The things I think Last.fm has going for them: * Charts (of course) * Recommendations: I use last.fm for two kinds of recommendations: (1) where it uses listener history to figure out similar bands and albums, which in some ways is inferior to feature-based recommender systems like Pandora, and (2) where it uses neighbor listening history (e.g., neighbor radio) -- I can't count the number of times I've been introduce…
http://musicmachinery.com/2011/02/22/is-the-kdd-cup-really-m...
Because it's entirely anonymised, not just the users but the artists too -- c.f. Netflix's problems with deanonymization:
http://33bits.org/2010/03/15/open-letter-to-netflix/
This means you can't use any interesting characteristics of the music itself, or the associated metadata, to aid the recommendations. All the interesting domain knowledge is stripped out, which likely means the best solutions still won't work as well as algorithms that use metadata (like Last.fm's) or content analysis (like Pandora's) or both, and certainly won't lead to any particularly interesting insights about what drives people's tastes.
Disclaimer: I work at Last.fm