Liquid democracy has developed into a broad field of research spanning political science, economics, computer science, artificial intelligence, and social choice. While transitive vote delegation is its most characteristic feature, theoretical models differ considerably in how they represent delegation, voting, participation, power, and the dynamics of collective decision-making.
In Modeling Liquid Democracy: A Survey of the (Computational) Social Choice Literature, the authors provide a systematic survey of research on liquid democracy in computational social choice. The paper addresses the fragmentation of the existing literature by identifying the fundamental modeling choices underlying different approaches and organizing them through a common taxonomy.
The taxonomy distinguishes, among other things, between different decision-making settings, different types of delegation, and systemic features of liquid democracy models. It makes explicit assumptions that are often treated only implicitly in individual studies and enables models to be compared according to their structural characteristics. This provides a common framework for locating existing research, identifying relationships between different approaches, and revealing areas that have received comparatively little attention.
The survey also argues for a broader understanding of liquid democracy than is captured by models focusing exclusively on transitive vote delegation. In practice, liquid democracy combines direct and representative participation through features such as topic-specific delegation, transitive delegation, and the ability to revoke or reassign delegations. Practical systems such as LiquidFeedback have additionally applied these mechanisms beyond conventional issue voting, including to continuous processes and collective moderation.
Building on the survey, the authors identify several directions for future research. These include extending formal models beyond one-time issue voting, studying hierarchical delegations with overlapping scopes, analyzing how delegation networks and patterns of representation evolve over time, developing more adequate concepts of power and influence, investigating partial delegation mechanisms, and improving the empirical basis of liquid democracy research through real-world delegation data and more systematic methods for generating synthetic data.
An important part of the work is a community-maintainable online taxonomy of the literature. Rather than providing only a static snapshot of the field, the accompanying website offers a curated and searchable collection of research papers classified according to the taxonomy developed in the survey. The resource is designed to evolve as new research appears and to support systematic comparison and further development of the field.
The paper also discusses the relationship between theoretical research and practical implementations of liquid democracy. LiquidFeedback appears both as an example of the practical evolution of the concept and as a source of research questions that are not yet fully represented in the theoretical literature.
The paper was presented in the Survey Track of IJCAI-ECAI 2026, the 35th International Joint Conference on Artificial Intelligence, in Bremen, Germany, on August 19, 2026, and will appear in the conference proceedings.
Publication: Proceedings of IJCAI-ECAI 2026, forthcoming
This research was supported by the European Union under the Horizon Europe project PERYCLES (Participatory Democracy that Scales) for Davide Grossi and Andreas Nitsche. Georgios Papasotiropoulos was supported by the European Union through the ERC project PRO-DEMOCRATIC (101076570).
