Can we design *a* community for *scientists* My test essay on designing an online community has gotten some comments on FriendFeed and has been linked-to from twitter (this searched works at the time of writing). It's great to have this kind of quick feedback.

First, I think Deepak is absolutely right. He questions whether there is a generic type of scientist such that you can design a community for all scientists. The literature shows that you need to design a community with a clear purpose and clear policies. The participants have to have common interests to bring them together. Maybe what I meant (and didn't say!) is that these are minimal requirements for communities to be useful to people involved in science, and that additional features and functionality are required for the specific group of users or purpose for the site.

Could there be very generic tools (facebook, friendfeed), generic science-y tools, and then much more specific research area or even a specific collaboration tool? Sure, but they we also get into overload. Where do you post this update? Check each of these or get e-mails and feeds from each of these? What happens when your contacts post the same info in facebook, twitter, and friendfeed... and it's not even their own content, but someone else's that you also follow. Kind of annoying - is there a way to deduplicate? Should there be? maybe the content only appears once with comments and links showing from all of the various places?

What Mr.Gunn says is right, too, there has to be some up front value to get people to even try any site. There has to be a percieved relative advantage, there has to be various types of knowledge, and there has to be a low-risk way to trial the site. I don't really agree that just being first is enough. There were other photo sharing sites before Flickr that weren't as popular. Flickr did some things that make it more useful.

Recommender systems are great - ScienceDirect's related articles usually works pretty well for me (ducking bcs I liked something from the Evil Empire - sorry Bill!). PubMed doesn't seem to. From what I can tell, SD uses the full text whereas PubMed uses only the MeSH indexing. I'm not sure how many people have to be on Mendeley for it to really work. Individual papers might be saved for lots of different reasons, too, - it's almost like you need to say why you're keeping it - same organism, good method, author from x lab, person a recommended, etc. I think there's a lot of value about collaborating and communicating around scholarly publications. I need to probably post something more thoughtful about recommender systems for articles... I have some thoughts on that, too.  ¶ 6:57 PM| |cites (technorati) |