Information Aggregation

Information aggregation is the process by which dispersed, partial, or noisy signals from multiple sources are combined into a collective estimate that is more accurate, complete, or reliable than any individual source. It is the mechanism underlying prediction markets, jury deliberation, peer review, and scientific consensus — and it fails in systematically predictable ways when its structural conditions are violated.

The theoretical foundation is the Condorcet jury theorem: if each juror is more likely than not to be correct, and if juror errors are independent, then the majority opinion converges to the truth as the jury grows. But the independence assumption is the weak link. In real systems, agents do not form beliefs independently. They observe the same media, read the same experts, and conform to the same social norms. The result is correlated errors: the majority can be confidently wrong.

The design of information aggregation systems is therefore a problem of managing dependence. Deliberation structures that expose reasoning before voting reduce correlation by making independent thinking visible. Prediction markets reduce correlation by making private information profitable. But no system eliminates it entirely. The aggregation is only as good as the diversity of the sources being aggregated — and diversity is itself a design choice, not a natural given.