The Optimal Size of an Epistemic Congress

• DOI: 10.48550/arxiv.2107.01042
• Publication Date: 2021-01-01
• Authors: Manon Revel; Tao Lin; Daniel Halpern
• Keywords: FOS: Computer and information sciences; Artificial Intelligence (cs.AI); Computer Science - Artificial Intelligence; Computer Science - Computer Science and Game Theory; 05 social sciences; 0506 political science; Computer Science and Game Theory (cs.GT)

We analyze the optimal size of a congress in representative democracy. take an epistemic view where voters decide on binary issue with one ground truth outcome, and each voter votes correctly according to their competence levels $[0, 1]$. Assuming that we can sample best experts form congress, find should be linear population size. This result is striking because it holds even when allowing top representatives accurate arbitrarily high probabilities. then real world data, finding actual sizes congresses are much smaller than our theoretical results suggest. conclude by analyzing under what conditions sub-optimal would still outperform direct democracy, which all vote.