Listening to a song just now from the great compilation I am the Resurrection – A Tribute to John Fahey, I was thinking that I’d really, really like to hear more music like what I heard in that song. But where do I start? How do I describe that music? How would I search teh intarwebs to find such things? If I put “acoustic jangly music” into Google, I doubt I’d get what I’m looking for. Further, I wondered what it’d be like to search for images with only basic cues, like “cat fuzzy” and be presented with images specifically of fuzzy cats and nothing more.

What I’m looking for is a more abstract search – which I think is something that humans do every day in their minds. The equivalent of asking the video-store clerk “You know that movie with that guy with that shiny thing on his head,” or like what Kelly refers to as “fall music” – music that somehow evokes the feelings of the season that is “fall”. Obviously this isn’t something that you can easily divine from a filename, and even the search terms like “fall music” mean different things to different people, so it’s a tough challenge.

As you might imagine, Google’s already pushing things in this direction. In a small step to make their image search better (it’s already fairly awesome), they introduced the Google Image Labeler a while back. It puts you head-to-head with someone else to help “tag” the images with words of what you see in the photo. If you see a cat in the image, you can put in the word “cat”. Your description doesn’t get accepted unless the anonymous person on the other end puts in the same word, thereby ensuring some level of quality.

The problem is that humans are intelligent, but not quite intelligent enough to transfer equivalent intelligence into machines. This is a whole branch of science, as you might imagine, and we are certainly pushing things forward every day. It’s only a matter of plumbing the depths of our own minds and learning just how we learn.

The mind, it boggles.