What chance really looks like: get your vision here.

I have seen the Loch Ness Monster – and it was on vacation over Bognor Regis.

Or is it a camel?

I know what you’re thinking, that it’s not Nessie herself. But who’s to say it isn’t her spirit?

Cynics suggest this is what statisticians might call noise, and any resemblance to anything meaningful no more than pure chance. They said the same about Jesus on toast. Believers know better. (This week, someone saw Prince playing the guitar in a BBC weather map).

These visions are useful. They’re a lesson that what noise looks like is easily interpreted (by those so inclined) as the real thing. They remind us to check our natural bias to over-interpret observed patterns. Almost like Monte Carlo simulations – they serve as comic illustrations of what goes wrong when we fail to control that impulse. The old point is that despite the fact that they are simply the result of enough chimpanzees with enough typewriters… etc… we’re still tempted to attach meaning to them.

But that old point can’t be made often enough. In fact, I’m thinking of setting up a sideline selling pictures to researchers of Nessie or Jesus on Toast. They could stick them above their desks – a kind of memento mori (a memento forti?) – a measure of the risks we run whenever we seek patterns in data. I used to keep something like that pinned to the wall when I was writing – lest I was tempted to see supporting evidence everywhere for whatever crackpot theory I was hung-up on at the time – a sign made by my daughter that said: ‘oh no it isn’t.’

Do we need telling any of this? Sensitivity to the precocity of chance is a mark of anyone who fancies themselves the least bit statistically savvy. The science replication saga suggests that even sophisticates need the lesson hammering home twenty times a day.

One illustration of the point is here, in a little Excel model I did for a piece for the BBC a few years back (pic below) to show how chance alone can produce the appearance of a killer hospital. On the BBC site, it’s interactive. Try it.


What struck me was that people who know all about chance still found the results surprising: we were able to generate random numbers that intuitively looked like a smoking gun, even to sophisticates: ‘Sure, I know all about variability, but they couldn’t be that bad without a reason…’

Oh yes they could. Which is why – as I repeat ad nauseam – that we should never stop fussing about how spectacularly the world can produce the mere image of meaning. We can’t stop saying this because the opposite tendency likewise never stops – the tendency to see patterns everywhere – and it’s the most dogged, prolific, cognitive impulse of the lot.