DemocracySim
How do different voting rules influence the temporal evolution of participation rates and inequality in a simple multi-agent system with adaptive agents?
This thesis investigates how different voting rules influence the temporal evolution of participation rates and inequality in a simple multi-agent simulation with adaptive agents. Agents are heterogeneous in their preferences and resources and repeatedly decide whether to participate in elections that aggregate individual preferences into collective decisions with payoff consequences. The environment is stationary in its update rules but evolves over time in response to collective decisions, affecting the distribution of rewards among agents. Agent behavior adapts via a fixed, explicit learning mechanism based on experienced outcomes, ensuring non-random, time-dependent dynamics. The study compares a small set of canonical voting rules (2–4) while keeping all other model components constant. Outcomes are evaluated primarily using participation rates and inequality measures (Gini index) as time-series, supported by descriptive summary statistics on collective outcomes. The thesis explicitly excludes strategic voting, complex learning models, empirical validation, policy recommendations, and normative notions such as optimal democratic design.
This project is kindly supported by OpenPetition and supervised by the Swarm Intelligence and Complex Systems Group at University Leipzig.
Read Explore Code