Wednesday, 6 November 2013

Are you a nanokiller?

I have been intrigued for some time with Ronald A. Howard's idea of micromorts: a way of putting the risks we take in our lives on a human scale. The idea is that one micromort is a 1-in-a-million chance of dying. So, for example, if we say doing a skydive has a risk of 7 micromorts (as that Wikipedia page that I've linked to claims), that means 7 jumps out of each million lead to somebody dying. Or, in other words, if you jump from a plane there's a 7/1,000,000 chance you'll die (assuming you've used a parachute - without the parachute I suspect the chances are far worse). As we'll all die one day, of course, just being alive carries a background risk level of more than 30 micromorts, as the article also explains.

Anyway, I wondered if we might apply a similar principle to road deaths, as a way of making salient a very important point: each time you drive a motor vehicle, there's a small chance someone will die. I've long thought about how, each time I drive, I am effectively killing a tiny fraction of a person because I'm complicit in the overall number of deaths that take place. Today I realised that something analogous to the micromort concept provides a useful way of quantifying this.

So let's find some statistics! The Department for Transport statistics web page reveals that in the United Kingdom in 2012, motor traffic travelled 302.6 bn miles and led to 1754 deaths. Let's do the maths:

  • 302.6 bn / 1754 = 172,519,954.39 miles for each death
  • 1 billion / 172,519,954.39 = 5.8

And so, ladies and gentlemen, I present you with the nanokilling. Every mile you drive, you commit 5.8 nanokillings. Drive 12,000 miles in a year and you've committed 69,600 nanokillings, or 0.0000696 killings.

So clearly, the typical individual is fairly unlikely to kill over the course of their driving career. Let's say someone drives 10,000 miles per year for 50 years. 50 * 10000 * 5.8 = 2,900,000 nanokillings, or 0.0029 killings. This means you'd need to get together with about 344 other people before you could be reasonably sure that, collectively, you've managed to kill somebody.

But that's the thing, isn't it? 345 people isn't really that many. There's probably that many within a few streets of you. And there are a lot of streets in the country, aren't there?

Obviously the nanokilling would need to be recalibrated from time to time as new statistics on numbers of deaths and the amount of travelling that took place to cause them emerge, but of course that's also kind of beside the point. The point is that as long as there is motorized travel and deaths on our roads, the number of nanokillings will never be zero, which means the fundamental point of this article will endure - when we use a motor vehicle, we commit nanokillings. Unless you foreswear motoring (and the products of motoring, and do nothing to push for alternatives) you're to some extent complicit in causing little bits of a death. I know I am, even if I'm not happy about it.


J said...

I assume these calculations are based on road traffic deaths from collisions / injuries. Probably some are not related to motor vehicles but this excludes stroke from hypertension from exposure to traffic noise; cardio-respiratory deaths from air pollution from motor vehicle exhausts; low physical activity levels from the deterrent effect of cars on walking & cycling; and the reduction in social networks (and its effect on increasing morbidity and mortality) form community severance due to motor vehicle volume and/or speed.
So probably don't need anywhere near 354 other people.
sobering thought!

Ian Walker said...

Absolutely, J - my estimate only included direct deaths caused by collisions. If we were to factor in air pollution deaths (of which I've seen estimates of over 20,000 per year) then it's far far worse. Sobering indeed!

Mike said...

Yes I think it is a useful concept in a similar vein to micromorts. It does of course suffer from the same problem though, that it presents a homogenised statistic of the risk on average. However of course that average risk is in fact an integral of lots of different individual risks, with some individuals being below the average and others above. e.g. if you could subcategorise individuals by for example driving style then some drivers are likely to be using more nanokillings per mile than other drivers. But of course most people regard themselves as above average for driving skills, so would probably then think themselves below average for nanokillings!

Anonymous said...

The British Parachute Association quotes a figure of 1/100,000 fatalities/jumps rate for jumps by experienced sky divers generally, an order of magnitude higher, based on figures in their database. Novice freefall jumps are 1:40k. For display skydiving their quoted rate is 1/12500! Their best figure, tandem jumps with highly experienced instructors is still 1:380k. See:

So that'd put parachuting at 2.6 micromorts in the very safest case of a tandem jump, but for people actually doing skydiving for themselves its 25 micromorts while learning, going down to about 10 micromorts when experienced. However, taking part in displays increases it to 125 micromorts!

Interesting comparison :).

dr2chase said...

So, if the background level is 30 micromorts, and if the annual mortality rate for bicycle commuters is 28% lower (results of a biggish Danish study, there are others with similar results listed here: )
does that mean that for bicycle commuters, the background level is really 21.6? I.e., the total risk of death (including poor exercise outcomes) from driving to work is greater than cycling to work plus a daily hop out of an airplane?