“FRANKFURT and LONDON—The crash of an Aviation Week)airlifter that killed four people on May 9 may have been caused by new software that cut off the engine-fuel supply, industry sources have said.” (
Was this a landmark crash? Was this the first airplane crash directly attributable to a software error?
There have been multiple crashes due to shortcomings in human factors design and the related software that can confuse pilots. Uncertainty about so-called “mode awareness” is a leading cause. Crews don’t always recognize what autopilot setting they have chosen or how the fly-by-wire system will behave when they chose a certain mode; for example, one that is supposed to maintain speed or decent rate or altitude.
In 1988, an Airbus A320 flown by test pilots at an airshow flew into trees following a low pass over the runway, in part because their flight control setting did not give them full authority to climb. More recently, the crew of an Asiana Boeing 777 landing in San Francisco realized too late that the mode they had selected for the autopilot caused the autothrottle to cut thrust and slow the aircraft. As a result, decent rate increased and jet got low, hitting a wall short of the runway.
All aircraft manufacturers can do a better job designing flight control software from a human factors standpoint. Pilots should not need to be Elf Lord-caliber gamers to understand how the software-based systems and aircraft will interact.
Combine those design demands with the fact that the latest transport planes are reported to use over 5 million lines of code, and the challenge to create perfect, error-free programming is daunting. The crash of this military cargo plane, at least from the preliminary reports, was not a function of the interaction between crew and automation. A flaw in the software code may have been the root cause. Was that a function of system complexity?