A Take on Quick Slots
Co-authored with @sarah.allen, @guayabyte @BrunoMr, @Christoph, @dataalways, @Hasu, @Quintus
EIP-8198 introduces infrastructure for variable slot timing, allowing Ethereum to gradually shorten its slot duration as client and network bottlenecks are better understood.
We support this direction. Shorter slots reduce confirmation latency and can enhance user experience, while also improving the responsiveness of on-chain markets. Faster feedback and settlement can, in turn, benefit trading activity and market efficiency, helping to ensure the continued robustness and relevance of Ethereum as a platform for decentralized finance. The proposal’s gradual approach allows these benefits to be pursued while empirically testing the limits of Ethereum’s infrastructure.
One consideration in this process is geographical decentralization [1]. Our research [2, 3, 4] studies how latency and the geographical distribution of validators, builders, relay intermediaries, and information sources shape incentives in block production. We already observe these forces in today’s PBS market, where improvements in relay and builder infrastructure have made minimizing builder— relay latency increasingly important, thereby amplifying the payoff from colocating with relays [2]. Shorter slots can further tighten the time available across the block-production pipeline. As communication latency becomes a larger fraction of this available window, participants geographically distant from major network and infrastructure hubs can face a greater disadvantage, strengthening incentives to colocate with payoff-relevant entities.
However, this is unlikely to be a material constraint at the slot times currently under discussion. In latency-calibrated simulations, reducing the slot time from 12 seconds to 6 seconds increases regional reward disparities but leaves geographical centralization outcomes largely unchanged [3]. Internet propagation delays remain small relative to the available timing window. This gives us confidence that reductions within the currently discussed range, such as from 12 to 10 seconds and potentially to 8 seconds, are unlikely to materially increase pressure for geographical centralization under today’s network conditions.
We therefore support making slot timing configurable and experimenting with gradually shorter slots. At the same time, geographical incentives should remain part of the evaluation as Ethereum continues reducing slot times, alongside continued exploration of mechanisms that counteract concentration incentives, e.g., through diversity-aware reward design across dimensions such as geography, operator, or infrastructure provider, and efforts to make the protocol less latency-sensitive in the first place. Geography is less a constraint on the reductions being considered today than a potential bound on how far slot times can ultimately be pushed without creating additional pressure toward geographical centralization.