From Wired News:
My playlist has a total of 17 songs by the band, so it seemed highly unlikely that two of them would be bunched so close together in a random order. But I was wrong about that.
The problem, it turns out, isn’t that the programs aren’t randomizing my playlists. They are. According to Jeff Lait, a mathematician and author of randomm3u, it’s what’s happening between my ears, specifically, in my expectations of what it means for something to be random.
To illustrate his point, Lait referred to a phenomenon statisticians call the birthday paradox. Roughly stated, it holds that if there are 23 randomly selected people in a room, there is a better than 50-50 chance that at least two of them will have the same birthday. The point: Mathematical randomness often contradicts our intuitive expectations of randomness.
What we want, Lait says, isn’t a list that’s been randomized, but one that’s been stratified, or separated into categories that are weighted by a listener’s preferences. A stratified playlist might select songs randomly but would be smart enough to throw out choices that, say, would repeat a band within 10 songs.
The article goes on to describe iTunes’ new Smart Shuffle that will randomize music with selected parameters.
Of course, until iTunes can read your mind and determine what you really want to hear next, it’ll never totally satisfy. But they’re getting scarily closer.