During Ars Electronica in Linz, I got a chance to hang out with Michael from last.fm. I would have blogged about this earlier, but they have been having server problems and they wanted me to wait until they had stabilized the situation.
Last.fm has been around for awhile now and they've even been covered in Wired so many of you may already know about them. It is a music site based on collaborative filtering. Using one of the many Audioscrobbler plugins, you can set your music player to upload the titles of the music you are playing to their site. This starts to create your profile. You can also go to the site and browse songs and artists and add them to your profile. It will recommend similar artists and also show other fans of those artists. You can browse the profiles of those fans as well. Eventually, you will have enough songs in your profile for it to calculate your neighborhood. These are other members with similar taste. It's quite uncanny how similar some people's taste in music can be. You can visit these people, see what they are listening to, send them messages or add them as friends.
Once you have a healthy neighborhood and profile, the next thing you do is start listening to the radio. Last.fm is MCPS/PRS registered and has a paid license to broadcast music internationally from the UK. Only music registered with MCPS/PRS or registered directly with last.fm will be streamed, but you can listen to your own music collection, anyone else's music collection or your profile neighborhood as an mp3 stream. The web based player window will show the name of the artists, the track, the cover art, the person who's profile it is coming from and a button for "love", "ban", "skip". Anything you like will be added to your profile.
You can configure last.fm to use your local Amazon.com. You can buy most of the albums you browse on Amazon. In addition, labels can sign up on last.fm and sell music directly via downloads. Labels can set their own price. The collaborative filter allows labels to target new songs into the clusters that are most likely to be receptive to a track and the collaborative filter takes over after that.
I think this is an amazing synthesis of traditional business models from the music industry and collaborative filters. I also love how your music becomes your identity. My last.fm page shows what I'm listening to and what I kind of music I like most.
DISCLOSURE: I don't have any official relationship with last.fm yet, but I'm currently talking to them a lot and giving them my feedback and thoughts.
Michael, let me know if I got the facts right.