Metamerist was musing abut Pigeonholing Algorithms & Self-fulfilling Prophecies this morning, a subject that I’m kind of interested in as well.   We’ve all used these systems which try to evaluate our choices in music or books, and then make suggestions based upon what we say.   They are sometimes useful, but most often useless in trying to find new material that’s of interest.

The reason is actually pretty easy to understand.  Let’s take books as our example domain.  Amazon knows that recently I’ve bought books on baseball, on the chinese language, on filmmaking, and on computer graphics.   So what does it do?  It gives me more of the same.   It’s not a bad idea really: there are undoubtedly lots of books in these subjects I don’t have.   But they are unlikely to actually suggest books in other, even related subjects that might be of interest.  Does an interest in filmmaking suggest an interest photography?  If I’m interested in Chinese language, might I not be interested in Chinese food or Chinese history?

It seems most of these systems work to dramatically reduce the number of possibilities that they present to you rather than expand them.  In that sense, they just reinforce your narrow tastes, rather than help you expand and educate them.  That’s too bad.

What’s the answer?  What algorithm will help?  Frankly, I don’t know.   The way that I deal with it is to read a wide variety of sources, keep a wide variety of interests, and actively work to break out of the ruts in your thinking.  Don’t be afraid to explore deeply, but also work to explore broadly.   Make it your responsibility, not just the responsibility of software.  Invest time in making yourself more interesting.

[tags]Rants and Raves[/tags]