gnod.net is a fun site that will take up a few minutes of your time if you let it. At it’s most basic, gnod (global network of dreams) recommends books, music, and movies based on things you like. Created by Marek Gibney of Hamburg, Germany, gnod is, by the author’s definition:

…my experiment in the field of artificial intelligence. Its a self-adapting system, living on this server and ‘talking’ to everyone who comes along. Gnods intention is to learn about the outer world and to learn ‘understanding’ its visitors. This enables gnod to share all its wisdom with you in an intuitive and efficient way. You might call it a search-engine to find things you don’t know about.

gnod interviews you by asking you to create a set of your favorite artists, then moves you into a series of “do you like ?” questions. This form of learning tree is not so new, is slightly annoying to me, and gleefully employed in some form by big commerce like amazon and blockbuster. However, gnod does one better by displaying all related artists/books/movies in a floating, JavaScript-y cloud view.

Normally, I’d start to babble about how cool the algorithm is (Gibney does not reveal the details) or how we love applied technology and it makes our lives so much better, but in this case, I think the results are sort of interesting, but somehow wack ironic.

Case in point… Al Green. gnod will place Al green at the center of the cloud, and the closer an artist is to Al, the more likely people’s tastes will overlap. The O’Jays are pretty close to Al, which makes sense to me, but Marvin Gaye and Curtis Mayfield are hangin’ out the same distance from Al as… UB40. The same goes for Bill Withers. And you know what? The Bee Gees are pretty freakin’ close to Al on that map. Am I missing something here, or are we just dealing with a system that hasn’t seen the sample it needs? I suppose in my own insulated way, I don’t encounter the Al Green x Bee Gees fans that gnod does.