In the context of decentralized blockchains, accurately simulating the outcome of order flow auctions (OFAs) off-chain is challenging due to adversarial sequencing, encrypted bids, and frequent state changes. Existing approaches, such as deterministic sorting via consensus layer modifications (e.g., MEV taxes) (Robinson and White 2024) and BRAID (Resnick 2024) or atomic execution of aggregated bids (e.g., Atlas) (Watts et al. 2024), remain vulnerable in permissionless settings where limited throughput allows rational adversaries to submit "spoof" bids that block their competitors' access to execution. We propose a new failure cost penalty that applies only when a solution is executed but does not pay its bid or fulfill the order. Combined with an on-chain escrow system, this mechanism empowers applications to asynchronously issue their users a guaranteed minimum outcome before the execution results are finalized. It implies a direct link between blockchain throughput, censorship resistance, and the capital efficiency of auction participants (e.g., solvers), which intuitively extends to execution quality. At equilibrium, bids fully reflect the potential for price improvement between bid submission and execution, but only partially reflect the potential for price declines. This asymmetry unbounded upside for winning bids, limited downside for failed bids, and no loss for losing bids - ultimately benefits users.
Watts et al. (2025) studied this question.