When Random isn’t so Random

I’ve noticed for some time now (yet, despite what I thought, have not blogged about) that when I listen to all my music on random, it doesn’t appear to really be random. It’s mostly random, but not random enough as I noticed some songs coming up more often than they should. At first I chalked this up to the fact that humans are horrible at figuring out stats and how often things should occur. But then I had somewhat concrete proof, I had some artists with lots of songs and those artists were being played less than artists with just a few songs. Recently I found the culprit – when Rhythmbox plays random songs, it weights the songs you like heavier so they appear more often.

Author: Eric Mesa

To find out a little more about me, see About Me

3 thoughts on “When Random isn’t so Random”

  1. My first guess would have been that “random” is actually “pseudo-random” most of the time – a seed number and an algorithm which repeats eventually. That’s how it’s done in most systems. To get real randomness into a computer, you need entropy. I forget the site, but there’s one guy who generates random numbers on his server derived from dangling a microphone out the window into the wind and translating the data.

    When I ran that method that I blogged this last week where I mutiplied ($RANDOM * $RANDOM % range), I was astonished to see the exact same two words pop up twice!

  2. Of course, it’s entirely possible that Amarok uses a similar randomizer. q;o)

    Perhaps with KDE4 I’ll got back to being a KDE fanboy and then I’ll use Amarok again. Amarok 2 looks really sweet, I can’t wait until the release.

Comments are closed.