The art of stating the obvious (and still getting it wrong)

It’s a fool that looks for logic in the chambers of the human heart.
Ulysses Everett McGill

weatherman.jpgOne of the reasons why so many people all around the world are scared stiff about the climate change, real or imagined, is that quite obviously, weather matters a lot in our daily lives—even if it’s not about large-scale calamities, hurricanes, droughts or deluges. A simple and relatively modest downpour is enough to ruin a garden party and thus undo the plans for the weekend. Cloudy instead of clear can make all the difference between whether we’d rather spend a day on the beach or in the office. And because for most of us, all those choices take a little planning ahead, we need something to rely on regarding what the weather is going to be like in days and weeks ahead. Luckily, the help is readily available nowadays, as every single TV program and newspaper offers a five-day weather forecast with glimpses even further into the future. If your experience is anything like mine, you probably feel that by and large, those forecasts do their job. There’s an odd miss every now and then—sometimes it rains when it shouldn’t have and sometimes it doesn’t when you were expecting it will, but in general, it is something you can depend on. And frankly, it’s not like you have anything better.

But perhaps you do.

Some guy from northwest state of Missouri with a penchant for statistics and some quite remarkable perseverance decided to find out exactly how well do the weather forecasts actually do when giving us guidance for planning our weather-dependent activities. To that end, he tracked the weather reports of four local Kansas City TV channels plus a government-run meteorological website over the period of 220 days. Although there’s a lot more in the article, probably the most relevant part from the practical point of view is the performance drill-down concerning the chance of rain. As it turns out, when predicting the chance of precipitation for one day out, four stations under review got their stuff right in 85 percent of the time, with the success rate declining to a still respectable 73 percent for a period of the next seven days.

Not perfect, but not too shabby either, one would have to conclude. There are people running some serious models in order to come up with those predictions, but weather is a complex thing and getting it completely right would certainly be too much to ask. Well, that much is true, but there’s a way to do better than that.

And a key to improving the precision of those predictions lies in the observation that rain or dry is not a 50/50 thing. Apparently, in Kansas City it only rains in about 14% of all days. And what this means is that if one would go in the air and predict every night that it will not rain, it would beat the average forecast success rate by 1% for the next day, and a whole 13% for a week out.

The irony of this situation is of course that while demonstrably superior to the current means of prediction, people would probably have serious difficulties taking the “it won’t rain” forecast seriously to the point of organizing their lives around it—which they nonetheless feel perfectly happy to do based on a method, however sophisticated, that clearly delivers inferior results. When faced with uncertainty, we seem to be remarkably reluctant to accept that this is what the things are – uncertain. Instead we have to come up with a method that would try to explain away the unpredictable and project it back on our worlds which then suddenly start making sense to us, even if our predictions actually yield results that are farther off the mark than if we’d simply accept the prevalence of chance. This is the common root of persistent success that, in addition to weather forecasting, things such as astrology columns or stock market reports enjoy. Or, like goes the tagline of a movie that the quote above comes from: “We have a plan, but not a clue”.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s