This week I have been helping some friends insulate their house. It is an old, solid-walled, terraced house, and we are aiming to achieve energy performance not too far from that of a modern house by fitting insulated panels to the inside of the outside walls. The construction is basically 50mm of rigid, foil-backed insulation board plus a layer of normal plasterboard, ready for redecorating.
Now, the standard size of an insulation board or plasterboard sheet is 2.4m by 1.2m, and up until today (when these materials arrive) my main concern had been getting these sheets up the narrow stairs. However, today is very, very windy, and now I'm much more worried about actually getting them into the house in the first place. This recalls Venturi's Law*, that windspeed is directly proportion to the surface area of the material you are attempting to carry. Practical applications include design of racing yachts, and not choosing to transport sheet building materials on windy days.
I have lived in traditionally windy cities - Sheffield and Newcastle - for most of the last 20 years, and I know that even in such cities there are probably only about 10 or 15 days per year when it is genuinely, seriously windy. That means on any given day there is about a 1 in 24 chance of high wind. I would guesstimate that I carry sheet materials outdoors only about 4 times a year, so on any day there is a 1 in 91.25 chance that I'll be carrying a sheet of plasterboard across the street. Therefore, if I stuck a pin in the calendar on, say, 4th February, the chances that I'd find myself carrying plasterboard in high winds on that day would be 1 in 2,190. That would mean only one day in every six years would I encounter this unfortunate coincidence. However, anecdotal evidence contradicts this, since I'm pretty sure it's windy almost every time I try to carry something.
I'm not paranoid and I don't believe I'm being singled out. I first noticed the phenomenon when carrying architectural models from home to university in Newcastle, where the flimsy polystyrene versions of my designs were always destined for some destructive weather-testing before a tutorial. Since moving to Sheffield I have often noticed other architecture students putting their models through the same testing processes.
This got me to thinking: supposing I'm not a victim of Venturi's Law, but of something altogether more insidious: unfortunate unlikelihood. That's to say, the less likely it is that I'm doing something, the more likely it is that something unlikely will happen at that moment. The evidence for the existence of this phenomenon is strong. If you use a train every day, it will occasionally be late; if you use it once a year, it is almost bound to be late that day. If you cycle every day, you will occasionally get wet; if you cycle once a month, you are certain to get caught in a monsoon. If you eat oysters every day, you may get a bad one and be sick once every couple of years; if you eat oysters only on Christmas Eve, you will inevitably spend Christmas Day feeling very poorly and vowing never to eat oysters again. If you drink alcohol every day you might have five nasty hangovers during the course of a year, and maybe one evening to be embarrassed about; if you only drink on bank holidays, you'll still have five nasty hangovers per year, and at least three causes to be ashamed of yourself.
The question is, is it possible to harness this force of nature, this convergence of statistical fragility with anecdotal certainty, and turn into something positive and useful? Does fortunate unlikelihood also work? For example, let's say I currently spend 1 day in every 10 days looking for new business, and actually win a commission once every 90 days. Now supposing I only spent 1 day in 90 looking for new business, is it more or less likely that my marketing efforts would pay off? Or, if my friend Filbert asked 10 women to marry him, would he be more or less likely to get a positive response than if he only asked one woman?
I suspect the only way to answer these questions is by empirical experimentation, but there is certainly a direct application to research methods. A good example is the kind of questionnaires used in transport or retail research, which begin, "How often do you use this service/shop, all the time, often, or rarely?" Conventionally, the subsequent answers of frequent customers will be considered more significant, because their repeated patronage makes them more valuable customers. However, the law of unlikelihood dictates that the customers who use the service/shop very rarely are much more likely to have 'outlier' experiences - good or bad - and therefore focusing the research on those occasion patrons will enable more detailed study of how, and why, things might go spectacularly right or wrong for any given patron on any given visit.
A closing conundrum: my grandmother always used to speed up the arrival of a bus by lighting a cigarette. She probably smoked about 10 cigarettes a day, and used the principle that a bus was more likely to arrive if it was interrupting a smoke by doing so. It often worked, but not always. Given that I've smoked fewer than 10 cigarettes in 37 years, it is almost infinitely unlikely that I would light up whilst waiting for a bus, so I couldn't actually use a cigarette to summon a bus without destroying the unlikelihood of my doing so. On the other hand, a chain smoker would always be smoking when the bus arrived, but his decision to smoke would have no effect on the bus's arrival. Therefore, does the greatest overall power lie in actions that are only moderately unlikely?
AW.
*Note: Venturi's Law is different from, though not wholly unrelated to, the Venturi Effect. The main difference is that the latter actually exists.
