Profiting From Deciding Which Transaction Goes First
On a public blockchain, pending transactions are visible before they execute and whoever orders them into a block can capture value from that ordering. The extraction is substantial, measurable, and largely unavoidable.
The Property That Creates It
On a public blockchain, a user broadcasts a transaction and it waits in a public queue before being included in a block.
Anybody can see it. And whoever produces the next block chooses which transactions to include and in what order.
That combination, public pending transactions plus discretionary ordering, creates a value opportunity that would not exist if either half were absent.
The value extractable through including, excluding, and ordering transactions is called maximal extractable value.
The Forms It Takes
| Type | Mechanism | Harms the User |
|---|---|---|
| Arbitrage | Correcting price differences between venues | No, generally beneficial |
| Liquidation | Closing undercollateralised positions | No, required by the protocol |
| Front running | Placing an order ahead of a seen transaction | Yes |
| Sandwich attack | Buying before and selling after a user trade | Yes, directly |
The distinction matters and is frequently collapsed. Arbitrage between venues improves price consistency and would be regarded as useful in any market. Liquidating an undercollateralised loan is a function the protocol requires somebody to perform.
A sandwich attack is different in kind. Seeing a user pending purchase, buying the same asset first to move the price up, allowing the user trade to execute at the worse price, and immediately selling, extracts value directly from that user with no offsetting function.
Front running a visible pending order is what securities regulation prohibits and what this design makes structurally available to whoever orders the block. The difference is not the ethics, it is that here the ordering right is sold openly.
How the Market Organised Itself
Early extraction produced a destructive competition. Bots competed to have their transactions ordered first by bidding up transaction fees, which congested the network and raised costs for everybody.
The response was infrastructure separating the roles.
Searchers identify opportunities and construct transaction bundles.
Builders assemble complete blocks from bundles, optimising for total value.
Proposers, meaning validators, select the most valuable block offered without seeing its contents.
That separation, generally called proposer builder separation, moved the competition into an auction conducted off the public network, which removed the congestion problem.
It also concentrated the activity. A small number of builders produce a large share of blocks, which is a centralisation outcome in a system designed around distribution.
Where the Value Ends Up
The extracted value flows substantially to validators, because builders bid most of it away to win block inclusion.
That has a consequence for the economics of the network: staking returns include a component derived from extraction, which means the yield on validating depends partly on activity that harms users.
Attempts to quantify the total have produced figures in the hundreds of millions to billions of dollars annually depending on the period and methodology, with the harmful categories a meaningful minority of it.
The Mitigations
Several approaches reduce the harmful forms rather than eliminating extraction.
Private transaction relays submit transactions directly to builders rather than to the public queue, which prevents them being seen and sandwiched. This is now widely used and it requires trusting the relay.
Batch auctions execute all trades in a period at a single clearing price, which removes the value of being first entirely. That is the same mechanism equity markets use in opening and closing auctions, applied for the same reason.
Encrypted mempools conceal transaction contents until after ordering is fixed, which removes the information the extraction depends on. This is technically demanding and is where the most serious protocol level work is directed.
Slippage limits set by the user cap how much price movement they will accept, which bounds the loss from a sandwich without preventing it.
The Comparison Worth Making
Traditional markets have the same underlying problem and address it differently.
Order flow information is valuable, front running client orders is prohibited, and the prohibition is enforced by regulators with the power to fine and to bar individuals.
A permissionless system has no such authority, so the response must be technical rather than legal. That is why the work is on encryption and auction design rather than on rules, and it is a genuinely interesting case of the same market structure problem being solved by different instruments.
The Bottom Line
Maximal extractable value arises because pending transactions are public and somebody chooses the order, which makes front running available by design rather than by misconduct. A substantial share of it is ordinary arbitrage and liquidation that any market needs, and the sandwich attack is direct extraction from a user with no offsetting benefit. Because there is no regulator to prohibit it, the mitigations are architectural, principally hiding the transaction until ordering is fixed or removing the value of being first through batch auctions.