Communication between people who would never otherwise have met (or even heard of one another) is one of the things which the internet and world wide web can be very good at. Finding and developing the best ways to harness this is always a live issue, and one promising model was launched as a beta in July of 2007, and as a full live product at the end of the year, by publishers Elsevier. I’ve been using and exploring it since October (late in the beta phase), both through my own account and with the coöperation of other users willing to act as stalking horses.
In essence, 2collab is a networking site – like Facebook or Bibo for social life, LinkedIn for business reputation structures, Advogato and ITToolbox in the computing world, Friends Reunited and so on – but designed for researchers. Not necessarily science or computing researchers, but the material which each user sees is easily tailored. To quote the original announcement of the 2collab beta: 'The history of science is based on working together – it’s all about sharing and learning. And this is what 2collab is all about.'
The basic unit of 2collab is the bookmark: a fairly full bibliographic reference for an item which is available through a web URL. The bookmark is added by a user, given as much detail as that user sees fit, then associated into information structures in various ways. Buttons can be added to your browser toolbar, one of which provides one-click bookmarking of the currently viewed site to 2collab. A bookmark can also be tagged for authorship – indicating that the user posting the item is one of its authors. Key term lists are also maintained at the bookmark level. Other users can rate, vote for, and comment on a bookmark, each of which promotes it in the viewing pecking order (see below). Clicking on the bookmark takes the viewer to the document concerned, or a 'details' button opens up a page of fuller information. Bookmarks can, individually or collectively, be imported from or exported to bibliographic managers.
The top level of the structure tree is the user, who 'owns' the bookmarks. As with most systems, the user has a profile page which carries some introductory information about her or him – name, affiliation, interests, location, email address, photograph and so forth – of which as much or as little as desired can be provided or withheld. In my own case I’ve filled in a bout two thirds of the boxes, electing to limit myself to a Europa Science frame of reference (I could have added others if I had so chosen) and the resulting example can be seen at http://www.2collab.com/user:felixgrant/profile.
Below the user comes the user group. Once again, this is owned by the user who set it up. A group can be completely open (anyone can add bookmarks to it), private (completely hidden from anyone except its members), or closed public (anyone can view, but only members can add or alter). Members are invited into the group by the user who created it. Researchers collaborating on a project, particularly in the early stages, which they wish to keep, will go for the private option. My demonstration example group (containing the last 40 or so items which I have contributed to Scientific Computing World) is closed public and can be found at http://www.2collab.com/group:Felix%20Grant%20-%20My%20SCW
An 'all bookmarks' option can also be selected, or subsets assembled by searching for specific terms, bypassing the user and group structure. Regardless of view, key terms are visible in a right hand panel (viewable as list or cloud, sorted or filtered in various ways) and can be clicked to narrow the selection.
Four ways are provided to sort bookmarks and, while useful as it is, this is an area which I feel could do with expansion. The present options are most recent, number of votes, highest user rating, or number of comments. There is what I (and several of my guinea pigs) see as a weakness in the 'most recent' setup: it shows not the date or publication, or of addition to 2collab, but how recently the bookmark was edited. If I spot a small spelling error in one of my earliest bookmarks, and correct it, that bookmark therefore appears at the top of the 'most recent' sort again. I would like to see 'publication date' and 'addition date' added to the sort choices. Less important, though useful, would be an alphabetic sort. There is an active user feedback process available, some of my suggestions have already been actioned, so I’m sure that this issue will be addressed in due course.
Use of 2collab is free to both users and readers so, having a cynical and suspicious and nature, I asked what Elsevier gains from this. It’s obviously good PR; and Elsevier’s own answer is that it gives them an edge in spotting what people want to read. If large numbers of people start following bookmarks to a particular topic, Elsevier will know about it first – which makes good sense. And are closed groups read by Elsevier? They assure me not – and, since reputation is essential in a field such as theirs, I believe them.
It's a good system which deserves to succeed.