Information cascade
People rationally copy earlier choices and ignore their own private signal.
What it means
An information cascade occurs when individuals, deciding in sequence and observing the choices (but not the private information) of those ahead of them, find it rational to disregard their own signal and imitate the crowd. The mechanism is Bayesian: once a few early actors have chosen the same way, their accumulated choices appear to carry more evidential weight than any one person's private signal, so each subsequent chooser optimally follows suit — and contributes no new information, since their action reflects the herd rather than what they actually knew. This is what makes cascades fragile and potentially wrong: because private signals stop being aggregated, the entire crowd can lock onto an inferior option, and the cascade can reverse abruptly when a salient new piece of public information arrives. It differs from simple conformity in that the actors are behaving rationally on the available evidence, not merely caving to social pressure. It matters for understanding fads, fashion, bank runs, technology-adoption races, viral content, and financial bubbles, and it underlies why early movers and visible 'social proof' can disproportionately steer collective outcomes.
Examples
Diners choosing between two empty-looking restaurants pick the one that already has a few customers, and the gap widens all evening regardless of which kitchen is actually better.
An app that reaches the top-ten chart keeps climbing, because each new download is prompted by the ranking rather than the app, and adds no fresh evidence that it is any good.
Depositors who see a queue outside a bank join it; the queue, not the bank's books, becomes everyone's reason to withdraw, and the run can start on a rumour.
First described in Bikhchandani, Hirshleifer & Welch (1992).