Every swap you sign is an instruction to reorder the market, and someone is paid to do that reordering. That is MEV, and the useful way to think about it is not as theft but as a queue with a price: value that exists because blocks are ordered rather than simultaneous.
The interesting question is who captures it. The answer changed in 2022 and the change is why the topic stopped being an obscure one.
The three ways value shows up
Arbitrage between venues is the largest and least controversial source. Two pools price the same asset differently for a few blocks; someone closes the gap and keeps the difference. It is roughly a public service with a tip.
Liquidations and backruns are the second: the right to be the first transaction after a large trade or a price move. Sandwiching a retail swap — buy in front, sell behind — is the third, and the only one that takes value directly out of a user's pocket.
Most of the value is in the first two. Most of the outrage is about the third, which is exactly the ordering the public mempool invites.
Who does what
Searchers scan the mempool for opportunities and encode them as bundles: an ordered set of transactions that must succeed or revert together. They are the ones who actually find the trade.
Builders assemble blocks by selecting and ordering bundles, competing to produce the most valuable block possible. A handful of builders produce most blocks, which is a concentration problem worth watching on its own.
Relays pass the winning block from builder to proposer without revealing its contents, and proposers — validators — pick the most valuable header. MEV-Boost is the middleware that standardised this, and it means the validator proposing a block is no longer the party deciding what goes in it.
Proposer-builder separation is what that arrangement is called, and it turned MEV from a validator engineering problem into a market with an auction.
What separation fixed and what it broke
Before PBS, every validator needed custom infrastructure to capture MEV, so it concentrated among professionals and pushed smaller operators toward pools. After PBS, an ordinary validator with MEV-Boost captures most of the value with none of the search infrastructure, and the rewards have been large enough that most Ethereum blocks are now built this way.
The cost is a new dependency. Validators now trust relays, and relays are a small set of mostly permissioned intermediaries that could, in principle, censor transactions. The mitigation is having more than one relay configured and watching builder diversity — a subtly centralised block-building market is a worse outcome than MEV itself.
What a user can actually do
Assume your transaction is visible and priced accordingly. Submit through a private RPC so it is not sitting in the public mempool waiting to be sandwiched. Set slippage limits you are willing to accept rather than defaults. On some venues, use a batch auction or an intent-based router that matches your order in competition rather than exposing it.
None of this eliminates MEV. It redirects it from the extraction you notice to the arbitrage you do not.
Where this is going
Cross-chain MEV is the obvious next surface as more activity moves through bridges and intents. Encrypted mempools, if they work at scale, would remove the opportunity for front-running by hiding order flow until it is final. And decentralised block building aims at the concentration problem directly.
Judge each proposal by who ends up with the value and who has to trust whom. MEV cannot be legislated away, because ordering is a fact of the system; it can only be redistributed, and every mechanism so far does so imperfectly.
Next: how sequencing works on the networks where the ordering happens off-chain. Join our Discord.
References
Next, read L2 Scaling Solutions 101. More in Infrastructure & Scaling.