From Federated Assemblies to Grassroots Federation: Protocols for Fair Sortition at Scale

Citizens’ assemblies selected by sortition can deliver legitimate and diverse representation, but implementing them at scale often presumes a centralized sampling authority—potentially undermining feasibility and local sovereignty. We therefore propose and analyze a grassroots alternative: many local communities that self-organize and self-govern via sortition, and coordinate through higher-level bodies. This raises a basic design problem: how should higher-level assemblies be sampled from lower-level ones so that representation remains fair, despite overlap, hierarchy, and the integrality of seats?

In Federated Assemblies [1], the federation is a *static* hierarchy (a DAG), and parent assemblies are selected from child assemblies. The central question is which fairness constraints can be guaranteed—e.g., individual representation and proportional representation of child assemblies—both ex ante and ex post under randomized selection, and how these guarantees depend on structural assumptions. In Grassroots Federation [2], the federation *dynamically* evolves through community formation, mergers, and splits. We introduce fairness notions that separate persistent minimum guarantees from eventual proportionality and participation guarantees, and study a greedy repair-style protocol that achieves the combined objectives under stabilization. Finally, I report on ongoing simulation work aimed at quantifying convergence—how quickly different protocols approach their long-run fairness targets across synthetic and real-world federation structures.

References

[1] Daniel Halpern, Ariel D. Procaccia, Ehud Shapiro, Nimrod Talmon. Federated Assemblies. AAAI 2025.
[2] Ehud Shapiro, Nimrod Talmon. Grassroots Federation: Fair Democratic Governance at Scale. AAMAS 2026.