I’ve been reading about last.fm and Audioscrobbler for a few months now, and was intrigued by what I’d heard. But I didn’t totally understand it, and I didn’t have time to explore it—the last thing I needed during these last couple of months was another computer-based time sink.

But now that the Lab’s gone live, and the holiday break has begun, I’m getting a chance to try it out—and I’m totally delighted with it. It’s a brilliant idea.

Here’s how it works:

1. You sign up for a free account with last.fm
2. You download a free Audioscrobbler plugin to work with your music player of choice and configure it with your last.fm login info
3. You play enough music for the system to learn about your tastes. (I put iTunes on “party shuffle” and let it play continously for a while, turning off the sound when I didn’t want to listen.)
4. You go to the last.fm site and click on the “Profile Radio” button near the top of the page. The system finds people with musical tastes similar to yours, and starts playing music from their collection. (This is all legal, btw…read the FAQ for details.) If it plays a song you love, click the “love” button and it gets ranked higher in your profile; if you don’t like it, click “skip” and it goes to the next song. Hate it? Click “ban” and you’ll never hear it again.

How cool is that? A personalized radio station that (a) learns what you like, (b) lets you skip songs you don’t want to hear, and © doesn’t play music you’ve said you don’t like.

There are other social features built in—you can add friends (people you like, who are different from “neighbors” that share your musical tastes), chat with people, participate in forums, etc. But the beauty of this for me isn’t in the explicit social behavior, it’s in the implicit recommendation and customization process.

Which got me thinking about definitions of social software and social computing. Most of the ones I’ve seen have focused on direct, intentional communication between two or more people. But what about systems where the communication is implicit, where the social component is the emergent information that comes from multiple users, rather than any direct exchange between or among those user? Food for thought as I work on the LSC wiki.