InfoDelphi Framework for Multi-Agent Forecasting via Information Asymmetry

InfoDelphi improves multi-agent forecasting by partitioning evidence into shared public and disjoint private subsets to prevent herding. By introducing designed information asymmetry, the framework reduces inter-agent error correlation during deliberation.