Zipf's law and the diversity of biology newsgroups

Abstract

Usenet newsgroups provide a popular means of scientific communication. We demonstrate striking order in the diversity of biology newsgroups: Submissions to newsgroups obey a form of Zipf's law, a simple power law for the frequency of posts as a function of the rank, by posting, of contributors. We show that a simple stochastic process, due to Günther et al. (1992, 1996), Levitin and Schapiro (1993), and Schapiro (1994), accounts for this pattern and reproduces many of the properties of newsgroups. This model successfully predicts the relative contribution from each poster in terms of the size, the number of posters and total posts, of the newsgroup.