[ethresear.ch] [Execution Layer Research] Proprietary AMMs and Ethereum

Really interesting analysis, especially the point about PBS potentially making stale-quote extraction an equilibrium outcome rather than just an occasional failure of transaction ordering.

One thing I’m curious about is whether the builder/PropAMM interaction could be modeled explicitly as a mechanism-design problem.

Suppose a market maker submits a parameter update at time t, but the builder controls whether that update executes before or after incoming trades.

The builder effectively has an option on the AMM’s stale liquidity, with the value of that option depending on external price movements, quote age, available liquidity, and the market maker’s pricing curve.

It seems like the builder’s incentive to censor or delay updates would increase nonlinearly with volatility and quote staleness, rather than simply with transaction fees.

Have you considered whether there exists a fee or commitment mechanism that makes honoring the update economically preferable to extracting the stale-quote arbitrage?

My intuition is that without an enforceable ordering commitment, this becomes difficult because the builder can condition its behavior on the value of the arbitrage opportunity.

I’m also wondering whether the relevant metric should be expected LVR conditional on block-builder behavior, rather than conventional LVR under an exogenous execution process.