Have you ever wondered what insurance companies do with all the money you pay them every month? It seems that one UK-based insurance company decided that a good way to use up some of the spare cash was to discover that, every day, people in the UK are carrying around over 2,000 tons of redundant keys. I'm surprised they didn't come up with some conclusion such as this requires the unwarranted consumption of 10,000 gallons of fuel, which emits enough carbon to flood a small Pacific island.
It seems they questioned several hundred people about the number of keys they carry and which locks they fit, and the results indicate that everyone carries around two keys that they don't know what they are for. However, I did a quick survey amongst six family members and friends and discovered only a single unknown key amongst all of us. So, on average, we are only carrying one sixth of an unknown key around. Extrapolating this percentage across the country means that there must be several hundred people carrying bunches of keys around when they don't know what any of them are for.
Of course, statisticians will tell you that you can't just average out results in this way - it's not mathematically logical. It's like saying that, because some cats have no tail, the average cat has 0.9 tails. Yet insurance companies rely on averages every day to calculate their charge rates. They use the figures for the historical average number of accident claims based on your age and driving record, or the average number of claims for flooding for houses in your street. Or your previous record of claims for accidental damage to your house contents.
What's worrying, however, is how these numbers affect your premiums. I just changed the address for my son's car insurance when he moved a quarter of a mile (to the next street in the same town) and the premium came down by nearly a quarter. I've suggested that he move house every month so, by next year, we won't be paying anything. Though it probably doesn't work like that...
Anyway, if the way they calculate premiums is already this accurate, just think what it will be like in a few years' time as more powerful processors, parallel algorithms, and quantum computing continue to improve the prediction accuracy. The inevitable result is that, when you apply for an insurance quote, the company will actually be able to tell exactly how much they will need to pay out during the term, and will charge you that plus a handsome profit margin. So you'll actually be better off not having insurance at all, and just paying the costs of the damage directly!
And this is the interesting point. Insurance is supposed to be about "shared risk". When insurance first started after the severe fires in London in the 17th Century, the premiums were based on the value of the property and the type of construction. Wood houses cost twice as much to insure as stone or brick houses. Other than that, everyone paid the same so they shared the costs equally. Of course, you can't really argue with the concept that people who have lower risk should pay less, or that people whose property is more valuable (and will therefore cost more to replace) should pay more. But I wonder if we are starting to take this to extremes.
Ah, you say, but even with the pinpoint accuracy of the current predictions of risk, they are still averages. So if you are lucky (or careful) you can beat the odds and not need to claim, while if you are unlucky (or careless) you will get back more than you paid in premiums. True, but next year they'll just use an even more powerful computer to recalculate the risk averages. Like a derivative function in calculus, the area of variability under the curve can only get smaller.
But I suppose it will be useful in one respect. When you get your motor insurance renewal telling you that this summer you will collide with a 2004 registered blue Volkswagen Beetle at a traffic signal in Little Worthing by the Sea, you can simply cancel your policy and use the money you save to go on holiday to Spain instead.