Liquid democracy systems with transitive and suspendable delegations, as implemented in LiquidFeedback, raise a fundamental question: how should the power of individual participants be measured when delegations can propagate through a network but can also be suspended by voting directly?
In Power in Liquid Democracy: A Network Centrality Approach, the authors develop a computationally tractable framework for measuring power in such systems. They introduce expected voting weight, which captures the voting weight a participant can expect to exercise when voters independently decide whether to vote directly or rely on their delegations.
The authors show that expected voting weight corresponds to Random Walk Decay centrality, thereby connecting the analysis of liquid democracy with methods from network analysis. They provide an axiomatic characterization of the measure and compare it with alternative centrality-based approaches.
The framework also supports the analysis of power distributions in delegation networks and the identification of representative sets of influential participants.
The paper was first presented by Davide Grossi at the 3rd Workshop on New Directions in Social Choice, co-located with the ACM Conference on Economics and Computation (EC 2026), in Rome, Italy, on 6 July 2026.
Access paper on arXiv | DOI: 10.48550/arXiv.2608.13188
This research was supported by the EU Horizon Europe project PERYCLES for Davide Grossi and Andreas Nitsche, the ERC project PRO-DEMOCRATIC (101076570) for Georgios Papasotiropoulos and Piotr Skowron, the National Science Centre (2023/50/E/ST6/00665) for Oskar Skibski, and the EPSRC (EP/X038548/1) for Tomasz Wąs.
