The Wiki Man

Why averages don’t add up

12 October 2019

9:00 AM

12 October 2019

9:00 AM

I recently learned from a doctor friend that salt isn’t necessarily bad for you. Yes, there is a minority whose blood pressure isdriven haywire by eating the stuff, but most people can consume it without much risk. The reason we are formally advised to avoid salt is that lowering salt consumption improves public health on average: salt reduction is helpful to the few who are affected, while being generally harmless for everyone else.

This makes sense at first. Except it leads to a problem. Because if you demonise every food that is harmful only to a minority, you risk recommending so many dietary restrictions that life becomes intolerable. A better answer may lie in personalised medicine: ‘In your case, I’d concentrate on avoiding three things first.’

The notion that any solution which is good on average must also be good overall is dangerously plausible, but often wrong. If the government designed pizzas, each would be 25 per cent covered in pineapple, since that is what the average pizza customer wants. In reality nobody wants this at all: they either want the stuff slathered everywhere or else nowhere in sight, believing that the first Hawaiian pizza emerged from the bowels of Satan himself.

The very idea of an average is also an obstacle to identifying and solving problems, because it represents a problem as monolithic when it is not. As I suggested recently, the aim of reducing train overcrowding is silly: if 100 people have to stand 5 per cent of the time it’s not a problem: if five commuters have to stand 100 per cent of the time it is.

Why then do policy-makers obsess about averages? First, most statistics are presented in this form, as a composite snapshot. Second, the interventions government mostly uses — legislation and economic incentives — are often blunt instruments which must be imposed on everyone equally as though we were all representative of the wider population. But we aren’t.

When Lt Gilbert Daniels, a physical anthropologist, was hired by the US Air Force in the early 1950s, he was tasked with designing a cockpit for ‘the average’ man. He already knew from his prior work that an average hand is not a typical hand. He soon found that an average human body, one that is near-average across a range of dimensions, is also surprisingly rare: when you design a cockpit for an average male you are designing for an almost nonexistent body-type. Not one of the 4,000 pilots tested was within the average range on all ten bodily dimensions. Daniels’ solution? Seats and controls which could be moved to suit the individual pilot.

His lesson still needs to be learned today. Stop trying to solve for the average. It’s a principle well understood in the design of websites, where different customers may have very different objectives — browse, buy a specific item, make a return.

One bugbear of mine is that few online retailers give me any choice over who delivers my order. If I am ordering a USB cable, I would willingly pay £1 extra for them to hand it to the Royal Mail, since our postie will be passing by my door in any case. But no, they insist on putting it in a bloody great van and ringing on my doorbell, presumably because that is the best solution on average.

If you want people to feel comfortable with any course of action, leave them some control in the matter. In a future article I plan to explain how the environmental movement could be much more successful if it stopped shouting at everyone to change every-thing all at once, and gave us a list of truly significant things we could adopt one at a time.

Got something to add? Join the discussion and comment below.

You might disagree with half of it, but you’ll enjoy reading all of it. Try your first 10 weeks for just $10

Show comments