Maximal extractable value is the profit available to whoever decides the contents and order of a block, beyond the fees and rewards the protocol pays. It exists because transactions are broadcast publicly before they are executed, and because the party assembling the block is free to reorder, insert or omit them. Anyone who can see a profitable transaction before it settles can act on it.
The forms are well catalogued. Arbitrage between venues, where a price difference across two pools is closed by a bot in a single transaction, is the benign case: it makes prices consistent and the profit is compensation for doing so. Liquidations are similar — someone must repay undercollateralised debt, and the discount is the incentive. The extractive case is the sandwich: a bot sees a pending swap, buys ahead of it to push the price up, lets the victim's trade execute at the worse price, and sells immediately after. The victim receives less than they would have. That difference is the cost, and it is paid by the person who never knew the bot was there.
Front-running of this kind is possible because slippage tolerance is a user-set parameter that most users leave wide. A tight tolerance causes more failed transactions and less extraction; a loose one does the opposite. The economics are therefore partly a user-interface problem, and interfaces have been slow to make the trade-off legible.
The response has been to restructure block production rather than to prohibit the behaviour. Proposer-builder separation splits the role in two: specialised builders assemble blocks and bid for the right to have them included, and the validator proposing the block simply takes the highest bid. Extraction still happens, but the value flows to the proposer as a bid rather than accruing to whoever happens to be sophisticated, and ordinary validators capture a share without running extraction infrastructure themselves. Private transaction relays go further, letting users submit transactions without broadcasting them publicly, which removes the sandwich opportunity for those who use them.
The centralisation concern is the part worth watching. Building blocks profitably is a specialised, capital-intensive business with strong returns to scale, and a small number of builders producing most blocks is a meaningful concentration of ordering power — including the power to censor. Several measurement efforts track builder and relay market share for exactly this reason. A network can be decentralised at the validator layer and quite concentrated at the layer that decides what validators sign.
The uncomfortable conclusion is that MEV cannot be removed from a system with a public mempool and discretionary ordering; it can only be redistributed, made competitive, or hidden from the parties who would exploit it. Designs that encrypt transactions until they are ordered, or that impose a deterministic ordering rule, change the problem rather than deleting it. For a user, the practical mitigations are narrow slippage tolerances, private relays where available, and awareness that a large swap into a shallow pool is an advertisement.