Analyzing Air Force Bombers

I read an interesting story recently about risk management in a book titled Red Blooded Risk by Aaron Brown.

During World War II, the Air Force was trying to figure out the optimal amount of armor that should be added to it’s bombers.  Since every pound of armor means one few pound of bombs, it means that more bombing runs will be needed to deliver the same payload.

The Air Force collected data about what parts of the bombers were hit the most – leading edges took more damage than trailing ages, for example, thinking that the places that took the most damage needed the most armor.

They gave this information to a statistician named Abraham Wald and his response was surprising: he recommended that they put armor where damage had not been recorded and no armor on the parts of the plane that had recorded the most damage.

When the Air Force asked him why he had put armor in places that had never suffered any damage, he said, “the bombers hit in these places never come back.”

The story is meant to illustrate that you have to think carefully about problems in a sometimes out-of-the box kind of way.  While the Air Force was right to consider where previous bombers had been hit, that evidence alone wasn’t sufficient to make optimal decisions.