The asymmetry is obvious once you notice it.

Protocol fees are a deliberate mechanism set by governance, routed to LPs or treasuries, distributed according to protocol logic. MEV is also a product of protocol activity. But MEV doesn't route to protocols. It routes to validators, block builders, and the infrastructure layer that intermediates between transaction creation and block inclusion.

Both are generated by the same user activity. One is internalized. One isn't. Understanding why requires looking at the architecture that makes each possible, and what that implies for protocols that want to close the gap.

Fees Are Inside the Transaction. MEV Is Outside It.

Protocol fees are collected inside the execution boundary of the swap. When a user trades, the fee is computed, deducted, and routed atomically, as part of the same operation. The protocol doesn't need to coordinate with anyone external. The fee logic is baked into the contract. There's no race to win, no external party to share with, no infrastructure dependency.

MEV is different. The arbitrage opportunity that exists after a swap, the price dislocation between the pool and external markets is realized by a separate transaction. A searcher observes the post-swap state, constructs a backrun, and submits it to the mempool. That transaction then competes for block inclusion. Validators (or block builders under PBS) order transactions to maximize their own revenue. The surplus flows to whoever wins the blockspace market.

The protocol created the opportunity. The protocol saw none of the proceeds. The reason isn't governance or tokenomics, it's execution architecture. Fees are captured inside the transaction boundary. MEV escapes it.

Why the Blockspace Market Extracts MEV from Protocols

Once an arbitrage opportunity hits the public mempool, it enters a competitive market with its own extraction logic.

Searchers compete through priority fees, builder relationships, and latency optimization. Block builders maximize revenue by selecting and ordering transactions optimally. Validators select the most profitable block. At each stage, the actors with the strongest position in the blockspace market extract value, and they have no obligation to share with the protocol that generated the opportunity.

This isn't adversarial in a simple sense. Searchers, builders, and validators are doing exactly what the system incentivizes them to do. The protocol architecture just wasn't designed with MEV internalization in mind. MEV externalization is the default, not the outcome of any deliberate decision.

The result: protocols charge 5-30bps in explicit fees while generating arbitrage opportunities worth multiples of that in implicit value, most of which exits the system entirely.

The Structural Asymmetry

Protocol fee revenue scales with fee rate and volume. MEV leakage also scales with volume but it scales with price impact and volatility too, which means high-value swaps generate disproportionately large arbitrage opportunities relative to the fees they generate.

For a large, volatile swap, the explicit fee might be 30bps. The post-trade price dislocation could be 50-100bps of the trade size. That differential goes to the blockspace market.

This is why LVR is such a useful lens: it measures exactly this transfer. LPs provide liquidity, generate swap volume, charge fees, and bleed LVR to external arbitrageurs who correct the prices their AMM couldn't update fast enough. The fee revenue and the LVR leakage are both functions of the same user activity. They just have different destinations because they have different execution architectures.

What It Would Take to Internalize MEV

The asymmetry suggests a clear solution: move MEV execution inside the transaction boundary.

If backrun arbitrage happens in the same atomic operation as the swap triggered by a hook, bundled by a multicall, or embedded in an EIP-7702 session then it operates inside the execution boundary where fee logic already lives. There's no external transaction. No mempool exposure. No blockspace race. The surplus doesn't escape to an external market because the transaction that captures it is the same transaction that created it.

This doesn't require speed advantages, builder relationships, or latency optimization. It requires execution topology, specifically, the architectural property that execution happens inside the transaction that triggered the opportunity rather than after it.

The protocols that have built this have a categorically different MEV outcome than those that haven't. Not marginally different categorically. One has MEV as a revenue line. The other has MEV as a cost.

Why This Default Persists

If same-transaction MEV capture is technically feasible and economically superior for protocols, why isn't it the default?

Several reasons. First, the tooling is recent. Hook architectures that enable afterSwap execution are a feature of Uniswap v4 and similar systems, they didn't exist at scale until recently. Second, the design assumption in most AMM architectures is that MEV is an externality to be minimized rather than a resource to be captured. Third, integrating execution mechanisms requires protocol-level changes, which have governance and engineering overhead.

But the more fundamental reason is that MEV externalization was never a deliberate choice. Protocols didn't decide to give MEV to validators. They built fee mechanisms and left everything else to market structure. Market structure optimized for validator revenue, as it was designed to do.

Changing the outcome requires a deliberate architectural decision: explicitly designing for MEV internalization rather than accepting the default.

The Question Worth Asking

Protocol fees go to protocols because fee logic lives inside the transaction boundary. That was a design choice, someone wrote the contract to collect and route fees.

MEV goes to validators because MEV execution lives outside the transaction boundary. That's also a design choice or more precisely, the absence of one.

The asymmetry isn't structural in any deep sense. It's architectural. And architecture is something you can change.

The protocols that recognize they own valuable order flow and build the execution infrastructure to exercise that ownership will have a fundamentally different cost structure than those that don't. MEV isn't a tax on protocol activity. It's a revenue stream that's currently being collected by someone else.