The Comprendia Blog has an entry about social networks in the sciences: Are Any Social Networks for Life Scientists Gaining Traction?
It is a report of a 'quick and dirty' analysis of visitors to a range of sites. Not everything you might expect is included; Mendelay for example does not feature because of its local component.
As I mentioned, this is a quick analysis, and I invite others to copy the data and add to it (here's the Google Spreadsheet), but a clear trend emerged from this short study. The top site, biomedexperts.com, which has an estimated 128,000 visitors per month, utilizes scientific publications to connect and segment scientists based on interests or regions. The second most popular site, citeulike.org, is a social bookmarking site with about 100,000 visitors per month who use it to search for, store, and browse related peer-reviewed publications. What's interesting is that these sites are based on completely different formats, but they both rely on scientific publications.
Is it surprising that scientific publications are the 'glue' that is connecting and attracting scientists? Not really-when I think back to my days in the 90's as a bench scientist, PubMed and my many folders of papers were what helped me to connect with other scientists and learn who the thought leaders of my field were. In a sense, PubMed was scientists' first 'social network.' [Comprendia]
An important strand in discussion of networks suggests that people connect and share themselves through 'social objects', pictures, books, or other shared interests, and that successful social networks are those which form around such social objects. Think of LibraryThing or GoodReads and books, or Flickr and photographs. What I thought was interesting in the post was the focus on the incentives required to create network effects and a suggestion that these related to making connections around 'social objects', in this case scientific publications. It is also interesting to see the suggestion that the literature itself has always had this social element.
There is also a link to an interesting post by Cameron Neylon 'What should social software for science look like' which discusses various scientific research outputs as social object.