In this paper we describe an evaluation of behavioral descriptors generated from an analysis of a large collection of Usenet newsgroup messages. The metrics describe aspects of newsgroup authors’ behavior over time; such information can aid in filtering, sorting, and recommending content from public discussion spaces like newsgroups. To assess the value of a variety of these behavioral descriptors, we compared 22 participants’ subjective evaluations of authors whose messages they read to behavioral metrics describing the same authors. We found that many metrics, particularly the longevity and frequency of participation, the number of newsgroups to which authors contribute messages, and the amount they contribute to each thread, correlate highly with readers’ subjective evaluations of the authors.