Monday, April 11, 2011

On Randomness

Humans have a hard time grasping true randomness. We look for meaning and patterns everywhere, even when we know that something really is random.

For example, when people are asked to distribute dots randomly on a page, the dots end up mostly evenly spaced, when truly random would be, well, random - meaning they are most likely to not be evenly spaced. Each sequential dot has an equal chance of being anywhere on the page as the first one did. If the dots are purposefully not put close to the earlier dots, then it's no longer random - it's following a certain set of rules. Out of the billions of possible combinations, each dot being placed on top of the earlier one, creating what appears to be one single dot in the middle of the page, is just as likely as any other combination. Similarly - as another popular example goes - if you have a thousand monkeys typing on typewriters forever, one is bound to create a Shakespeare.

At an iPod and iTunes media event a few years ago, Steve Jobs announced that Apple was introducing a setting in iTunes that would allow users to adjust the Shuffle feature so that it would play songs from the same artists less frequently. Prior to that adjustment, the Shuffle feature was random. That meant that if you had 500 songs on your iPod, each song had a 1/500 chance of being played right after you pressed "shuffle." Then each of the other songs has an equal chance of being played second and so on. It's equally possible that the first ten songs are from the same album as that the first ten songs are from different albums and different artists (although not equally probable). Our expectation of the Shuffle feature was that the songs would come from different artists - which, by following a certain set of rules, isn't random. When you have millions of people using iPods, a certain number are bound to see a disproportionate amount of sequential songs from the same artist or album after pressing Shuffle.

No comments:

Post a Comment