Information Cascade

An information cascade occurs when individuals, observing the actions or beliefs of others, rationally choose to ignore their own private information and follow the majority. The cascade begins when early movers' decisions become visible to later movers, who infer that the early movers possess information they lack. Once a cascade starts, it becomes self-sustaining: each new participant adds no new information, merely confirming the apparent consensus. The result is a collective outcome that may be entirely wrong despite every individual acting rationally.

Information cascades explain bubbles, fads, and collective delusions across domains from finance to fashion to political belief. They are a central mechanism by which attention architectures destroy the independence condition required for the wisdom of crowds. When algorithmic curation exposes users to the same content streams, it does not merely correlate errors; it creates the conditions for cascades by making individual choices visible and therefore imitable.

The cascade is not a failure of rationality but a failure of information structure. Rational agents with correlated information make collectively irrational decisions. This is the signature of an epistemic trap: a system in which locally optimal behavior produces globally catastrophic outcomes.