MEV-Boost Architecture: PBS, Relays, and Protection

Frequently Asked Questions

01.What is proposer-builder separation in MEV-Boost?

Proposer-builder separation splits Ethereum block production into two roles. Builders assemble the most valuable block from mempool transactions and MEV bundles, competing on total block value. The proposer, an ordinary validator, selects and signs the highest-bidding block header offered through the relay. MEV-Boost implements this split off-protocol, meaning the consensus layer itself does not enforce it. A trusted relay mediates between builders and proposers, holding the block payload in escrow until the proposer commits by signing the header. This design lets thousands of small validators capture MEV revenue without running sophisticated block-building infrastructure.

02.What does a MEV-Boost relay do and why is it trusted?

A MEV-Boost relay is a trusted mediator between block builders and validators. It receives full blocks from builders, simulates them to verify bid value and validity, then forwards only the block header and bid to the proposer while holding the full payload in escrow. After the proposer signs the header, committing to that block, the relay releases the full payload. This commit-before-reveal sequence stops validators from copying builder transactions. The relay is trusted because the consensus protocol does not yet enforce this split natively. Operators mitigate relay trust by connecting to multiple relays simultaneously so no single relay policy or outage is decisive.

03.What is optimistic relaying and what trade-offs does it introduce?

Optimistic relaying removes synchronous block simulation from the critical auction path. Instead of simulating each block before admitting its bid, the relay marks a collateralized builder's bid active immediately and simulates asynchronously. This removed roughly 150 milliseconds of simulation latency in early deployments, improving builder competitiveness and proposer fees. Builders post collateral that covers proposer losses if an invalid block is delivered. The trade-off is that validity is no longer guaranteed upfront. A builder can submit an invalid block that passes the optimistic check, and the asynchronous simulation catches it only after the proposer may have committed. Collateral compensates the proposer for a missed slot but does not prevent the miss.