# MEV-Boost or No MEV-Boost? ## Introduction The Glamsterdam fork introduces a major change to the Ethereum protocol in the form of ePBS(enshrined Proposer Builder Separation). Builders are now enshrined within the protocol(link to builders field in beacon state). ## What does MEV-Boost do today? MEV-Boost does the following: 1. Connect to multiple relays simultanousely. 2. Per relay configuration via the config file. 3. Strategic timing games(contraversial usecase for clients) 4. Bid verfication 5. Relay multiplexing ## MEV-Boost with ePBS With ePBS, proposers directly talk to builders rather than with a trusted relay entity. Proposers ideally would want more expressivity in selecting bids from builders. They would want to pick a builder in accordance with their value for min-bid, pick a builder who pays either via the trustless or trusted mechanism and other types of use cases. Proposers With ePBS, we can imagine MEV-Boost being a sidecar which queries multiple builders and selects bids based on the proposer's per-builder preferences. These preferences might change with time and having MEV-Boost can allow us to update these preferences without making chagnes to clients. ## Arguments For Keeping MEV-Boost The main argument for keeping MEV-Boost IMO is that, its an external piece of software which doesn't require client co-ordination to make changes. We can iterate on different bid selection mechanisms and proposer preferences without client changes. Teams can come up with their own sidecar implementations like hwo commit-boost did allowing more customization in the way we select bids and add proposer preferences. Generally making changes in clients is not trivial and usually requires co-ordination. ## Arguments Against Keeping MEV-Boost 1. MEV-Boost historically had an ownership problem due to which Builder-API testing used to always be delayed and done poorly during hard forks. 2. It's main logic to do bid multiplexing and to maintain status checks over multiple relays are not too complicated to be absorbed into a client.