A brief analysis of why MySpace is successful as a space for music (for which it was originally designed):
- You can listen to full tracks by each band.
- There's no need to find and navigate complex Flash-based artist websites.
- There's a direct communication channel to and from the artist - via the blog outwards and via the comments inwards.
- You can see the band's social network in context - their friends, people who like them.
- There's a standard gig listing template (though this is the weakest point, it should really be location-aware and aggregatable to show you which of your favourite artists are playing nearby in the next few weeks - see SonicLiving).
How about an analogy to science publishing:
- You can't read full papers (most of the time).
- There's no central (or distributed) repository of papers, though PubMed Central is a start. Finding papers is easy, but navigating journal sites can be awkward. Finding author profile pages (generally home pages on university sites) is hard, and their list of publications is often out of date.
- There's no public communication channel with the authors: you can send them an email, and some journals provide space for comments, but they're not very visible and I don't know if the authors are notified when a comment is posted.
- There's no indication of an author's collaboration network.
- There's no unique identifier for authors (except in some places: Scopus and SciLink), so getting a list of all an author's publications can be difficult.