New arXiv paper proposes Shapley-capped auction to decentralize MEV searchers
Sarenche & Liu's model targets searcher centralization with fairness metrics and anti-Sybil rules.
Maximal extractable value (MEV) markets have long worried blockchain researchers due to centralization among builders and validators, but a new preprint takes a different angle: the searchers who compete to capture profitable opportunities. In "Towards Decentralized Searcher Competition in MEV Markets," authors Roozbeh Sarenche and Yunwen Liu develop a heterogeneous model where searchers differ in opportunity coverage and execution efficiency. They introduce two metrics to quantify decentralization: a Shapley-weighted Jain fairness index, which checks whether rewards track each searcher's marginal contribution, and an expected-reward Herfindahl-Hirschman Index (HHI) to measure long-run reward concentration across searchers. Using these tools, they show that the standard first-price, winner-take-all auction systematically rewards rank-dominant searchers rather than those with higher marginal value, concentrating rewards and undermining fairness.
To fix this, the authors propose an entry-filtered Shapley-capped auction mechanism that distributes rewards more equitably among admitted high-quality submissions. But implementing this in permissionless blockchains is tricky: malicious actors could create Sybil identities by submitting copied or degraded versions of the same strategy, while validators might collude with searchers to inflate joint profits. The paper addresses these risks with Bayesian security constraints designed to deter copied-code Sybil deviations and validator-searcher coalition deviations. The result is a formal framework that goes beyond simple auction tweaks, offering both quantitative metrics to grade decentralization and a concrete mechanism design that resists game-theoretic attacks. For blockchain infrastructure teams, this work provides a rigorous way to benchmark MEV auction designs and could inform future protocol upgrades that want to preserve openness while keeping MEV extraction competitive.
- Introduces two metrics: Shapley-weighted Jain fairness index and expected-reward HHI (Herfindahl-Hirschman Index) to measure searcher reward concentration
- Shows first-price winner-take-all auctions favor rank dominance over marginal contribution, concentrating rewards among top searchers
- Proposes an entry-filtered Shapley-capped auction with Bayesian security constraints to resist Sybil attacks and validator-searcher collusion
Why It Matters
Gives blockchain protocols measurable fairness criteria and a concrete auction mechanism to prevent MEV centralization from undermining decentralization.