The Weatherman is Not a Moron.

Snippet 1:

Our views about predictability are inherently flawed. Take something that is often seen as the epitome of randomness, like a coin toss. While it may at first appear that there’s no way to tell whether a coin is going to come up heads or tails, a group of mathematicians at Stanford is able to predict the outcome virtually 100 percent of the time, provided that they use a special machine to flip it. The machine does not cheat — it flips the coin the exact same way (the same height, with the same strength and torque) over and over again — and the coin is fair. Under those conditions, there is no randomness at all.

Snippet 2:

For centuries, meteorologists relied on statistical tables based on historical averages — it rains about 45 percent of the time in London in March, for instance — to predict the weather. But these statistics are useless on a day-to-day level. Jan. 12, 1888, was a relatively warm day on the Great Plains until the temperature dropped almost 30 degrees in a matter of hours and a blinding snowstorm hit. More than a hundred children died of hypothermia on their way home from school that day. Knowing the average temperature for a January day in Topeka wouldn’t have helped much in a case like that.