By Daniel Stern, DVN Chief Editor
Two weeks ago a jet plane just two years old crashed in Ethiopia six minutes after takeoff. Last October a jet plane just two months old crashed into the Java Sea. There were no survivors either time. Both crashes involved the same make and model of plane: the Boeing 737 Max 8, a state-of-the-art version of the well-proven 737. As investigation progresses, it’s looking more and more as though both crashes were caused by an advanced pilot-assistant system, a new automation called MCAS, for Manœuvring Characteristics Augmentation System, intended to prevent aerodynamic stall (when the wings no longer generate lift) during low-speed, nose-up flight. MCAS uses sensor data input to detect when that kind of condition has developed. The main input is from an angle-of-attack sensor (AOA)—just one of the plane’s two AOAs, for some reason—which is supposed to tell the computer about the plane’s nose angle relative to the air rushing past the plane. Then as output, MCAS pushes the plane’s nose down to compensate. Boeing put in the system to counter the Max 8’s tendency to nose-up more than usual.
Both doomed planes zoomed wildly up and down in the minutes and moments just before crashing, and now it’s becoming clear the up-down zigzagging was on account of the pilots’ tug-of-war against the MCAS output. MCAS got confused, falsely determining the plane was nose-up and nearing a dangerous aerodynamic stall, and aggressively tried to “correct” the nonexistent condition by forcing the nose down. Every time the pilots manually pulled out of the resultant dive, the recovery lasted only ten seconds before this pilot assistant, this safety system, would once again point the plane down. The pilots fought heroically, but lost their battles with a computer programmed and empowered to act as though it knows best.