On the way home today I was reminded, and very strongly, of an article in the weekend's paper about the randomness of an iPod's shuffle.

Combining, as it does, discussion on high-tech gadgets, music and mathematics, I was particularly drawn towards that article which was about how some iPod owners get suspicious of the random shuffle feature on their toys because they seem to keep playing tracks by the same artists all the time.

The trouble with random is that it is, as the name implies, random. The one thing it does not mean is 'even distribution'. Almost by definition, with any truly random distribution you will encounter clusters – most famously with incidences of cancer. Tossing a coin 10 times and getting heads every time is just as random as getting alternate heads and tails.

The rational, scientific part of me knows that perfectly well, so when I find clusters of music by the same artist or the same genre on my iRiver's random play I resist the temptation to  write to the Koreans and complain about their algorithms.  Even so, I was impressed when the iRiver played Manu Chao's song Mr Bobby and then segued straight into Patchanko Hop by Les Fils de Teuhpa – a track from a Mano Negra tribute album. Mano Negra being Manu Chao's old band.

So somehow, the iRiver followed a Manu Chao song from 2001, with an old song of his old band's from the early 90's, covered by another band 5 years ago. Brilliant. In between this, the machine also played 3 tracks from the Pet Shop Boys' Bilingual album during a sequence of not more than 20 songs. Remarkable when you consider that there are between 9,000 and 10,000 tracks to choose from on the machine's disk.

You can almost see how religions get started.  The natural inclination to associate random with evenly distributed means that if a pattern occurs it leads to suspicion of some hidden source of direction; some external influence.  Believing that Apple (or iRiver)  have written some sort of artificial intelligence which can discern your mood or select songs based on a qualitative assessment of the content or knowledge of provenance not contained in the filename or path is not really so different to attributing unusual coincidences or spectacular natural phenomenon to a supernatural cause is it?

Indeed, go onto the iRiver forums and  you will find there are periodic holy wars between those who insist the shuffle is not random and those who say it is, or that two songs by the same artists are not proof that it isn't.