Nick Bradbury has written about attention data on his blog several times. What I see as an interesting use of attention data (that fortunately costs you some money. Yes, fortunately. You get 10 hours free to try it out.) is the site Pandora.com. While it may sound like you would be opening up a world of trouble by clicking on that link, I challenge you to check it out. You can type in artist names and Pandora will then start selecting songs that might fit your taste. What’s interesting is that you can rate the songs (click on the song in the player and choose from the popup whether or not you like it) and then they are able to take a list of songs that you’ve liked and can then compile a list of songs that you might like based on your attention data. I am amazed at the value of this sort of attention data.

Pandora’s musical evaluation of the songs is based on a complicated grid of characteristics that basically rank each song by each artist that has been rated by their group of music evaluators by various qualities so that if you’re a big fan of certain songs with various characteristics you’ll most likely get other songs with similar characteristics. Other bands with songs that sound similar will be brought in so that you can be introduced to new music. Sure, you could listen to the radio and hope that the record labels happen to be promoting the next thing that isn’t just a cookie cutter cut out of other music but this is quite amazingly good at nailing my musical preferences. As a former music department manager I can attest to the challenge of doing this manually. A database of music is just amazing in this context.

So, you pay them to use your attention data to be exposed to quality (ok, that’s a bit subjective) music, and then they learn from you to help play what you like but also they can take your attention data and discover what various tastes are and can give even better suggestions to other users.

Brilliant.

Thanks to Robin for pointing me to Pandora.