Digital Apollo: Human and Machine in Spaceflight Book

As Tom Wolfe tells in much detail in The Right Stuff, early American astronauts were test pilots who wanted to fly their spacecraft, not just passively sit inside and let ground control or onboard computers fly them. Note that the computers of the time were very primitive by the standard of today, and could be downright dangerous: in one of the flights of the North American X-15 suborbital spaceplane, an adaptive autopilot amplified pilot error and caused the plane to break apart, killing the pilot. Apollo missions flew through a combination of onboard computer control and semi-manual control: even when the flight commander took control of the spacecraft, he did not physically move the nozzles; the computer was still the intermediary.

The most interesting part of the book is about the design of the Apollo Guidance Computer, of which both the command module and the lunar excursion module had a copy. It was one of the first computers built out of integrated circuits, each with 3 NOR gates. The ROM storing its software was physically in the form of wires woven through ferrite cores by retired seamstresses: one kind of loop was a 0, another was a 1. An early design had removable modules that astronauts could replace in flight should they fail; by the middle of the Apollo program the computer became reliable enough for this to become unnecessary. The CPU ran at 1.024MHz; double-precision floating-point multiply took 1ms. Despite being half as powerful as the later Intel 8080, the computer ran a proper operating system with jobs, priorities and interrupts. Some jobs were written in assembly language, and some in a high-level language compiled to threaded code, like Forth. During the Apollo 11 landing, the computer became overloaded processing rendezvous radar data, which it should not have been doing because the lunar module was not doing a rendezvous; it showed alarms. The guidance officer at ground control told the astronauts to ignore the alarm; he was praised by President Nixon as "the young man [who], when the computers seemed to be confused and when he could have said "Stop," or when he could have said "Wait," said, "Go." " During the Apollo 14 mission, a ball of solder in the abort switch had shaken off loose and was floating in weightlessness and shorting a circuit; programmers wrote a patch to ignore the rogue signal, and had Mission Control dictate it to the astronauts.

The next American spacecraft after Apollo was the Space Shuttle. By the time it flew, computers had become much more powerful. During its design there were voices within NASA calling to allow it to fly fully automated missions, but they were overruled because no such missions were planned. Curiously, the only part where astronauts needed to do anything was deploying the landing gear, and after the Columbia disaster even that was automated, so if reentry protection were damaged, the astronauts could evacuate to the ISS and land the shuttle on autopilot. The Soviet Shuttle clone Buran made its only flight fully on autopilot.