by R. Passov
Over the course of the Apollo missions, two criteria governed the role of the Apollo Guidance Computer (AGC): 1) A program built into the AGC had to be absolutely necessary to the mission and; 2) it had to be without doubt that the computer bested humans in performing the task. One person stood at the center of every tradeoff.
On October 17th, 1966, Bill Tindall, Deputy Chief, Mission Planning and Analysis, released the delivery dates for the “manufacturing of the computer programs” that would guide the Apollo spacecrafts to the moon.
Described in a 1965 NYT article as an “Unassuming 40-year-old engineer on the ground,” Tindall’s job was to arbitrate between Mission Control’s evolving list of program objectives, the demands of the astronauts, the capabilities of a lab at MIT, and the fixed capacity of the Apollo Guidance Computer. When he felt the best tradeoffs had been reached, the pattern of bits representing the output of negotiations was sent to a factory.
Inside the factory, women sat in front of rectangular grids containing thousands of magnetic rings, an eighth of an inch in diameter. Copper wires were passed either between or around each ring. Each wire represented a ‘bit’ – an instance of a ‘1’ or a ‘0’ – and each bundle of sixteen wires, or Rope, a ‘word’ in permanent memory.
It could take up to 6 weeks to complete a weave. The pattern was the machine language (‘1’s and ‘0’) for the computer programs, crafted by mathematicians and physicists at MIT, that would land the Lunar Module on the Moon. The set of programs for a particular mission was contained in just over four-thousand ’16 bit words’ of woven copper wire.
The weaving took place in a factory owned by Raytheon, located in Waltham, Massachusetts near the textile mills from which the weavers were recruited. Known as ‘Rope Memory’ by its inventors, inside the factory, the blocks of woven wire were referred to as ‘LOL Memory.’ Little-old-lady memory.
* * * * * * * * * *
A few years back, working from the footage of a BBC documentary, I had managed to track down a Rope Lady, likely the last remaining one. An old fashion search through phone records led to a woman living on the outskirts of Boston.
In our first call, after wondering why I would be interested in something that happened so long ago, she opened up. “The Astronauts would come down to the factory to see what we were doing,” she said. “We were real proud. I really didn’t do a lot of weaving,” she added, “I was a supervisor.”
I left my phone number and address and asked for hers. Then sent a hand-written note that tried to explain my interest and why meeting her in person was so important. I was hoping to use her story as a path through that computer.
After giving her time to read my note, I got her back on the phone. Cautiously, she offered that while she didn’t walk that much, there was a coffee shop near enough and we could meet there.
When the week of our visit rolled around, I made a courtesy call. “Well,” she said, in a voice as relaxed as a late spring Sunday, “looks like I won’t be meeting you. A tree fell in my yard.” Then, in a quieter voice, she added, “And, I have the cancer.”
A real writer would have stuck to the story. But I had no business in her life and could tell from her voice that that was what she wanted me to understand.
* * * * * * * * * *
In 1962, contemporaneous with Kennedy’s launch of the Apollo Mission, Purdue established the first computer science degree program in the US. The Apollo Guidance Computer (there were two identical machines, one in the Command Module and one in the Lunar Module) came from young engineers, physicists and mathematicians, learning as they were building.
At the turn on the 21st century, NASA complied oral histories from Apollo veterans, including one from the person who helped me track down Mary Lou, the Rope Lady. At the launch of the Apollo program, Jack Garman was a 22-year-old engineer out of Michigan, assigned to help build the software for the Mission Control Computer.
Jack recalls being trained on everything: the physics of the hardware all the way through the math and logic that became the basis for the software. Like a 1950’s hot rod mechanic, he came to know everything about how the guidance computer worked – at every level. Something no longer possible
Jack offers that since there were no software layers, the output of the computer was actually the inner-workings.
From: Johnson Space Center Oral History Project – 27 March 2001
John (Jack) R. Garman:
Have you seen the movie Matrix? Anybody that stares at this years later is going to laugh at me for this … everything’s driven by computers, and there’s one scene where there’s a fellow staring at this sort of waterfall of numbers on a screen, and the main character says, “Well, what are you looking at?”
He says, “I’m looking at that virtual world.”
“You can stare at those numbers and actually see it?”
“Oh, yeah…,” and he’s just staring at numbers. That was deja vu for me on a much small[er] scale, because we couldn’t get them to put anything on the screen out of the computers except octal numbers.”
What Jack and his co-workers saw was the bit stream running across a crude display: ‘1’s and ‘0’s that comprised the consequence of the actual instruction that the (ground-based) computer was executing. By watching this stream, Jack interpreted what the computer was doing. During critical parts of a mission, fully absorbed by the output of the computer, he lost track of hours.
At some point, consoles were given lights – green, yellow and red. Good, not so good and not good. The lights were meant to help those who unlike Jack couldn’t read the instructions as they came across the screen. Then, someone built a map of the heavens to plot the path of the space craft. It was a hand-painted map on a table. As the computer followed the trajectory of the space craft, it drove a pen across the map, marking the trail in ink. Jack referred to this as a giant etch-a-sketch.
The ultimate abstraction was the camera that took real time pictures of the etch-a-sketch and then fed those images into the cathode ray tubes in the control room, giving mission control the feeling of looking at real-time output: A screen that showed a television picture of a computer-driven, mechanical drawing of the Apollo 11 as it headed toward the moon.
* * * * * * * * * *
Bill Tindall authored thousands of ‘Tindall Grams.’ They stand as a monument to the last time humans bested computers.
From: FM / Deputy Chief, Mission Planning and Analysis Division Jul 18, 1966
Subject: Rendezvous Terminal Phase Guidance Program in the Apollo Computer
“On July 7, 1966, a team of …’experts in rendezvous’…met at MIT to discuss and review the preliminary Guidance System Operation Plan (GSOP) which MIT has unofficially distributed …
…..In my opinion, this meeting was highly successful; and, since these processes – the terminal phase and the External [delta]V — are the most significant new requirements and the most controversial of the mission programs, I feel we are probably over the hump …
I would like to point out here the … items given the most attention at this meeting since they serve well to describe the character of the terminal phase rendezvous guidance philosophy:
Based on Gemini experience, the crew has emphasized that there is no requirement for automatic execution of the braking maneuvers by the G&N (guidance & navigation) system. … it is felt that this task can be carried out just as well, if not better, by the crew … However, there are occasions when automatic control of these maneuvers by the G&N might be mandatory. …Recognizing that procedures are available … to carry out the G&N controlled braking maneuvers by proper pilot manipulation of the computer, we deleted the requirement for automatic computer logic for this task. The point is, we felt that there was insufficient justification to carry out the extra programming, debuting, verification, and documentation, as well as using some 50 to 100 words of precious computer storage, for a program which was not needed, except in rather remote contingency situations…
One hundred words of computer storage today, if you could physically purchase such a small amount, would cost 1/10,000,000th of a penny.
From: FM / Chief, Apollo Data Priority Coordination Division May 29, 1969
Subject: DPS (Descent Propulsion System) low level propellant light
During our final review of the Descent Mission Techniques on May 28 th, GAEC (Grumman Aircraft Engineering Corporation) presented a comprehensive review of the low level DPS (Descent Propulsion System) propellant light – its operation and accuracy. The most significant piece of information coming from this was that we are assured of about 98 seconds more DPS operation at the lower thrust level after the light comes on…
We are proposing the following technique. The crew should commit to landing or else they should abort one minute after the low level light comes on. That is, the descent is continued in a normal manner for one minute after the light, which time the crew must decide that they can assuredly land or they should abort right then. By aborting right then they have approximately eight to ten-seconds DPS capability remaining at full thrust prior to propellant depletion. Selection of one minute as the go / no-go point came about based on an intuitive feeling that approximately eight to ten seconds of DPS thrusting is a reasonable minimum to get the LM the hell out of there coupled with the operational simplicity of keeping track of a[n] integer minute during this busy and exciting time…
Over $10 billion (in today’s dollars) worth of creativity and hardware, fully committed to the most expansive mission ever undertaken by humankind, all resting on the intuition of a single individual.