Ouroboros Taktikos: Regularizing Proof-of-Stake via Dynamic Difficulty

In any Nakamoto-style distributed ledger technology, participants must eventually come to a deterministic ordering of proposed extensions as a necessary precondition to consensus. To prevent Sybil attacks, this process encodes a bias toward selecting proposers who commit a limited resource. In proof-of-work (PoW) schemes, block proposals are secured using a hashing mechanism that preferences miners with greater computational power. In Nakamoto-style proof-of-stake (PoS) paradigms, proposers’ eligibilities derive from their staked holdings, relying on a thresholding mechanism to fairly select the next block proposer by favoring those who have committed more stake. This eligibility threshold controls which parties may propose a block in a given slot, with easier thresholds increasing block density. However, higher density also increases forking – periods of uncertainty where consensus remains (temporarily) unsettled. Therefore, the selection of the PoS eligibility threshold critically affects both security and throughput in the associated chain. Previous work relies on static threshold values, which simplifies security analysis. In this work, we extend the static eligibility threshold to a dynamic one, and introduce the concept of a local dynamic difficulty mechanism in which thresholds follow a non-monotonic difficulty curve. We implement this mechanism in a novel PoS protocol, Ouroboros Taktikos, finding that the dynamic regime regularizes slot intervals and improves block throughput. The pseudo-predictable nature of the protocol also penalizes covert attacks, simultaneously increasing security. We compare Ouroboros Taktikos to Ouroboros Praos and show that the addition of local dynamic difficulty improves throughput by \(\sim \) 2.9 \(\times \) and reduces 99 \(^\mathrm{{th}}\) percentile block latency by \(\sim \) 5.7 \(\times \) compared to the state of the art.