"When Markets Go Mad" by David Appell. Scientific American, May 2002, pages 28-29.
Being able to predict extreme behavior of complicated systems like stockmarkets and traffic jams would carry huge benefits. While some extreme eventsstem from random external happenings (like the stock market's post-September 11dive), others are internal to the system. The article postulates a sort ofunintentional cooperation among the various agents (e.g. traders, drivers) ineach system that leads to large changes. Normally, there is a great diversityof belief among the agents about the future behavior of the system. But when,for whatever reason, most agents begin to have similar feelings about where thesystem is headed, the stage is set for big change.
--- Rafe Jones