Wednesday, April 08, 2015

Little and dangerous

It’s really difficult for people to get their heads around stupidly large or absurdly tiny quantities. “I worked it out in nanoseconds” is a colloquialism that doesn’t really encompass quite how tiny a nano-anything is. For what it’s worth, there are around pi seconds in a nanocentury. So we end up using prefixes to produce sensible numbers. Seconds; not nanocenturies. You don’t go into Tiger Treats and ask for 0.0001 tonnes of liquorice unless you’re deliberately trying to be annoying; you ask for 100 grammes, or 4oz, or “a quarter” if you’re still Church of England.

This is not a metrication rant. It’s about the measurement of small but fatal risks.

Typically, for every one million scuba dives there are five fatalities. (75 deaths in 14 million dives, according to BS-AC in 1998-2009). Probability of dying: 0.000005

In nearly 4.9 million skydives over 1994 to 2013 there were 41 deaths: around 0.000009 chance of not surviving a jump.

All those leading zeroes are difficult, hence the concept of the micromort. One μmt = 1/1,000,000 chance of death. Thus:-
  • One BS-AC scuba dive 5 μmt 
  • One parachute jump 9 μmt 
  • Climbing Mount Everest 39,427 μmt
It’s relatively easy to collate micromort data for regulated activities such as diving, or parachute jumping, or flying as a fare-paying passenger. You simply add up the total number of deaths and divide by the total number of customers. For air travel, I found an ICAO safety report for 2009-2013 that says the average risk was 16.6 μmt/departure.

I found a separate figure of 22 μmt/departure elsewhere on the internet. This matches ICAO’s 2009 figures. It’s possible to calculate a risk in terms of micromorts per kilometre travelled, but as no-one has ever yet collided with the sky and most aviation incidents inevitably involve terrain, counting takeoffs is probably more meaningful. Because air travel generally involves vast distances, comparing it with other modes of transport on a risk per kilometre basis probably isn’t fair.

Ignoring airport transfers (the thrill-ride fun bus between Manila and Batangas springs to mind), if you fly to some exotic dive destination and do fewer than seven dives, the flights are more hazardous than the diving. So dive more. Three or four dives a day seems about right.

Problems with micromorts arise when attempting to assess risks associated with unregulated activities such as riding a motorbike. At last, we arrive at what I want to write about.

No amount of dressing up the figures will ever show that motorcycling is the safest way to travel. The rider is projected at speeds well in excess of what the human body was ever designed for, with almost nothing in the way of protection beyond a plastic pot on his head and some animal skins wrapped around his body and limbs. Hardly the same level of protection that is offered by being strapped inside a tonne or more of metal that is designed to crumple and absorb energy in a crash. I’m not going to attempt to show that bikes are as safe as cars; as the figures I’ve obtained from the University of Google offer extremely variable results, I’ll not discuss the subject of the private car and its relationship with the micromort.

Fortunately for me, the Google Elves have provided more coherent data with regard to motorcycles and micromorts.

US figures for 1999-2003 rate motorcycling risk at 5.37 μmt/trip. (3112 fatal injuries in 580 million person trips). Compare this with UK figures: over 2007 to 2013, the average annual number of motorcycle deaths was 460 from a fleet of some 1.1 million motorbikes. That’s 418 μmt/year.

In 2009, the NHTSA rate was 723μmt per US-registered motorbike. That’s 1.7 times the comparable UK figure. Perhaps there is something in mandatory staged licensing and compulsory helmets. More likely is that there’s insufficient data here to formulate any theories.

The US figures include a ‘per hundred million person trips’, and to offer any degree of confidence, the number of trips has to be accurate. How is this achieved? What is a ‘trip’? Is Chicago to Los Angeles one trip, or is each day of the Route 66 tour a separate trip? Commuting to and from work is two trips a day, right? There’s clearly an extremely wide variation in annual mileage too. You have daily bikers who clock hundreds of miles per week come rain or shine, and others who only take their bikes out to shine them and trailer them to Sturgis once a year. On the US-based Concours Owners Group forum, there are members who claim figures like 100,000 miles in three years, or “My 2009 Connie has 185,000 miles and has never missed a beat…” That’s 53,000km/year and 59,000km/year. And another selling a mint 2010 with only 15,000 miles, never seen rain, full service history, etc… (6,000km/year).

How can anyone possibly estimate a single risk based on trip numbers or distance travelled?

The UK figures produce their own set of difficulties. They fail to take into account trip numbers or distance travelled at all. Some owners have more than one machine, and can obviously only ride one at a time. It is theoretically possible by reductio ad absurdum, to die in a bike crash in the UK while your road-legal machine is parked in the garage and you’re at home watching the telly.

My personal motorcycle experience amounts to 385,000km over 35 years; what I regard as a moderate 11,000km/year. According to the University of Internet’s Faculty of Wikipedia, there’s a 2009 article in The Times newspaper that includes risk of mortality as “1 μmt = 6 miles on a motorbike,” or 0.1 μmt/km. That’s 1100 μmt/year with my annual mileage.

Adopting the US figure of 5.37 μmt/trip, the arithmetic throws out 205 trips per year and a typical trip of 54km. Both of these figures seem broadly in line with my own behaviour. Most of my motorcycle trips are commutes to and from the office, and regrettably few are epic tours.
  • UK figures: 460/1.1M = 418 μmt/year 
  • US figures: 205 x 5.37 = 1,100 μmt/year
I could get my personal risk figures down to the UK values if I did 4200km/year. Does the Average Motorcyclist really cover that little distance?

High mileage implies more experience and therefore lower risk, right? If you have a record of riding tens of thousands of miles without falling off or being hit by another vehicle, you’re obviously doing something right.

Conversely, you can argue that low mileage reduces your exposure to a given risk of riding an unstable machine among other vehicles being piloted by idiots who are too busy updating their Facebooks to look out of their windows... The truth is probably somewhere between the two. There's a low-risk sweet spot where the emptying bucket of Luck and the filling bucket of Experience together produce a minimum overall Risk.

As none of this blog post actually comes to any coherent conclusion, please feel free to cherry-pick the parts that best suit the case that you’re trying to argue.


No comments:


The opinions expressed in this weblog are the works of the Grumpy Goat, and are not necessarily the opinions shared by any person or organisation who may be referenced. Come to that, the opinions may not even be those of the Grumpy Goat, who could just be playing Devil's Advocate. Some posts may be of parody or satyrical [sic] nature. Nothing herein should be taken too seriously. The Grumpy Goat would prefer that offensive language or opinions not be posted in the comments. Offensive comments may be subject to deletion at the Grumpy Goat's sole discretion. The Grumpy Goat is not responsible for the content of other blogs or websites that are linked from this weblog. No goats were harmed in the making of this blog. Any resemblance to individuals or organisations mentioned herein and those that actually exist may or may not be intentional. May contain nuts.