Innovations in Decentralized Exchange — CBER Forum Symposium
UniswapX is an intent-based trading protocol where searchers compete to execute swapper orders on-chain. On chains that use priority gas auction (PGA) ordering, that competition drives surplus value to block proposers — an actor who adds no value to the trade. MEV taxes, based on the theoretical framework from Dan Robinson and Dave White's "Priority Is All You Need" paper, let the smart contract read the priority fee as a signal for MEV and enforce that the swapper's output increases as a function of it, redirecting the majority of competitive MEV back to the user who created it. This talk, presented at the Crypto and Blockchain Economics Research (CBER) Forum Symposium at Columbia University, covered how we turned that framework into a production system at Uniswap Labs and what we've learned from operating it.
I also discussed practical challenges — auction coordination timing, participation costs from reverted transactions on fast block time chains, and the tradeoffs around scaling factor selection — as well as extensions including layering MEV taxes on top of Dutch auctions and applying the mechanism to AMMs to mitigate loss-versus-rebalancing. The symposium also featured Max Resnick (Anza Labs) presenting on Prop AMMs and emerging market structure on Solana.
When a user swaps tokens through a decentralized exchange, the trade often creates surplus value — the gap between what a competitive filler can realize from executing the order and the minimum output the user requires. On chains that use priority gas auction (PGA) ordering, fillers compete for the right to fill orders by bidding up their transaction's priority fee. Under competitive conditions, fillers bid away their entire surplus, and that value accrues to the block proposer or sequencer as priority fee revenue.
The block proposer is not a counterparty to the trade. They didn't create the value, and they don't contribute to execution quality. From the perspective of the swapper and filler — the two parties who actually participate in the trade — this value leakage is deadweight loss in the system. The question is whether we can redirect it.
MEV taxes, a mechanism introduced by Dan Robinson and Dave White in their 2024 paper "Priority Is All You Need", provide an answer. The core insight is that on a chain where transactions are honestly ordered by priority fee, the priority fee itself becomes a universal signal for how much MEV exists in a transaction. A smart contract can read that signal during execution and charge its own fee as an increasing function of it.
This only works under a specific assumption: the block proposer must faithfully order transactions by descending priority fee without censorship or manipulation. On Ethereum L2s with centralized sequencers — such as Base, Unichain, and OP Mainnet — this assumption holds because the sequencer commits to priority ordering by default through the OP Stack. On Ethereum L1, where block construction is outsourced through a competitive builder auction that maximizes proposer revenue, MEV taxes would not function as designed. The mechanism is therefore most naturally suited to environments where the sequencing rules can be trusted or enforced.
UniswapX is an intent-based trading protocol. Rather than submitting a transaction with a fixed execution path, a user signs an off-chain message — an intent — specifying what they want to trade and a minimum acceptable output. Fillers compete to fulfill that intent by submitting on-chain transactions.
The MEV tax is embedded directly in the intent structure. The smart contract enforces that the swapper's output is a multiplicative function of the transaction's priority fee: the output equals the minimum output scaled up by a factor that increases linearly with the priority fee per gas. For each additional wei of priority fee a filler specifies, the contract requires them to deliver an additional fraction of output — denominated in units called mps, or thousandths of a basis point — to the swapper.
The result is that fillers can still compete by bidding priority fees, but the higher they bid, the more output flows to the swapper rather than the proposer. The filler's total cost is unchanged — what changes is the split between swapper and proposer. The scaling factor, which controls that split, is a tunable parameter. In practice on L2s, where gas costs are low and block capacity is large, it can be set high enough to redirect nearly all competitive MEV to swappers while still maintaining fine-grained filler competition.
The scaling factor itself has been a relatively forgiving parameter to set. The more challenging design problems have been around auction coordination.
The MEV tax mechanism requires meaningful filler competition within a single block. UniswapX coordinates this by setting a target settlement block approximately four seconds after the swapper submits an intent. Fillers submit competing transactions into that block, and priority ordering determines the winner. The four-second window balances competition against latency — shorter windows reduce participation and risk turning the auction into a latency game, while longer windows degrade user experience.
On chains with very fast block times, such as Base with flashblocks, fillers struggle to reliably target the correct sub-block and resort to submitting duplicate transactions across multiple intervals. Most of these revert, turning the auction into an effectively all-pay mechanism where losing bids carry real costs. On chains where the base fee is non-negligible, these revert costs accumulate and can discourage filler participation — undermining the competition the mechanism depends on.
These issues are addressed on Unichain by introducing transaction scheduling, which allows fillers to target a specific block, and revert protection, which excludes reverting transactions from the block entirely so that fillers pay nothing for losing bids. Together, these features reduce participation costs and move the auction closer to a clean first-price sealed-bid structure. The formal implications of these design choices — how participation costs affect filler entry, swapper welfare, and equilibrium auction dynamics — remain open questions of interest to the mechanism design research community.