This is a very interesting paper that quantifies patterns in user generated content. First, "There is no such thing as an average user. UGC production is not a normal distributed process. From the contributing users, the majority contribute few items, whereas few contribute a lot." And second, "Three groups can be inferred from the data. 1) Amazon Reviews, Digg, FanFiction and SlideShare seem to have a similar distribution of contribution. The 10% of the users contribute from 40% to 60% of the content. 2) a group is integrated by the 'fatbelly' IPPs: Furl, LibraryThing and Revver. In those cases, the 10% of the users contribute between 60% and 80% of the content. 3) Scribd and Merlot form the third group. In this group, the most prolific sources seem to have a big impact in the overall number of items contributed." I want to venture as a hypothesis that the first group constitutes 'network' behaviour, and will demonstrated considerable diversity, while the second and third represent increasingly 'group-like', or less diverse, behaviour; that there will be characteristic network 'shapes' that correspond to each of these three groupings. Via Erik Duval's Weblog.