Imaginary interview with Margaret Hamilton
by Charactorium · Margaret Hamilton (1936 — ?) · Technology · Sciences · 6 min read
It is in a cluttered room at the MIT Instrumentation Laboratory, in Cambridge, that one winter morning in 1976, Jim Lovell meets Margaret Hamilton. On the tables, stacks of printed listings sit alongside the gray coils of core rope memory, and the air still smells of teleprinter ink. The astronaut and the engineer have known each other since Apollo: one night in December 1968, it was Hamilton's software that caught a mistake by Lovell, thousands of miles from Earth. Today he comes, notebook in hand, to finally understand what the invisible man inside the machine, watching over his life, was doing.
—Margaret, do you remember Apollo 8, late 1968? I accidentally erased navigation data. How did your program save me?
Of course I remember, Jim — you more than anyone felt the void when the screen went blank. What saved you was not a miracle, but a safeguard I had added against a fault that NASA deemed impossible. They had told me repeatedly that a trained astronaut would never launch the wrong program in mid-flight. I insisted on planning for it anyway, and I wrote the recovery routine that allowed navigation data to be reloaded from the ground. When your error occurred, the software already knew what to do. I didn't triumph that day: I mostly thought we were lucky to have defied their certainty.
They swore to me that an astronaut would never launch the wrong program. I planned for it anyway.
—They say your daughter, at the lab, put you on the trail of that flaw. Is that true, Margaret?
It's absolutely true, and it's a story I wouldn't have told a journalist. Evenings and weekends, I came back here with my daughter Lauren, because I didn't want to choose between her and work. One day, she played with a simulation console and pressed a key that crashed the entire navigation program in mid-simulated flight. Instead of scolding her, I thought: if a child can cause this, a man exhausted three hundred thousand kilometers away can too. I wanted to harden the system against that error. NASA refused. A few months later, you made it for real. Lauren, without knowing it, had written the specification.
If a child can crash the program, a man exhausted three hundred thousand kilometers away can too.
—You use a term no one understood when we worked together: software engineering. Where did that idea come from, Margaret?
It came from frustration, Jim. In the sixties, programming was not considered a real engineering profession — it was seen as tinkering, a matter of people who typed away. But I was writing the code your life depended on when you launched. So I started talking about software engineering to command respect: to say that my work deserved the same rigor, testing, and documentation as civil engineering or aeronautics. At first, people smiled, sometimes mocked. The word seemed pretentious. But I held firm, because a bridge that collapses kills, and poorly thought-out software in your capsule killed just as much. The name was a way to make this profession visible and responsible.
A bridge that collapses kills; poorly thought-out software in your capsule killed just as much.
—When I visited you in the evening, your team was mostly women, in a world of engineers in ties. How did you experience that?
I experienced it by working twice as hard, that's the truth. I was often the woman in a printed dress among white shirts, and I constantly had to prove that our software would hold. Aeronautical engineers, used to metal and equations, doubted that lines of code written by a team with many women could be trusted. So I imposed ruthless testing and documentation standards: every emergency case had to be anticipated, written, verified. We stayed late, sometimes with Lauren asleep in a corner, running through failure sequences no one else imagined. What I wanted was not to be liked: it was that none of our missions would experience a single fatal bug in flight. And we succeeded.
I didn't want to be liked; I wanted no mission to experience a single fatal bug in flight.
—Those nights spent here, poring over pages of code by hand — what were they really like, Margaret?
They were long, silent vigils, Jim. By day, I coordinated subgroups, discussed with NASA engineers, wrote specifications. But the real groundwork happened in the evening, when the lab emptied and the machines were finally free. I used the calm to run long test sequences, write documentation, and above all, search for failure scenarios we hadn't yet covered. I read through the printed listings line by line, pencil in hand, because there weren't enough screens for everyone. Lauren often slept nearby. It was in those unwitnessed hours that the reliability you benefited from up there was won: not in the glare of the launch, but in the darkness of a Cambridge office.
Reliability was not won in the glare of the launch, but in the darkness of a Cambridge office.
—Let's talk about Apollo 11. Three minutes from the lunar surface, the computer screamed the 1202 alarm. What was happening, Margaret?
What was happening was that the computer was receiving more work than it could handle in the allotted time. The 1202 code indicated an overload: an extra signal came from the rendezvous radar, and the little four-kilobyte machine was drowning in tasks. In any ordinary system, this would have frozen everything, and the landing would have been aborted. But our software didn't just compute: it knew how to prioritize. Faced with the overload, it automatically set aside secondary tasks to keep only those essential for descent. It, in a way, decided on its own what mattered for landing the module. It was that intelligence, planned years earlier, that let Neil Armstrong focus on the ground rising toward him.
The software didn't just compute: faced with danger, it decided on its own what mattered.
—How could such a small machine 'choose' to obey you rather than panic, as a man in its place might?
Thanks to a principle we called execution priority. Instead of treating all tasks as equal, we ranked each according to its importance for mission survival. The machine executed vital tasks first — guiding the descent, monitoring altitude — and only handled others if time remained. When the overload hit, it simply dropped the non-essential and cleanly restarted the essential functions, losing nothing that kept the module safe. It's the opposite of panic: an overwhelmed man often does everything at once and freezes; the system, on the other hand, let go of the non-essential without hesitation. We designed it to continue functioning correctly even in the midst of error. That's called fault tolerance, and it was my obsession.
An overwhelmed man does everything at once and freezes; the system, on the other hand, let go of the non-essential without hesitation.
—On that table, I see a stack of listings almost as tall as you. And those strange woven coils — what are they, Margaret?
That stack, Jim, is the entire Apollo code, printed page after page. When stacked, it reaches my shoulder height, and it makes me smile: so many years of nights fit into these sheets. The coils you're looking at are our ferrite core memories — what we call core rope memory. The program wasn't written as it is today: women workers physically wove copper wires through tiny magnetic rings, one wire through the ring for a one, beside it for a zero. Once woven, the program was etched into the material, impossible to corrupt by accident. Your spacecraft carried a software that could, literally, be held and touched. That's what made me serene: what you carried was not an abstraction, it was an object verified, thread by thread.
The program was woven thread by thread into the material: your spacecraft carried software you could touch with your finger.
—Today you are founding your own company. What are you still seeking, after having taken men to the Moon?
I seek to prevent error before it is born, not just catch it in flight. On Apollo, we corrected faults in real time, and that was already a revolution. But I realized that most failures stem from how a system is designed, long before the first line of code. With my method, which I call Higher Order Software, I want to base development on mathematical axioms, so that a large portion of errors becomes literally impossible to write. It's the logical continuation of everything we experienced together: if a life depends on a program, that program must not only react well to disaster, it must be conceived so that disaster has no place to lodge. That is my new obsession.
A program on which a life depends must not only react to disaster: it must make disaster impossible.
—When you look back, Margaret, what would you like people to remember about this work that remained so invisible for so long?
I would like them to remember that behind every launch, there were human beings bent over pages of code for entire nights. They saw the rocket, they saw your face at the capsule window; they did not see the girl asleep in a corner of the lab, nor the women who checked each instruction by hand. Software was long considered invisible, secondary, almost negligible. Yet it brought you back, it landed Armstrong on the Sea of Tranquility. I claim no personal glory, Jim: I simply want this profession to be granted the dignity of a full-fledged engineering discipline. If that is remembered, then all those Cambridge nights will not only have saved missions — they will have founded a field.
They saw the rocket and your face at the window; they did not see the women who checked each line by hand.
This imaginary interview was generated by artificial intelligence from sources documented in Margaret Hamilton's profile. It dramatises what the figure might have said based on what we know about them, but does not constitute attested historical testimony. For primary sources and factual documentation, refer to the full profile.