A fun little story about “being data driven”​

I’m seeing outstanding amounts of hype around “data”: “being data driven”, “big data”, “data marketing”, “a data based marketing strategy” and so on…

And, in every case, data intended as huge, convoluted amasses of numbers – dutiful observation, since “data” by itself does not assumes a peculiar definition of its form.

I think it’s worth sharing a nice story about “being data driven”.

The Lancaster bomber “data driven” armor system

During World War 2, the Lancaster bomber was England’s main bomb delivery system. And, obviously, the Nazi forces had great interest in shooting them down before they would rain tons of hate and discontent upon their roofs.

British forces gathered to improve their bombers’ survivability, and choose to use a data driven approach to find the best solution: they analyzed all the surviving bombers, so to find out which parts have been hit the most.

They literally counted how many bullet holes the surviving bombers had, and made an average of the most hit sections. And then, obviously, proceeded to add more armor where the most of bullet holes are.

First part of the experiment: more armor where the most of the bullet holes are

A fleet of modified Lancaster bombers took off for new sorties.

Outcome? Same as before. Adding armor in the bomber’s most hitted parts didn’t improve their survivability.

Then, one voice (belonging to an Hungarian scientist: Abraham Wald) arose from the team:

“You should not improve armor on the sections who could withstand the more damage and still get back home: you should improve armor on the sections who were brought unscathed home – because that means that the bombers who got hit in there, never got home.”

Who, for the record, were mostly the engines.

Second part of the experiment: more armor where the most of the bullet holes aren’t

Outcome: success.

Abraham’s suggestion to arm more the zones with less bullet holes improved the Lancaster bomber’s survivability.

Moral of the story:

Data is not understanding

Data is information

Information is not understanding.

Information can lead to understanding, and therefore a possibility of finding a solution, only when elaborated by a brilliant enough mind.

Further readings:

Abraham’s Wiki page: https://en.wikipedia.org/wiki/Abraham_Wald

A longer read about this story: http://paristampablog.com/2014/09/25/abraham-wald-and-the-missing-bullet-holes/

A fallacy very similar to this situation: https://en.wikipedia.org/wiki/Survivorship_bias