Fooled by randomness

Stephen Downes has a terrific post about selective attraction. It’s a warning against ascribing magical properties to people and things that get to the top of popularity.

He describes an experiment in which we take 1000 tunes on an iPod and shuffle them. We add one small tweak:

Instead of selecting completely randomly, suppose you include a slight preference to songs you’ve already played. To say a song would have a 10 percent greater chance of being selected if a song were played 10 times more than another.

This sounds like a small tweak, but actually it soon generates a very uneven distribution, with a few tunes getting a lot more airplay.

Now the kicker. If you then reset the shuffle, with the exact same algorithms, you get the same unequal distribution – but with quite different tunes at the top. Read the whole thing for the explanation; its classic complexity. I’ll just skip to his conclusion, which I think we could all do with keeping in mind more of the time:

Don’t equate wealth with knowledge, fame with insight. Don’t equate the accidental properties of a network phenomenon to intrinsic worth or value. The differences between us are far slighter than the disparities in wealth, power or fame would indicate.

Hat tip: Tim Kastelle’s tweet

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.