The Power of the Little Details

Thanks to the ability of Apple's iTunes to share music collections over local networks, it is now possible to judge someone's taste in music -- or lack of it -- in a way that previously required a certain level of intimacy.

The ability to examine the music collections of co-workers, neighbors or fellow students is akin to peering into their souls: Someone who appears cool and interesting from the outside is revealed as a cultural nincompoop through the poor sap's terrible taste in music.

And now that iTunes is widely available for both Mac and PC users, it is becoming clear that there are social implications for sharing music.

While I often spend a great deal of time explaining why machine learning is not omnipotent, this shows the flip side rather well: A large enough collection of little details can be correlated reasonably well to certain traits, and it can be very hard to "lie" to such systems.

Given that a person's playlist is built without restrictions beyond "what the person likes", I would expect that we could write a computer program to give a reasonably accurate assessment of a person's personality based merely on a playlist. The evidence? People are using the playlists for information, so there must be information there.

Now, it's totally inappropriate for large-scale law enforcement uses because the accuracy would be too low, and that's the point I'm typically making on this weblog due to iRight's subject matter. But as long as you're not trying to use it for something so important, where false positives have a truly damaging effect on somebody's life, the information so extracted can be useful; it just takes a large enough sample. Marketing, for instance.

This also provides a vivid demonstration of how casually and unwittingly your privacy can be violated. These people didn't realize how much information about themselves they were leaking until they started to suffer the social repercussions. Sure, in this case it probably seems trivial, but this is just an exemplar; much more information can be gleaned from your buying habits and such.

A lot of little details can add up to much more then the sum of the parts.