Thursday, 9 September 2010

Three of the best articles I've read all year

This week I've been smacked sideways three times by three of the most interesting articles I've read all year. I just had to link to them. Warning: moderate levels of nerdiness.

Firstly this one courtesy of Wired: quantum chess. Wowzer. Normal chess is hard enough for me with enough permutations to keep me interested for ages (at our recent cheese-and-port-and-chess night, the games were always the centre of attention despite some pretty stupendous smelly cheeses), and I've never got into Fischer Random chess, but quantum chess sounds nuts. And impossible.

Then this story from the LA Times, which was picked up on by the Freakonomics crew. That sort of shit gets me seriously excited. Mapping crimes isn't new - and police forces are already aware, for example, that serial burglars tend to work within a very small area - but using simulations to predict the behaviour and movement of criminals while including factors like police movements into the model, adds up to a very powerful model. "Humans are not as random as we think". That sounds familiar! Will it work? Would it ever hit the UK? Who knows, but even the thought cruising around in an IRV with real-time probability heatmaps for crime on the MDT gets me fidgety with anticipation. Look out for further developments.

As an aside, another, similar story that Freakonomics reported was this one on using text analytics to predict whether a film will succeed or bomb. Instinctively, that sounds like bollocks to me; some of the best films have the simplest plot lines, and vice versa. Isn't there some theory about there only being five stories ever written, and they're all by Shakespeare, or something like that? Stories about using science and/or data modelling to predict wacky things tickle me every time, but this sounds like nonsense to me. I'm happy to be proved wrong.

Finally, I've always believed in some similarities between physics and marketing (I've lost count of the number of people ask why I wanted to work in marketing after doing a physics degree, but it makes complete sense to me). What a delight, then, to discover a talk given by Google's Dan Cobley comparing the laws of physics and those of marketing. ("H/T" as they say to Mike Cooke). It'll only take up seven minutes of your time. He splits things up neatly into four analogies - all of which are neat: large brands have larger inertia and are therefore more difficult to change direction quickly (cf Newton II - nice!); and that as time goes on, it becomes ever more difficult to keep control of a brand (cf entropy) - he didn't explicitly make reference to the fact we are living in a more fragmented society, but I wonder if that was what he was referring to. But the real killer was Heisenberg's Uncertainty Principle, and the observer effect that results from it. It's a simple idea (not so simple in practice) and many philosophical dilemmas result from it: that any observation you make actually affects the quantity being measured. As Cobley points out, the fact that consumer research can be affected by the very act of observing (consumers are less likely to give honest answers in the presence of an observer), can be compared obviously and simply to the Heisnberg-derived observer effect. Yes, yes, YES. I'd half had the same comparison stored up in my head but Dan Cobley has articulated is concisely. Some people in the comments whine about it not taking the true spirit of the Uncertainty Principle and how that's not the meaning of the observer effect, but that's just bluster and semantics - for all real-world, simple-analogy purposes, it holds true. Those moaners would probably have a fit if they saw my own analogy of flirting with quantum mechanics which I'm pleased to say has the seal of approval from my old flatmate Libby, a postdoctoral physicist who said she'd use the comparison with her undergrads!

Oh, and while we're on a science tip, a good piece by Evan Harris on the impending cuts to the science budgets.

No comments:

Post a comment