Develop a more realistic adversary model for blockchain fee analysis

Develop a formal adversary model for blockchain transaction and resource-demand generation that is more natural and realistic than an essentially all-powerful adversary, to better reflect constraints and behaviors encountered in real blockchain systems.

Background

The paper establishes O(1/sqrt(T)) average regret guarantees and a matching lower bound under an extremely strong (all-powerful) adversarial model that can choose transaction sets and utilities subject only to resource limits. While this yields robust guarantees, the authors note that the adversary’s power may be overly pessimistic for practical settings.

They explicitly call for a more natural adversary model, which could enable sharper analyses and potentially stronger performance guarantees for fee-setting algorithms.