Steven Levy on the iPod shuffle (Guardian article). Scientists have know for a long time that the folk perception of randomness and statistics is skew in many different dimensions of wrong, but it takes the bloody iPod to get this fact into mainstream culture.

(Queries of pseudorandomness aside, smartytrousers) I still like the idea of the iTunes/iPod software engineers meeting up, and “So, this random facility… should we just use a rand() function?” “OO, better than that, we should construct an elaborate AI algorithm based on the users preferences and a fourier analysis of the current song?” “Brilliant – we’ll get paid even more by the hapless ipod fools ha ha ha ha ha ha” etc

Separately, this might explain why TV’s Doctor Who fits a “randomiser” to his Tardis – the ultimate anti-satnav – and first ends up on the home world of his enemies, and then Earth. This is perfectly random, and it just *seems* to be really unlikely.

I like the idea that the TARDIS actually hates Who and therefore deliberately ends up at planets where he will either get killed, or at least captured for long enough that the Tardis can recycle all the stale air and the the smell of a man WHO ALWAYS WEARS THE SAME THING out of his system.

A Rand() function is still not strictly random – but can be easily be made so by adding in a non-computer controlled external operator (like, say, the user).

(Queries of pseudorandomness aside, smartytrousers)

I perhaps would prefer to tackle pseudorandomness, and perhaps that human minds, despite the alarming complexity of the pseudorandom algorythmn (or just picking a bit of pi) the human brain can sense something is up…

TARDIS HAETS Doctor but is sitting in a tree k.i.s.s.i.n.g. with Urf.

This is the usual failure to understand randomness, of course – all the patterns he claims to have found in his and others’ iPods are different things with few figures attached. It’s like claiming that getting three sevens in the first poker hand you deal proves the cards are non-random, especially as your pal was playing the other day and got a royal flush.

isn’t there some result that ANY finite string of figures is patterned rather than random — ie that “true” randomness can only obtain in a countably infinite string?

(i may have totally misremembered this — it is a long time since i studied the higher reaches of set and number theor, since when my BRANE has turned to CANTOR’S DUST)

No, you are right Mark, and more importantly you can prove that any generatable countably infinite string will fit a non-random pattern. A little trick with an analog of the incompleteness theorem it is easy to show that this does not cover all the coutably infinite strings, and hence the ones which are left are truely random BUT impossible to generate…

Hence Alan deliberate closing the door on pseudo-randomness…

best comments evah!