1969 United States meta

The Apollo Guidance Computer and the Birth of Software Engineering

Margaret Hamilton, MIT Instrumentation Lab, 1965–69 — wrote the code that landed humans on the moon and named the discipline that made it possible.

Margaret Hamilton led the Software Engineering Division of the MIT Instrumentation Laboratory during the development of the on-board flight software for the Apollo Guidance Computer. She and her team wrote the asynchronous, priority-driven executive that ran on roughly 64 KB of memory and successfully landed Apollo 11 on the Moon in July 1969 — including the now-famous moment when overload alarms during descent triggered the executive's priority logic to discard lower-priority tasks rather than crash, allowing the landing to proceed. Hamilton coined the term "software engineering" specifically to argue that what she and her team did deserved the same respect as hardware engineering and systems engineering. At the time the term was treated as a joke. The discipline she named is the one this book argues the industry has since forgotten.

In plain terms

In 1965 software was not a profession. It was something mathematicians and physicists did to get the real work done. No one had a degree in it. No one had a name for it. Hamilton was running a team writing the code that would carry three astronauts to the Moon and back, and the code had to work the first time. No patches. No updates. One shot. The Apollo Guidance Computer had about 64 KB of memory. The flight software ran an asynchronous priority-driven executive — meaning multiple tasks shared the CPU, and when the system was overloaded the executive killed low-priority work to keep the high-priority work running. Six minutes before Apollo 11 landed on the Moon, a hardware misconfiguration flooded the computer with extra work and the executive started throwing out the lower-priority tasks. The crew saw "1202" and "1201" alarms. Mission control had to decide, in seconds, whether to abort. They did not abort. The system was doing exactly what Hamilton had designed it to do — shedding load gracefully under overload — and the landing proceeded. Hamilton called what her team was doing "software engineering" because she wanted it to be taken seriously. At the time the senior engineers laughed. By the late 1980s the term was universal. By 2010 it was a degree program in every research university on Earth. But the discipline she named was not just the word. It was a set of practices: write specifications first, prove the program against the specification, design for failure rather than against it, treat the software as the system that has to keep working when everything else does not. This is the discipline Hoare's triples make possible, the one DO-178C codifies for aviation, the one seL4 demonstrates is achievable. It is the discipline that makes software a profession instead of a craft. She named the discipline so it could be respected. We forgot the discipline.