Because computers aren’t perfect

Last week’s Air France crash off the coast of Brazil will be puzzled over for a long time and it seems unlikely that the plane’s “black box” recorder will ever be recovered.

The entire tragedy reminded me that I don’t plan ever again to undertake a flight of more than a couple of hours and that even if I were to consider it, I would find long hours over water very unnerving. I used to love flying. Even dreamed of being a pilot when I was growing up — only to run into the “girls can’t be pilots, they can only be stewardesses” thing.

But I digress. My son sent me a link to an Information Week article about the Air France Airbus. I’d read elsewhere about its fly-by-wire technology maybe contributing to the crash, but this particular story explains how America’s take on flying by computer differs from France’s.

And I’ll take America’s take any day. Computers are no more infallible than the humans who program them. If a human doesn’t anticipate that conditions b, g, k, y, and z might all occur at once and include that possibility in the program, the computer won’t know what to do if it happens. That’s why we always need a manual override, a way for a human operator to step in and take control when a computer can’t get the job done.

We’ll probably never know if the Air France pilots, given a manual override, could have saved that plane. But I know I’d never have taken that chance away from them.

Sully Sullenberger evaluated a lot of improbable conditions very quickly, drew upon his many years of experience, and safely landed his plane in the Hudson River. The two situations are very different, but I’ll bet not one of the passengers on that NY flight would have preferred a computer to Capt. Sullenberger that day.