Divide or Confer: Aggregating Information without Verification

2025-07-23

AI Summary

This paper studies optimal information aggregation mechanisms for binary decision-making when a receiver faces multiple unverifiable, preference-biased senders. The core challenge lies in jointly achieving incentive compatibility and information efficiency when signals are dispersed across many senders and biases are unobservable. Method: The authors develop a Bayesian incentive-compatible-in-the-limit (ICL) framework under large-population asymptotics, constructing high-dimensional mechanisms that integrate private deliberation with consensus-based punishment. Contribution/Results: They establish that deliberative mechanisms are optimal for small populations, while large-population mechanisms converge to a limit exhibiting an irreducible asymptotic utility loss. The optimal mechanism explicitly depends on the receiver’s acceptance payoff and mitigates bias through penalties for deviations from consensus reports. Crucially, this work is the first to systematically uncover the structural advantage of information fragmentation in large-scale biased environments, providing a novel mechanism design paradigm and theoretical benchmark for distributed information aggregation.

Technology Category

• Multiagent Systems: Mechanism Design
• Game Theory and Economic Paradigms: Mechanism Design
• Reasoning under Uncertainty: Decision/Utility Theory

Application Category

• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

We examine receiver-optimal mechanisms for aggregating information that is divided across many biased senders. Each sender privately observes an unconditionally independent signal about an unknown state, so no sender can verify another's report. A receiver has a binary accept/reject decision which determines players' payoffs via the state. When information is divided across a small population and bias is low, the receiver-optimal mechanism coincides with the sender-preferred allocation, and can be implemented by letting senders emph{confer} privately before reporting. However, for larger populations, we can benefit from the informational divide. We introduce a novel emph{incentive-compatibility-in-the-large (ICL)} approach to solve the high-dimensional mechanism design problem for the large-population limit. We use this to show that optimal mechanisms converge to one that depends only on the accept payoff and punishes excessive consensus in the direction of the common bias. These surplus burning punishments imply payoffs are bounded away from the first-best level.

Problem

• Optimal mechanisms for aggregating biased information
• Handling information divide in large sender populations
• Preventing excessive consensus via surplus burning punishments

Innovation

• Private conferring among biased senders
• Incentive-compatibility-in-the-large (ICL) approach
• Surplus burning punishments for excessive consensus