The title is a quote of Michael Gove — to be fair on him I think he was referring to economists – a profession whose track record of consistently making inaccurate predictions is notorious. But still, as a PhD educated scientist, the quote does rankle a bit. Although that may be just my natural reaction to Michael Gove – he could read the phone book and I’d still get grumpy.
But the quote is memorable, and I thought about it while reading Chaos Monkeys by Antonio Garcia Martinez. The book, which I really enjoyed, is about Garcia Martinez’s time working in Silicon Valley – the part of California south of San Francisco where Google, Facebook, Apple, WhatsApp, etc are. Many British people may have had enough of experts but Google, Facebook, Apple, etc have not. They’re hiring them for six-figure salaries plus stock options, and putting them to work.
Some of these experts are busy experimenting on us, using the standard scientific method that my colleagues and I teach. They do this because the scientific method is the best way to increase their profits. For example, 20 minutes ago I went on the Amazon website and looked at some laptops, I then went to The Guardian website, which displayed an ad for laptops. This is not coincidence.
Very well paid experts are writing code that ran on ad servers when my computer requested The Guardian homepage. As well as the code that delivered the news items to my browser, separate code effectively sold ad space on the webpage to the highest bidder – in this case to Microsoft who want to sell me a Surface Pro. I find this quite impressive, in the time it took for the webpage to download, a number of companies, including Microsoft, bid automatically for screen space on my laptop, and Microsoft won.
The algorithms that run this auctioning and ad display are optimised via the good old scientific method of testing a hypothesis using experiments with controls. The experts at Google etc develop a new algorithm that they hypothesise is better, i.e., generates more clicks and so more money for Google. This is a testable hypothesis, so they test it. They partition some very large number of web users (i.e., us) into two groups, one group gets the new algorithm and the other the old. Then within a couple of days they can just tot up the ad revenue for the two groups, if the group with the new algorithm generated more money they can roll out the new algorithm, but if the ad income from control group is higher, they won’t.
A clear example of the scientific method at work, and a demonstration of how organisations that want to get the job done and make money have not had enough of at least some types of experts.