Once upon a time, there was a world war, and many planes were shot down. The King was very upset about this and asked his wizards, “How can we make sure our planes are safe, and do not get shot down?”
The wizards decided to study all the planes that had been shot. They recorded every shot and every detail about each shot. Then they worked for a long time, scribbling formulas and determining how to fix the planes so they would not be shot down.
But when they brought their results to the King, the court jester laughed at their wisdom.
“You’ve studied the wrong planes!” he said. “These planes were shot, but made it home safely! The planes you needed lie at the bottom of the English channel.”
But it was too late. The wizards had spent the King’s fortune fortifying planes in places where they could already be shot without bringing down the plane.
Moral of the story: Beware an irrational trust in data analysis, warns Ogilvy & Mather vice chairman Rory Sutherland, who shared this tale of Big Data folly from World War II. It can lead to an attitude of “false certainty,” where you trust what the data says without question.
“…I see big data as being a rather terrifying battleground which will be seized upon by people with pretty good statistical understanding, but who will tend to make perfect models out of things using various complex mathematical tools, which may not only be inappropriate to the task at hand but based on data that isn’t even particularly relevant,” Sutherland said at Digital Shoreditch’s Nudgestock event.
In this case, the problem was that the agency made incorrect assumptions that lead to collecting the wrong data. After studying the planes with multiple bullet holes, they spent “serious sums of money” reinforcing those areas with plating, only to realize what’s obvious to us now: The planes survived those shots in the first place, so there was no need to reinforce those particular spots.
Wrong assumptions are just one of the problems that can lead to faulty decisions. Data quality problems and data that’s been falsified can also lead to problems, as we’ll see in tomorrow’s fable, The Data Wizard Who Counted Bodies.
Be sure to also read Big Data Fables: The NSA and the Metadata.