Monday, 15 February 2010

Two tribes

So I've gone and read the report that 54% of people are talking about! Sorry, I mean 0.54%. In short, it is uninformative rubbish. Yes, there are lots of number about how much worse the worst areas in the country are compared to the best areas in the country, but, well, so what? Sure, life is going to suck in the worst places and be awesome in the best places. Here are some examples from page 7:

• Children are six times more likely to grow up in a household dependent on out-of-work benefits.
• Young people are almost twice as likely to be expelled from school.• Pupils are half as likely to get three ‘A’ grades at A-level as their peers in the least deprived local authorities.
• Teenage girls under 18 are almost three times more likely to become pregnant.
• Young people aged 18-24 are twice as likely to suffer from youth unemployment.
• People are twice as likely to be on a waiting list for social housing.• Households are four times more likely to be homeless.
• People are almost three times as likely to be a victim of violent crime.
• People are twice as likely to be a victim of common assault.
• People are thirteen times more likely to be a victim of robbery.
• People are twice as likely to have no qualifications at all.

Of course, this tells us absolutely nothing about what life is like for the vast majority of people in the country. It tells only us about outliers, which are definitionally rare and at the end of the spectrum. This is pure availability bias: tell us about the best and the worst because that's what sticks out. This is why people are frightened of terrorists, too: terrorist events (in the Western world at least) are so vanishingly rare that each one gets so much news coverage we feel they are far more likely than they are.

If the Tories had identified a line at, say, 40% median "area depravity" (they love the word "deprived"!) and shown that things were on average significantly worse below that one line than above it, perhaps their thesis would have some merit.

Here's another little vignette of the deception this report tries to pull, from page 2:

Since 1997 the Government has spent £473 billion on welfare payments alone – that’s as big as our whole economy in 1988.

Typical political trick: is your scary number not big enough? Then add it up over lots of years, then compare it to a number from a single year!

£473bn is approx £40bn per year, which, according to the Tories at least, is only 8.3% of government spending in 1988. But why did they choose 1988? £40bn is only 4.6% of 1997's GDP, and a measly 2.3% of 2008's GDP. Do we really resent paying 2-4% of GDP on welfare payments? Really?

We are not "two nations". We are one nation. And we're getting pretty sick of transparent political bullshit.

Friday, 5 February 2010

Brilliantly, the UK Statistics Agency has pwned Chris Grayling for, basically, lying about crime figures. Mark Easton's report here.

I'm surprised (well, perhaps just disheartened) that this isn't getting more play in the national media: here is what could be the future Home Secretary lying through his teeth about crime figures that he full-well knows are incomparable for political advantage.

Sure, this thing is probably all too common, but I still think it warrants flagging up when it happens.

Thursday, 4 February 2010


Here's a thought: everyone is arguing about which is better: AV, AV+, FPTP or STV. Or at least, lefties are - clearly STV is best! - but instead of gambling five years of potentially ruinous governance on a proposition with no local evidence, why don't we run some trials, eh? Take ten random constituencies and have them vote AV, another ten, AV+, another ten STV. See what happens. Take some time, get some evidence.

Is that really such a crazy idea?

Tuesday, 2 February 2010


So I am investigating building a 3d game in XNA. So far, my main conclusion is that all of the available "engines" are rubbish. Either poorly-documented, feature-lacking, or both. I am used to the relative elegance and acres of documentation of Java3d's Scenegraph, and despite the fact that J3d is lacking a ton of modern features (like, er, shaders), I have actually managed to get something approximately like what I want up and running in it.

This is quite frustrating. In the unlikely event that you have any tips, please contribute.