The Free Option Problem of ePBS
Bruno Mazorra, Burak Öz, Christoph Schlegel, Fei Wu
TL;DR
The paper analyzes the Free Option Problem arising from enshrined Proposer–Builder Separation (ePBS) in Ethereum's Glamsterdam upgrade, where builders gain a short-dated option to withhold payloads and prevent a block from becoming canonical. It develops a theoretical model showing that the option value $V^*$ and exercise probability $P^*$ rise with market volatility $\sigma$, option window length $\tau$, liquidity $L$, and external MEV share, and validates these predictions with historical block data, finding an average exercise probability of $P^* \approx 0.82\%$ (for $\tau=8$ s) that spikes during high-volatility periods. The authors quantify how builder heterogeneity and cross-domain signals affect liveness, and propose mitigations — shortening the option window and imposing penalties — including a dynamic penalty mechanism via online gradient descent that can approach target exercise rates with manageable costs. They also discuss broader implications for on-chain price formation and for other time-sensitive applications like prediction markets. Overall, the work highlights a fundamental scalability-security trade-off in ePBS and offers protocol-level strategies to preserve liveness while preserving scaling benefits.
Abstract
Ethereum's upcoming Glamsterdam upgrade introduces EIP-7732 enshrined Proposer--Builder Separation (ePBS), which improves the block production pipeline by addressing trust and scalability challenges. Yet it also creates a new liveness risk: builders gain a short-dated ``free'' option to prevent the execution payload they committed to from becoming canonical, without incurring an additional penalty. Exercising this option renders an empty block for the slot in question, thereby degrading network liveness. We present the first systematic study of the free option problem. Our theoretical results predict that option value and exercise probability grow with market volatility, the length of the option window, and the share of block value derived from external signals such as external market prices. The availability of a free option will lead to mispricing and LP losses. The problem would be exacerbated if Ethereum further scales and attracts more liquidity. Empirical estimates of values and exercise probabilities on historical blocks largely confirm our theoretical predictions. While the option is rarely profitable to exercise on average (0.82\% of blocks assuming an 8-second option time window), it becomes significant in volatile periods, reaching up to 6\% of blocks on high-volatility days -- precisely when users most require timely execution. Moreover, builders whose block value relies heavily on CEX-DEX arbitrage are more likely to exercise the option. We demonstrate that mitigation strategies -- shortening the option window or penalizing exercised options -- effectively reduce liveness risk.