Loading...
In 2011, Boeing faced a crisis. Airbus launched the A320neo with fuel-efficient engines, and American Airlines was ready to defect. Rather than design a new plane — a $20 billion, decade-long effort — Boeing chose to bolt larger LEAP-1B engines onto the 50-year-old 737 airframe. The engines sat farther forward and higher, changing the plane's aerodynamics dangerously: the nose pitched up in certain conditions, risking a stall. Boeing's fix was software, not engineering. They created MCAS — the Maneuvering Characteristics Augmentation System — which automatically pushed the nose down based on readings from a single Angle of Attack sensor. The 737 MAX had two AOA sensors, but MCAS only read one. The disagree alert, which warned pilots when the two sensors conflicted, was sold as a $13,000...
Popular framing: Boeing got greedy and the FAA was asleep at the wheel; one bad software patch killed 346 people.
Structural analysis: Competitive pressure from the A320neo forced a path-dependent retrofit onto a 50-year-old airframe, which forced a software band-aid (MCAS) sitting on a single sensor with no redundancy. Captured certification routed Boeing's safety analyses back to Boeing employees; the safety-disagree alert was unbundled into a paid option; normalization of deviance accepted each shortcut as the new baseline. The crashes were the predictable output of principal-agent collapse plus margin-of-safety erosion across regulator, manufacturer, and airline.
The popular frame demands accountability from individuals (executives, specific engineers) and modest regulatory reform (more FAA oversight). The structural frame suggests that without changing the incentive architecture — how executives are compensated, how regulators are funded and insulated, how self-certification authority is bounded — replacing the actors changes nothing. The gap matters because the post-crash settlements and reforms were calibrated to the popular frame, leaving the structural conditions largely intact, as the 2024 door-plug incident suggests.
