This RIG Open Problem (ROP) asks for numerical or formal theoretical validation of an argument to help decide between block or slot auction ePBS.
A possible upgrade to Ethereum would put proposers in a position where they would have to choose between one of two times, Early or Late, to sell their execution payload construction rights, a special, valuable right that the validator holds. In the Early auction, bidders only know their distribution of possible values. In the Late auction, bidders know their realized values. The proposer chooses the auction that maximizes its revenue. This ROP asks for simulations and formal arguments to show under what circumstances the proposer chooses which auction time.
This ROP invites contributions from those eager to contribute to core Ethereum development but who may not be involved with it, for example, students or contributors to other parts of the crypto ecosystem. The problem statement is stripped of much of the context usually required for core Ethereum problems, but results related to this ROP would be extremely valuable for core Ethereum development.
Enshrined Proposer-Builder Separation (ePBS), a potential upgrade to the Ethereum protocol, facilitates the fair exchange between a beacon proposer in charge of proposing the following block to the Ethereum blockchain and a sophisticated builder that pays the beacon proposer for the right to order transactions in this block. ePBS ensures that the beacon proposer receives its payment unconditionally based on the builder's actions (Unconditional payment) and that the builder gets the right to build the block (Honest Builder Safety). This fair exchange is vital to the decentralization of the validator set in Ethereum because it allows all validators access to the same market of sophisticated builders, ensuring that all validators have the same rewards in expectation regardless of their sophistication.
Notably, an ePBS design should satisfy the desideratum of No Trusted Advantage. This desideratum informally states a beacon proposer should be incentivized to use the in-protocol fair exchange that ePBS allows and should not be incentivized to sell its block construction rights via other methods that involve trusting a third party. If an ePBS design does not satisfy this desideratum, then it is expected that few beacon proposers will use ePBS, and thus, its development would largely be in vain.
Currently, two types of auction mechanisms are considered for ePBS. One is block auctions, which means that the auction winner pays the beacon proposer some fixed amount and immediately submits the block it wants to get on-chain. In slot auctions, the auction winner pays the beacon proposer some fixed amount, but the buyer still has some time, precisely 6 seconds, to construct the optimal block. The difference between the two may seem small, but they have vast implications for the structure of the builder market and the health of the Ethereum network.
Recently, an argument has been presented that Slot Auction ePBS does not satisfy No Trusted Advantage. The argument roughly goes as follows. Assume a beacon proposer can sell the block construction rights Early or Late. These two auctions are mutually exclusive. ePBS facilitates the Early auction, and a trusted third party facilitates the Late auction. In the Early auction, bidders know the probability density function of the value the block construction rights would have to them. In the Late auction, bidders know the exact value they assign to the block construction rights. Therefore, in the Early auction, bidders bid based on their expected values, whereas in the Late auction, bidders bid based on their realized values. Under reasonable circumstances, it can then be expected that the revenue in the Late auction will be higher than in the Early auction. This is because the probability that the highest realized value (the Late auction revenue) is higher than the highest expected value (the Early auction revenue) is large. The beacon proposer is thus incentivized to use the Late auction, meaning that the No Trusted Advantage is not satisfied.
Please read this post on Trusted Advantage in Slot Auction ePBS for a complete exposition of this argument.
Figure 1: Game tree of beacon proposer’s decision to choose the Early or the Late auction. In the Early auction, builders observe their distribution of possible values whereas in the Late auction builders observe their realized values. The beacon proposer chooses the auction that maximizes its payoff.
This argument has not been formally modeled, and no work has been done on estimating the magnitude of the problem. For example, how frequently would parties be incentivized to use the Late auction? What is the magnitude of the difference in expected revenue between the Early and the Late auction? This ROP asks for insights into these questions and welcomes submissions to related problems.
The deliverables are expected to provide more details on the following points:
A simulation could assume an ascending-bid first-price auction with private valuations, which is revenue equivalent to a sealed-bid second-price auction. Hence, it can be assumed that bidders bid their true value for the item, and the winning bidder pays the second-highest bid. The distributions of values for bidders could either be taken from standard distributions or informed by the historic bids that builders have placed in the MEV-Boost market, which is the current out-of-protocol instantiation of ePBS.