Timing Games on Ethereum in 2026: The Attester Externality of Block-Proposal Delay and the Absence of a Consolidation-Resolved Yield Gap

Ethereum block proposers can delay publishing their blocks within a slot to capture additional maximal extractable value (MEV), a practice known as timing games. Delay is privately profitable but imposes a negative externality on attesters: validators who have not received the block before their voting deadline cast an incorrect head vote and forgo the corresponding reward. Using public Xatu data across a contiguous 28-day sample (10 May to 6 June 2026, post-Pectra), we measure both sides of this practice simultaneously and resolve them by validator size for the first time on this data. We find that delay is the norm, the median block first appears roughly 1.5 s into the slot on every day, and that the correct-head-vote rate collapses to approximately 50% for blocks published after 3.5 s. The lower-bound attester cost averages approximately 3.0 ETH per day (range 1.8–4.9), concentrated in late blocks and elevated on weekends due to worse propagation conditions. The proposer’s captured delay premium averages approximately 0.004 ETH per block, exceeding the per-block externality by roughly an order of magnitude; the two quantities are uncorrelated across days (Pearson r = 0.08). Testing the centralization hypothesis using validator consolidation as a size signal, we find no MEV-capture gradient and only a negligible attester-side yield gap (approximately two basis points) along the consolidation axis. Because validator consolidation is a partial proxy for operator type and is distinct from committee share, this null is specific to the consolidation channel and does not rule out a committee-share effect, which our on-chain proxies cannot measure; committee share is thereby isolated as the only remaining viable route to a centralization effect.Manuscript under review at Ledger.