Information Cascades & Herding
When copying the crowd is rational — and wrong.
Sometimes it is perfectly rational to ignore what you know and copy the crowd — and that is exactly how the crowd goes wrong together. Once a few early choices line up, each newcomer sensibly follows the herd instead of their own private signal, so no new information enters and a fragile consensus locks in on a whim. How rational individuals produce irrational, reversible mobs.
A restaurant with a queue out the door pulls in more diners; a book near the top of the bestseller list sells more copies because it is near the top; an app with five downloads and an app with five million look identical on quality, yet the second one keeps winning. In each case people are not being stupid. They are doing something perfectly reasonable: when your own information is thin, watching what everyone else chose and copying it is often the smartest move you can make.
That is the unsettling heart of this course. An information cascade happens when rational people, seeing only each other’s actions and not each other’s reasons, sensibly abandon their own private evidence and copy the people ahead of them — so information stops flowing into the group, and a whim can lock a whole crowd onto a possibly-wrong choice. The individuals are rational; the herd is fragile and often wrong. This course builds the idea from the canonical urn experiment and its Bayesian arithmetic, shows why cascades aggregate almost no knowledge (and so shatter and flip at the first credible nudge), tours their real faces — queues, bestseller lists, bubbles, standing ovations, citation and research bandwagons, reputational herding — separates a genuine cascade from mere conformity and from real consensus, and ends with the cure: designs that keep independent signals alive so a crowd can actually pool what it knows.
In this topic
- 1 Why Smart People Copy the Crowd A two-minute orientation to information cascades and herding — why copying the crowd can be perfectly rational, how that same rationality lets a whole group lock onto a wrong choice, and how the course is laid out. 5 min
- 2 The Urn Game The canonical Bikhchandani–Hirshleifer–Welch experiment, worked with real Bayesian numbers by hand, showing the exact moment a rational person should ignore their own eyes and copy the crowd. 14 min
- 3 Rational Yet Fragile Why an information cascade is Bayesian-optimal for every single person yet aggregates almost no evidence — and why that information-poverty is exactly what makes the herd so easy to shatter and reverse. 13 min
- 4 Cascades in the Wild A field guide to information cascades in real life — queues, charts, viral hits, bubbles, and citation bandwagons — and how to tell each one apart from a crowd that actually knows something. 13 min
- 5 Cascades, Common Knowledge & Critical Mass How an information cascade sits between its two neighbouring models — common knowledge and critical mass — and how one shared public signal can either start a cascade or shatter one that had locked in. 13 min
- 6 Breaking a Cascade The capstone on where the cascade model misleads you — agreement is not accuracy — and the cure: designs that keep independent private signals alive so a crowd can genuinely pool what it knows. 14 min
- 7 Final Exam: Information Cascades & Herding A graded, one-way final exam on information cascades and herding — the urn mechanism and its Bayesian arithmetic, why cascades are rational yet fragile and information-poor, their real faces, reputational herding, the links to common knowledge and critical mass, and how to break a cascade. Pass mark 70%. 20 min
Mark course as finished
Done with every lesson? Lock it in — your progress is saved on this device.