The current edition of the peer-reviewed web research journal First Monday has a fascinating article on Long Tail statistical models by Kalevi Kilkki, a Principal Scientist at Nokia Siemens Networks, Finland. He takes data from markets that range from books and movies to web searches and names, and tests them against a classic Long Tail powerlaw model. By and large, they fit nicely .(Which isn't a surprise, since the observation that much of economics and human affairs can be described by a powerlaw goes all the way back to the "80/20 Rule" insights of Vilfredo Pareto in the 19th Century.)
I've just had a chance to skim it, so I won't say more than it looks well worth reading for the quantitative amongst you. My only criticism on this first pass is that in several markets, such as movie box office revenues, Kilkki does not really address distribution bottlenecks, such as the limited number of screens available. Instead, he blames the rapid fall-off of revenues on limited marketing, which is only part of the picture, so to speak.
[Thanks to Barry Ritholtz for the link]