering power, which ultimately belongs to block proposers. Under PBS, block builders bid for this power, transmitting MEV still to the proposers, but also retaining part for themselves.
Self-sequencing further moves MEV extraction towards the frontend. Instead of users submitting individual transactions that interact with the dapp smart contract on-chain, now the dapp will submit a whole bundle of user transactions on-chain. With account abstraction, this bundle becomes atomic, meaning nobody, including the proposer, can unbundle it and steal MEV.
This makes self-sequenced dapps very much like partial block builders. In today’s PBS, block builders build the whole block without proposers knowing the content. The latter only signs the block header. While for dapps, obviously they can only have authority over their own transactions — which is part of a block — and it’s possible for them to extract MEV by sequencing (building) those transactions.
What this creates is a power shift in the MEV supply chain. Conceptually, the first and last look in a value transmission process can capture the most value. Proposers have long taken advantage of their last look power to extract MEV, while the first lookers, namely dapps and wallets, only recently started to do so. With self-sequencing, the first lookers now have a perfect tool to eat proposers’ lunch. Not only proposers, traditional searchers and builders will also have less room for profit when they are faced with not transactions, but big bundles by dapps.
