I’ve been actively watching last.FM lately  . I believe that the value of this tool has yet to be truly uncovered (partially because it’s buggy as hell and there are key features missing). Still, i think that it is quite relevant for this discussion and important issues arise when considering it.
In an article entitled “The Focused Organization of Social Ties,” Scott Feld discusses the role of interests in social networks. You and i may both know five people, but if we also have five diverse interests in common, we are far more likely to get along. Furthermore, if you and i have five diverse and rare interests in common, we are very likely to know many people in common. In his article, he introduces foci to the structure of social networks, emphasizing that “foci tend to produce patterns of ties, but all ties do not arise from foci” (1018). Foci are not simply interests, but also people, places, social roles, etc. [Anyone interested in issues around modeling foci in social networks in applications *must* read this article.]
Music is one of the best focis out there. We naturally turn to our friends for music recommendations. People access music through their friends and people (most notably subcultural youth) find friends through music. Particularly among younger groups, i would posit that much about social network can be understood through music distribution and tastes. [Anyone have good research on this?]
Music is a cultural foci, one that i think has a lot of salience for these tools. It is present in most of the sociable articulated social networks and the most important factor for MySpace (built on indie rock bands) and Tribe.net (built on Burning Man culture which is fundamentally a music/art festival). Yet, it is last.FM that takes it to the next level and lets you connect for and because of the music, directly appreciating others’ music tastes.
This is not to say that there aren’t oddities involved. Your behavioral musical profile says so much about you that articulated versions of your tastes do not. I have already found myself temporarily banning certain artists because their dominance in my profile gives the wrong impression of my tastes. In other words, the visibility of my behavior has resulted in a behavioral change. That is indeed a very interesting end result of publicly visible behavior-driven social data.