Problem Description
MEV-Boost and associated relay systems in the Ethereum block production supply chain require the ability of actors to outsource the work of block construction. Ethereum validators that are selected as proposers outsource this work to relays who in turn interact with 3rd parties (”builders”) to complete this role. One challenge in doing this outsourcing is that the counterparty to whom work is outsourced may (rationally or maliciously) fail to behave as the outsourcing party expects - e.g. by failing to deliver a block. At the same time, parties doing the block building work are at risk of their work being (mis)used without compensation - e.g. by “unbundling” the candidate block and including transactions in a different block, and discarding the candidate.
Currently, the ecosystem has addressed this fair exchange problem by having a small number of trusted relays. Validators outsource block production to these relays, trusting that they will not misbehave, while relays only provide candidate blocks once the validator has committed to due compensation. When relays promise to deliver blocks but do not deliver, they will compensate validators for this, but there is not mechanism to enforce this. Block builders interact with relays in one of two ways:
- “optimistic relaying” in which the builder deposits collateral at the relay in order to interact with them, trusting that the relay will return this collateral if requested.
- “validated/standard relaying” in which the builder provides their work without commitment to payment, trusting that the relay will not abuse this information.
As a consequence of this market, barriers to participation for validators and builders are very low and not tied to identity (“permissionless”). However, participation of relays is based on trust and implicit reputation. Consequently, there are very few relays as both sides of the market must vet new relays who participate. It is an open question how this system can be adjusted such that participation in this block outsourcing market can be made completely programmatic, such that no human evaluation of reputation is required.
One class of solutions to this problem aims to prevent misbehaviour altogether with the use of cryptography schemes that involve many parties. Another is to use a fully bonded approach in which block production is only outsourced to parties who provide collateral, leaving the technical problem of ascertaining in some credible manner whether the outsourcing or providing party is at fault. Both approaches invite technical complexity and overhead (latency, infrastructure cost) and the latter also introduces capital cost. A third option which has not yet been explored in much depth is the use of programmatic reputation systems, perhaps in conjunction with a collateral-based approach. What are potential reputation schemes? Can such an approach improve capital efficiency and keep barriers to entry sufficiently low? Would such an approach lead to more or fewer failed exchanges? Can a reputation system be built without relying on a 3rd party mechanism which attributes fault?
This is an early description of the problem and is written without a thorough analysis of prior literature on reputation systems.
Related Notes
‣
‣
‣
‣
