WEBVTT 00:00:00.000 --> 00:00:02.220 I want you to close your eyes for a second. Imagine 00:00:02.220 --> 00:00:05.940 you are suspended in a fragile metal box, effectively 00:00:05.940 --> 00:00:10.400 a tin can, exactly 240 ,000 miles away from everything 00:00:10.400 --> 00:00:15.199 you have ever known. It is July 20th, 1969. The 00:00:15.199 --> 00:00:18.019 lunar surface is rushing up to meet you at thousands 00:00:18.019 --> 00:00:20.699 of miles per hour. The eagle is descending. This 00:00:20.699 --> 00:00:24.010 is the moment. Exactly. Three minutes. That is 00:00:24.010 --> 00:00:26.769 all that separates Neil Armstrong and Buzz Aldrin 00:00:26.769 --> 00:00:29.550 from becoming the first humans to walk on another 00:00:29.550 --> 00:00:31.850 world or, you know, becoming a permanent crater 00:00:31.850 --> 00:00:33.829 on the Sea of Tranquility. And the tension in 00:00:33.829 --> 00:00:36.229 Mission Control is just suffocating. You can 00:00:36.229 --> 00:00:38.549 feel it in all the recordings. And then right 00:00:38.549 --> 00:00:41.490 at that razor's edge of history, the one thing 00:00:41.490 --> 00:00:44.009 keeping them alive, the guidance computer starts 00:00:44.009 --> 00:00:46.380 screaming. And it wasn't just a warning light, 00:00:46.439 --> 00:00:48.039 not like a check engine light in your car. It 00:00:48.039 --> 00:00:50.539 was a full -blown go -no -go crisis. The alarms 00:00:50.539 --> 00:00:53.880 flashing were 1201 and 1202. Right. And here 00:00:53.880 --> 00:00:56.200 is the terrifying reality of that moment. For 00:00:56.200 --> 00:00:59.280 several heart -stopping seconds, nobody, not 00:00:59.280 --> 00:01:00.899 the astronauts and not the flight controllers 00:01:00.899 --> 00:01:03.280 staring at their screens in Houston, knew exactly 00:01:03.280 --> 00:01:05.000 what the computer was trying to tell them. They 00:01:05.000 --> 00:01:06.700 had no idea. They didn't know if the brain of 00:01:06.700 --> 00:01:09.120 the spacecraft was about to die. And if the brain 00:01:09.120 --> 00:01:12.780 dies, the Eagle crashes. It's as simple as that. 00:01:13.060 --> 00:01:15.840 The standard narrative here usually pivots to 00:01:15.840 --> 00:01:18.840 Neil Armstrong's piloting skills, which were, 00:01:18.879 --> 00:01:21.099 I mean, incredible. Don't get me wrong. Oh, absolutely. 00:01:21.260 --> 00:01:24.079 A legend. Or the steely nerves of Gene Kranz 00:01:24.079 --> 00:01:27.180 in Houston. But that's not the story we are telling 00:01:27.180 --> 00:01:29.439 today. We are looking at the story that happened 00:01:29.439 --> 00:01:31.819 years before the launch. We are talking about 00:01:31.819 --> 00:01:34.120 the person whose logic was sitting inside that 00:01:34.120 --> 00:01:37.140 computer, making life or death decisions faster 00:01:37.140 --> 00:01:39.640 than any human brain could process. We are talking 00:01:39.640 --> 00:01:42.620 about Margaret Hamilton. And while the name has 00:01:42.620 --> 00:01:45.200 gained some recognition recently, you might have 00:01:45.200 --> 00:01:47.060 seen that photo of her standing next to a stack 00:01:47.060 --> 00:01:49.540 of code listings as tall as she is. It's an iconic 00:01:49.540 --> 00:01:52.540 photo. It is, but I don't think most people grasp 00:01:52.540 --> 00:01:54.579 the technical magnitude of what she achieved. 00:01:54.760 --> 00:01:57.599 She didn't just, you know, write code. She literally 00:01:57.599 --> 00:01:59.939 invented the discipline required to write it. 00:02:00.019 --> 00:02:02.640 She coined the term software engineering. That 00:02:02.640 --> 00:02:06.920 is what gets me every single time. She gave the 00:02:06.920 --> 00:02:10.330 job its name. The mission for this date dive 00:02:10.330 --> 00:02:13.310 is to move past the Wikipedia summary. We are 00:02:13.310 --> 00:02:15.710 going to dig into the source material to understand 00:02:15.710 --> 00:02:19.069 how a student of philosophy Yes, philosophy ended 00:02:19.069 --> 00:02:20.789 up as the director of the software engineering 00:02:20.789 --> 00:02:23.870 division at the MIT Instrumentation Lab. Right. 00:02:23.990 --> 00:02:26.909 And we need to understand the Wild West of 1960s 00:02:26.909 --> 00:02:28.710 computing because there were no rules, there 00:02:28.710 --> 00:02:30.550 were no textbooks, and there was no stack overflow 00:02:30.550 --> 00:02:32.889 to copy -paste from. You were on your own. Yeah, 00:02:32.909 --> 00:02:34.710 you couldn't just Google the error message. Not 00:02:34.710 --> 00:02:36.830 at all. And we are going to do a proper technical 00:02:36.830 --> 00:02:39.349 breakdown of the code itself. We need to unpack 00:02:39.349 --> 00:02:42.330 the asynchronous executive and the specific logic 00:02:42.330 --> 00:02:45.849 that saved Apollo 11 because, and this isn't 00:02:45.849 --> 00:02:48.189 an exaggeration, without... her specific approach 00:02:48.189 --> 00:02:50.669 to error prevention, the moon landing simply 00:02:50.669 --> 00:02:52.930 doesn't happen. And we'll wrap up by looking 00:02:52.930 --> 00:02:55.789 at her career post -NASA, which I think is arguably 00:02:55.789 --> 00:02:59.169 even more relevant to us today. She pivoted from 00:02:59.169 --> 00:03:02.069 how do we fix errors to how do we mathematically 00:03:02.069 --> 00:03:04.889 prove errors can't exist. It's a whole different 00:03:04.889 --> 00:03:07.370 mindset. A total paradigm shift. But let's start 00:03:07.370 --> 00:03:09.830 at the beginning. Margaret Heffield Hamilton, 00:03:10.090 --> 00:03:14.849 born 1936. Indiana born, Michigan raised. When 00:03:14.849 --> 00:03:17.360 we look at her college years, I don't see rocket 00:03:17.360 --> 00:03:20.620 scientist on the transcript. Not at all. No, 00:03:20.719 --> 00:03:22.800 you see a really curious mix. She started at 00:03:22.800 --> 00:03:25.699 the University of Michigan in 55. Then she transferred 00:03:25.699 --> 00:03:28.939 to Earlham College. She graduated in 1958 with 00:03:28.939 --> 00:03:31.699 a degree in mathematics. Okay, so math is there. 00:03:31.800 --> 00:03:34.439 That makes sense. Right. But if you look closer, 00:03:34.599 --> 00:03:37.479 there is a minor on that transcript that does 00:03:37.479 --> 00:03:39.719 a lot of heavy lifting later on. Philosophy. 00:03:40.000 --> 00:03:42.659 Philosophy. It feels like a distraction, doesn't 00:03:42.659 --> 00:03:45.439 it? If you're studying hard math, why spend time 00:03:45.439 --> 00:03:47.699 on abstract thought? Well, she credits her father, 00:03:47.840 --> 00:03:50.560 who is a poet. and her grandfather a headmaster 00:03:50.560 --> 00:03:53.139 for that interest but think about what philosophy 00:03:53.139 --> 00:03:55.520 actually is it isn't just sitting around asking 00:03:55.520 --> 00:03:58.159 why are we here right it's the study of logic 00:03:58.159 --> 00:04:00.580 it's the study of systems of thought you learn 00:04:00.580 --> 00:04:02.840 how to construct an argument how to find flaws 00:04:02.840 --> 00:04:05.439 in a logical chain it's about abstraction so 00:04:05.439 --> 00:04:07.659 it's meta thinking thinking about how we think 00:04:07.659 --> 00:04:10.699 exactly and when you are about to enter a field 00:04:10.699 --> 00:04:13.900 that doesn't exist yet where there are no manuals 00:04:13.900 --> 00:04:16.620 and no established procedures you can't rely 00:04:16.620 --> 00:04:18.899 on rote memorization there's no Nothing to memorize. 00:04:19.620 --> 00:04:22.019 There are no rules to follow. You have to rely 00:04:22.019 --> 00:04:25.779 on systems thinking. You have to be able to look 00:04:25.779 --> 00:04:28.180 at a problem abstractly and derive the rules 00:04:28.180 --> 00:04:31.240 yourself. That philosophical background likely 00:04:31.240 --> 00:04:33.279 gave her the ability to see the architecture 00:04:33.279 --> 00:04:35.639 behind the code, not just the commands themselves. 00:04:35.819 --> 00:04:38.000 She could see the forest, not just the trees. 00:04:38.240 --> 00:04:41.180 So she has the math. She has the logic. She graduates 00:04:41.180 --> 00:04:44.259 in 58. The plan was to be a professor, right? 00:04:44.750 --> 00:04:46.750 She was heading to Brandeis to study abstract 00:04:46.750 --> 00:04:49.350 math. That was the plan. But, you know, life 00:04:49.350 --> 00:04:51.829 intervened. Her husband was at Harvard Law School. 00:04:51.889 --> 00:04:53.629 They needed money to support him through that, 00:04:53.709 --> 00:04:55.709 so she put her own graduate studies on hold. 00:04:55.889 --> 00:04:57.470 So she starts looking for work. She starts looking 00:04:57.470 --> 00:05:00.069 for work, and she takes a job in mid -1959 at 00:05:00.069 --> 00:05:02.629 MIT. She landed in the meteorology department 00:05:02.629 --> 00:05:05.610 working for a man named Edward Norton Lorenz. 00:05:05.649 --> 00:05:07.990 Okay, I recognize that name immediately. Lorenz 00:05:07.990 --> 00:05:10.649 is the chaos theory guy. He is the father of 00:05:10.649 --> 00:05:14.250 chaos theory. But wait, 1959. Chaos theory wasn't 00:05:14.250 --> 00:05:16.629 a thing yet, was it? That term comes later. Not 00:05:16.629 --> 00:05:18.529 officially, no. And this is where the story gets 00:05:18.529 --> 00:05:21.389 so fascinating. Margaret was there at the incubation. 00:05:21.430 --> 00:05:23.829 She was tasked with writing software to predict 00:05:23.829 --> 00:05:26.949 weather on these massive, clunky machines, the 00:05:26.949 --> 00:05:30.810 LGP -30 and the PDP -1. The LGP -30. I looked 00:05:30.810 --> 00:05:33.199 this up. It's the size of a desk. weighs 800 00:05:33.199 --> 00:05:35.860 pounds, and probably had less computing power 00:05:35.860 --> 00:05:38.199 than a musical birthday card. Oh, much, much 00:05:38.199 --> 00:05:40.579 less. And it was loud. And you had to feed it 00:05:40.579 --> 00:05:43.259 instructions on paper tape. It was a very physical 00:05:43.259 --> 00:05:45.740 process. So what was she actually doing? She 00:05:45.740 --> 00:05:48.639 was programming it to solve these huge sets of 00:05:48.639 --> 00:05:50.720 differential equations, trying to model weather 00:05:50.720 --> 00:05:52.899 patterns. They'd run these long simulations, 00:05:53.139 --> 00:05:55.100 printing out the results as they went. And this 00:05:55.100 --> 00:05:57.139 is where the famous discovery happens. This is 00:05:57.139 --> 00:05:59.740 it. One day, Lorenz wanted to rerun a simulation 00:05:59.740 --> 00:06:01.660 from the middle. So instead of starting from 00:06:01.660 --> 00:06:03.980 zero, he took a printout from halfway through, 00:06:04.220 --> 00:06:06.959 typed the numbers back into the computer, and 00:06:06.959 --> 00:06:09.160 set it running again. And he expected to get 00:06:09.160 --> 00:06:11.160 the same results as the first run, just from 00:06:11.160 --> 00:06:14.899 that point forward. Of course. Same inputs, same 00:06:14.899 --> 00:06:18.139 math, should be the same outputs. But the result 00:06:18.139 --> 00:06:20.680 was completely different. The new simulation 00:06:20.680 --> 00:06:24.100 diverged wildly from the original. Why? If the 00:06:24.100 --> 00:06:26.199 math is the same, the result should be the same. 00:06:26.360 --> 00:06:28.959 You would think. But the computer, internally, 00:06:29.220 --> 00:06:32.620 stored numbers to six decimal places. The printer, 00:06:32.759 --> 00:06:35.480 to save paper and time, only printed out three. 00:06:35.949 --> 00:06:38.889 A rounding error. A tiny, tiny rounding error. 00:06:38.990 --> 00:06:40.730 So when they typed the numbers back in, they 00:06:40.730 --> 00:06:42.670 were rounding off these microscopic fractions. 00:06:43.250 --> 00:06:47.170 Something like 0 .506127 became just 0 .506. 00:06:47.410 --> 00:06:49.970 And that tiny difference changed the weather. 00:06:50.069 --> 00:06:52.209 It changed everything. It completely diverged. 00:06:52.470 --> 00:06:54.709 This was the discovery of sensitive dependence 00:06:54.709 --> 00:06:57.110 on initial conditions, which we now call the 00:06:57.110 --> 00:06:59.990 butterfly effect. The idea that a butterfly flapping 00:06:59.990 --> 00:07:02.350 its wings in Brazil can set off a tornado in 00:07:02.350 --> 00:07:04.970 Texas. And Margaret Hamilton was the one programming 00:07:04.970 --> 00:07:07.250 these. machines when this realization was happening 00:07:07.250 --> 00:07:09.949 she was right there she saw firsthand that in 00:07:09.949 --> 00:07:12.750 a complex system a microscopic error doesn't 00:07:12.750 --> 00:07:15.389 just result in a microscopic mistake it can cascade 00:07:15.389 --> 00:07:18.889 into total chaos That is a terrifying lesson 00:07:18.889 --> 00:07:21.370 to learn right before you go to work on a spaceship. 00:07:21.529 --> 00:07:23.430 Yeah. That must have stuck with her. It absolutely 00:07:23.430 --> 00:07:25.730 did. It's a foundational experience. Because 00:07:25.730 --> 00:07:28.430 what you're saying is she learned firsthand that 00:07:28.430 --> 00:07:30.509 you cannot just hope the system holds together. 00:07:30.730 --> 00:07:33.350 You can't assume that small inputs will only 00:07:33.350 --> 00:07:35.829 have small outputs. I can have massive, unpredictable 00:07:35.829 --> 00:07:39.389 outputs. Precisely. She took that lesson, that 00:07:39.389 --> 00:07:42.470 systems are fragile and nonlinear, and she carried 00:07:42.470 --> 00:07:44.410 it with her for the rest of her career. And she 00:07:44.410 --> 00:07:46.470 did all this without ever taking a class in computing. 00:07:46.509 --> 00:07:49.269 Like, because they didn't exist. Right. She said 00:07:49.269 --> 00:07:51.689 the field was so new that nobody could tell you 00:07:51.689 --> 00:07:53.610 that you were doing it wrong because nobody knew 00:07:53.610 --> 00:07:55.870 what right looked like yet. It was pure on -the 00:07:55.870 --> 00:07:59.069 -job training you learned by doing and by, well, 00:07:59.149 --> 00:08:01.889 crashing the system and figuring out why. So 00:08:01.889 --> 00:08:05.209 she leaves the weather lab, and in 1961, she 00:08:05.209 --> 00:08:07.769 moves to the Lincoln Lab at MIT to work on a 00:08:07.769 --> 00:08:10.850 project called SAGE. The semi -automatic ground 00:08:10.850 --> 00:08:13.529 environment. This sounds like the setting for 00:08:13.529 --> 00:08:16.129 a Dr. Strangelove movie. It practically was. 00:08:16.329 --> 00:08:19.610 SAGE was a massive Cold War project. The U .S. 00:08:19.629 --> 00:08:22.370 military had built these colossal computers, 00:08:22.769 --> 00:08:25.910 the AMFSQ -7, which were the largest computers 00:08:25.910 --> 00:08:28.730 ever built. Each one took up an entire building, 00:08:28.930 --> 00:08:32.129 weighed 250 tons, and had something like 60 ,000 00:08:32.129 --> 00:08:34.389 vacuum tubes. Unbelievable. What was it for? 00:08:34.590 --> 00:08:37.529 It was a networked air defense system. It took 00:08:37.529 --> 00:08:39.809 in radar data from all over North America, and 00:08:39.809 --> 00:08:42.850 its goal was to identify unfriendly aircraft. 00:08:43.230 --> 00:08:46.090 Soviet bombers? Exactly. The software was supposed 00:08:46.090 --> 00:08:49.029 to track incoming Soviet bombers and help guide 00:08:49.029 --> 00:08:51.149 interceptor jets to shoot them down before they 00:08:51.149 --> 00:08:53.350 reached American cities. So if the weather software 00:08:53.350 --> 00:08:55.730 crashes, maybe you need an umbrella when you 00:08:55.730 --> 00:08:57.750 thought it would be sunny if Sage crashes? We 00:08:57.750 --> 00:09:00.409 all die. Okay, so the stakes were just a little 00:09:00.409 --> 00:09:03.019 bit higher. The stakes were absolute. This was 00:09:03.019 --> 00:09:05.679 military grade computing. It was the proven ground 00:09:05.679 --> 00:09:08.620 for high reliability systems. But there is a 00:09:08.620 --> 00:09:11.039 specific story from her time at Sage that I think 00:09:11.039 --> 00:09:13.120 perfectly illustrates who Margaret Hamilton was. 00:09:13.360 --> 00:09:16.259 The Greek and Latin story. Yes. This is a classic 00:09:16.259 --> 00:09:18.919 example of hazing in the engineering world. It 00:09:18.919 --> 00:09:21.299 was a sort of rite of passage. Explain this to 00:09:21.299 --> 00:09:23.399 me, because it sounds like a complete waste of 00:09:23.399 --> 00:09:26.240 taxpayer money to be hazing people on a critical 00:09:26.240 --> 00:09:28.669 military defense project. Well, think of it as 00:09:28.669 --> 00:09:31.309 a stress test. When a new programmer arrived, 00:09:31.570 --> 00:09:33.509 especially someone without a formal background, 00:09:33.950 --> 00:09:36.769 the veterans, the priesthood of coders, would 00:09:36.769 --> 00:09:39.490 assign them a specific program to fix. A trick 00:09:39.490 --> 00:09:42.429 program. Exactly. It was known to be impossible. 00:09:42.769 --> 00:09:46.070 It was incredibly complex, convoluted logic. 00:09:46.230 --> 00:09:48.990 It had never run properly. And just to make it 00:09:48.990 --> 00:09:51.769 impossible to debug, the original author had 00:09:51.769 --> 00:09:54.669 written all the comments in the code in Greek 00:09:54.669 --> 00:09:57.570 and Latin. So you open the file. And instead 00:09:57.570 --> 00:09:59.850 of instructions saying this function calculates 00:09:59.850 --> 00:10:02.230 altitude, it's it's literally all Greek to you. 00:10:02.350 --> 00:10:04.230 Essentially, it was designed to be unreadable. 00:10:04.370 --> 00:10:06.850 The goal was to watch the new kid panic and fail. 00:10:07.049 --> 00:10:08.730 Nobody was supposed to be able to run it. It 00:10:08.730 --> 00:10:11.129 was just a way to humble the new recruits and, 00:10:11.230 --> 00:10:13.669 you know, see who cracked under pressure. So 00:10:13.669 --> 00:10:15.590 they hand this to Margaret. What does she do? 00:10:15.789 --> 00:10:18.750 She doesn't panic and she doesn't try to translate 00:10:18.750 --> 00:10:21.649 the Greek. She realizes the comments are a distraction. 00:10:21.970 --> 00:10:25.850 They are noise. She looks past them at the raw 00:10:25.850 --> 00:10:30.120 logic. the binary the actual flow of the data 00:10:30.120 --> 00:10:32.419 through the machine she reverse engineered it 00:10:32.419 --> 00:10:34.500 completely she figured out what it was actually 00:10:34.500 --> 00:10:36.679 trying to do and she found the flaws in the logic 00:10:36.679 --> 00:10:39.179 and she fixed it she made it run the impossible 00:10:39.179 --> 00:10:41.639 he made it run and then for the ultimate flex 00:10:41.639 --> 00:10:44.519 oh i love this part When she got it to output 00:10:44.519 --> 00:10:46.879 the answers, she programmed it to print them 00:10:46.879 --> 00:10:49.799 out in Latin and Greek. That is fantastic. That 00:10:49.799 --> 00:10:51.840 is the coding equivalent of catching a punch 00:10:51.840 --> 00:10:53.539 and hitting the other guy with his own hand. 00:10:53.679 --> 00:10:56.440 It really is. But deeper than the joke, it established 00:10:56.440 --> 00:10:59.240 her reputation. It proved she wasn't just a coder. 00:10:59.360 --> 00:11:02.620 She was a systems thinker, a problem solver who 00:11:02.620 --> 00:11:05.639 couldn't be intimidated by complexity or, you 00:11:05.639 --> 00:11:08.419 know, deliberate obfuscation. She couldn't be 00:11:08.419 --> 00:11:10.720 bullied. She could not. And that reputation is 00:11:10.720 --> 00:11:12.440 what put her on the radar for the apocalypse. 00:11:12.490 --> 00:11:16.429 So let's go to the moon. It's around 1963 when 00:11:16.429 --> 00:11:20.049 she leaves SAGE. And then she joins the MIT Instrumentation 00:11:20.049 --> 00:11:22.409 Lab, which had the contract for the Apollo guidance 00:11:22.409 --> 00:11:24.850 system. Right. And she said she saw an ad about 00:11:24.850 --> 00:11:27.049 the Apollo project and just found the idea of 00:11:27.049 --> 00:11:30.750 a moon program very exciting. It was the challenge 00:11:30.750 --> 00:11:34.070 of a lifetime. She joins the lab. Yeah. And we 00:11:34.070 --> 00:11:36.110 need to be very clear about her role here because 00:11:36.110 --> 00:11:39.190 history tends to flatten these things. She wasn't 00:11:39.190 --> 00:11:42.000 just a programmer on the team. No, she was the 00:11:42.000 --> 00:11:44.600 first programmer hired for the Apollo project 00:11:44.600 --> 00:11:47.700 at MIT and obviously the first female programmer. 00:11:47.940 --> 00:11:50.100 That is a lonely position to be in. The first 00:11:50.100 --> 00:11:52.590 person through the door. Incredibly lonely. And 00:11:52.590 --> 00:11:54.490 she eventually rose to be the director of the 00:11:54.490 --> 00:11:56.590 software engineering division, leading a team 00:11:56.590 --> 00:11:58.889 of hundreds. But it was an uphill battle the 00:11:58.889 --> 00:12:01.289 entire way. Not just because of her gender, though 00:12:01.289 --> 00:12:02.950 that was certainly part of it, but because of 00:12:02.950 --> 00:12:05.190 her discipline. Right, the hardware versus software 00:12:05.190 --> 00:12:07.809 culture war. This was a real thing. It was a 00:12:07.809 --> 00:12:10.830 caste system. In the 1960s, engineering meant 00:12:10.830 --> 00:12:13.590 hardware. It meant building rockets, fuel pumps, 00:12:13.809 --> 00:12:16.710 circuits, metal. That was seen as real work. 00:12:17.049 --> 00:12:19.710 Tangible things you could hold. Exactly. Software. 00:12:20.399 --> 00:12:23.159 That was viewed as a secondary support task. 00:12:23.539 --> 00:12:26.120 It was seen as typing or something women did. 00:12:26.340 --> 00:12:28.500 It wasn't considered a science. It was almost 00:12:28.500 --> 00:12:31.120 like a craft. Which is so ironic because the 00:12:31.120 --> 00:12:33.480 hardware is just a heavy, expensive paperweight 00:12:33.480 --> 00:12:36.360 without the software to tell it where to go and 00:12:36.360 --> 00:12:39.019 what to do. Completely. But the hardware gurus 00:12:39.019 --> 00:12:40.779 didn't see it that way. They thought software 00:12:40.779 --> 00:12:42.620 was something you did at the end, like, you know, 00:12:42.659 --> 00:12:45.440 painting the rocket. Hamilton realized early 00:12:45.440 --> 00:12:47.879 on that this attitude was incredibly dangerous. 00:12:48.259 --> 00:12:50.440 Because if you treat it as an afterthought. You 00:12:50.440 --> 00:12:53.299 get bugs. And in space, bugs kill people. There 00:12:53.299 --> 00:12:55.840 are no second chances. So she realized they needed 00:12:55.840 --> 00:12:58.799 more rigor. more discipline they needed to be 00:12:58.799 --> 00:13:01.840 seen as equals to the hardware engineers so she 00:13:01.840 --> 00:13:04.480 starts fighting for legitimacy she did she started 00:13:04.480 --> 00:13:07.000 using the term software engineering she combined 00:13:07.000 --> 00:13:09.379 the words deliberately she was making a statement 00:13:09.379 --> 00:13:12.399 she was saying we are not just typing we are 00:13:12.399 --> 00:13:15.299 engineering a system with the same rigor the 00:13:15.299 --> 00:13:17.799 same mathematical foundation and the same discipline 00:13:17.799 --> 00:13:20.179 that you use to build a thruster and they laughed 00:13:20.179 --> 00:13:22.779 at her is that true it was an ongoing joke for 00:13:22.779 --> 00:13:25.100 a while The sources say the hardware guys would 00:13:25.100 --> 00:13:27.740 kid her about her radical ideas. They thought 00:13:27.740 --> 00:13:29.679 it was hilarious that this woman from the typing 00:13:29.679 --> 00:13:32.779 pool wanted to call what she did engineering. 00:13:33.340 --> 00:13:36.120 Unbelievable. Until they realized she was right. 00:13:36.509 --> 00:13:38.549 Until the problem started getting so complex 00:13:38.549 --> 00:13:40.610 that they couldn't solve them without her team. 00:13:40.809 --> 00:13:43.850 There was a turning point, a specific meeting 00:13:43.850 --> 00:13:47.330 where a very respected hardware expert, one of 00:13:47.330 --> 00:13:50.509 the gods of the lab, finally stood up and admitted, 00:13:50.590 --> 00:13:52.929 essentially, she's right. What they are doing 00:13:52.929 --> 00:13:55.370 is engineering. That must have been a huge moment. 00:13:55.490 --> 00:13:57.330 Hamilton said it was. It was the moment they 00:13:57.330 --> 00:13:59.250 stopped being the help and started being partners 00:13:59.250 --> 00:14:01.029 in the mission. And they had a lot of work to 00:14:01.029 --> 00:14:03.720 do. Her team was responsible for all the onboard 00:14:03.720 --> 00:14:06.419 in -flight software. Everything. The command 00:14:06.419 --> 00:14:08.740 module software that got them to the moon and 00:14:08.740 --> 00:14:11.460 back, and the lunar module software for the landing 00:14:11.460 --> 00:14:13.740 itself, and later they even did the software 00:14:13.740 --> 00:14:16.059 for the Skylab space station. What was the key 00:14:16.059 --> 00:14:18.539 focus? What was the philosophy behind the software? 00:14:18.940 --> 00:14:21.769 It was all about reliability. error detection, 00:14:22.129 --> 00:14:25.330 error recovery, and this concept of man -in -the 00:14:25.330 --> 00:14:27.970 -loop capabilities. How do the humans and the 00:14:27.970 --> 00:14:30.669 computer interact safely under extreme stress? 00:14:31.070 --> 00:14:34.190 That was her obsession. Let's get into the weeds 00:14:34.190 --> 00:14:36.269 of that code then, because this is Section 4, 00:14:36.370 --> 00:14:38.750 the architecture of survival. And I want to really 00:14:38.750 --> 00:14:40.870 understand this, because the critique we've heard 00:14:40.870 --> 00:14:43.009 of other explanations is that they dumb it down 00:14:43.009 --> 00:14:45.750 too much. Let's do it. We need to talk about 00:14:45.750 --> 00:14:48.129 the asynchronous executive. Yeah. The previous 00:14:48.129 --> 00:14:50.789 explanation I've heard used a grocery store analogy, 00:14:51.029 --> 00:14:54.509 like checking out customers. But that feels too 00:14:54.509 --> 00:14:57.169 linear, too simple. It is too linear. A grocery 00:14:57.169 --> 00:15:00.269 line is a queue. First come, first served. You 00:15:00.269 --> 00:15:02.610 wait your turn. The Apollodidon's computer could 00:15:02.610 --> 00:15:05.070 not afford to wait its turn. A better analogy 00:15:05.070 --> 00:15:07.370 is an emergency room doctor on a Saturday night. 00:15:07.549 --> 00:15:10.549 Okay, I'm the doctor. The CPU. You are the CTU. 00:15:10.549 --> 00:15:12.590 You have a patient with a sprained ankle. Let's 00:15:12.590 --> 00:15:14.450 call that job updating the astronaut's display. 00:15:14.809 --> 00:15:17.049 It's important the patient is in pain, but it's 00:15:17.049 --> 00:15:18.750 not life -threatening, so you're working on that. 00:15:18.850 --> 00:15:21.350 I'm wrapping the ankle. Then suddenly the ambulance 00:15:21.350 --> 00:15:24.629 doors burst open and a patient comes in on a 00:15:24.629 --> 00:15:27.309 stretcher with a massive heart attack. That job 00:15:27.309 --> 00:15:30.750 is fire the descent engine to stop us crashing 00:15:30.750 --> 00:15:33.129 into the moon. Okay, that's a bit more important 00:15:33.129 --> 00:15:35.730 than the ankle. Just a bit. In a normal linear 00:15:35.730 --> 00:15:38.470 computer, a grocery line computer, you'd have 00:15:38.470 --> 00:15:40.169 to finish taking the ankle before you could even 00:15:40.169 --> 00:15:42.009 look at the heart attack patient. And the heart 00:15:42.009 --> 00:15:44.720 attack patient dies. Exactly. The eagle crashes. 00:15:45.679 --> 00:15:48.019 Hamilton's team, utilizing a brilliant design 00:15:48.019 --> 00:15:51.299 by a man named Jay Halcombe Laning, built an 00:15:51.299 --> 00:15:54.820 asynchronous executive. It's a system based on 00:15:54.820 --> 00:15:56.940 interrupts and priorities. So what does that 00:15:56.940 --> 00:15:59.379 mean in the ER? The moment the heart attack patient 00:15:59.379 --> 00:16:02.539 arrives, the system literally freezes the ankle 00:16:02.539 --> 00:16:06.080 taping job mid -motion. It saves the state of 00:16:06.080 --> 00:16:07.960 that job, where the tape was, what you're thinking 00:16:07.960 --> 00:16:11.000 about, and it shoves it onto a back burner. Instantly, 00:16:11.059 --> 00:16:13.620 your entire brain, all your processing power, 00:16:13.759 --> 00:16:16.799 switches to the heart attack. It's ruthless prioritization. 00:16:17.159 --> 00:16:20.039 Ruthless and automatic. It creates a hierarchy 00:16:20.039 --> 00:16:22.860 of survival. The computer's programming had a 00:16:22.860 --> 00:16:25.679 list of jobs with priority ratings. Landing the 00:16:25.679 --> 00:16:27.879 vehicle was at the top. Everything else was below 00:16:27.879 --> 00:16:30.879 it. The computer knew, without being told, that 00:16:30.879 --> 00:16:33.379 keeping the ship upright was more important than 00:16:33.379 --> 00:16:35.480 talking to the astronaut. But this introduces 00:16:35.480 --> 00:16:38.500 a new problem, right? The human interface. If 00:16:38.500 --> 00:16:40.919 the computer is jumping between tasks like that, 00:16:41.059 --> 00:16:43.220 how does the astronaut know what's happening? 00:16:43.500 --> 00:16:46.220 That was the priority display challenge, and 00:16:46.220 --> 00:16:49.039 it was a huge focus for Hamilton. She designed 00:16:49.039 --> 00:16:51.080 routines where the computer could interrupt the 00:16:51.080 --> 00:16:53.179 astronauts. It could overwrite their screen if 00:16:53.179 --> 00:16:55.600 it had something urgent to say. But she realized 00:16:55.600 --> 00:16:58.539 this created a deadly new risk. The man -in -the 00:16:58.539 --> 00:17:00.960 -loop problem. The human is too slow. Exactly. 00:17:01.440 --> 00:17:03.860 Imagine you are the astronaut. You're about to 00:17:03.860 --> 00:17:05.680 press a button to say, turn on the landing radar. 00:17:06.299 --> 00:17:09.180 your finger is moving toward the button. In that 00:17:09.180 --> 00:17:11.200 split second, the computer has a problem and 00:17:11.200 --> 00:17:13.339 decides to interrupt you. It changes the screen 00:17:13.339 --> 00:17:15.740 to a different menu, and now that same button 00:17:15.740 --> 00:17:18.259 means jettison the engine. And my finger is already 00:17:18.259 --> 00:17:21.980 moving. I press the button. And you've just jettisoned 00:17:21.980 --> 00:17:24.079 the engine you need to land because the computer 00:17:24.079 --> 00:17:26.440 changed the context faster than your brain could 00:17:26.440 --> 00:17:29.619 react. That is a UI nightmare. How did she solve 00:17:29.619 --> 00:17:33.140 that? With a surprisingly simple, low -tech human 00:17:33.140 --> 00:17:35.619 rule that was drilled into the astronauts during 00:17:35.619 --> 00:17:38.299 training. Count to five. Count to five. That's 00:17:38.299 --> 00:17:40.819 it. That was the core of it. Yeah. If a priority 00:17:40.819 --> 00:17:43.720 display pops up, if the screen changes unexpectedly, 00:17:44.220 --> 00:17:46.920 the rule was you take your hands off the controls 00:17:46.920 --> 00:17:49.759 and wait five seconds. Let the system settle. 00:17:50.019 --> 00:17:52.579 Make sure the switchover is complete before you 00:17:52.579 --> 00:17:55.140 touch anything. It's a human buffer for a machine 00:17:55.140 --> 00:17:58.140 that's too fast. Precisely. It's such a low -tech 00:17:58.140 --> 00:18:00.480 solution to a high -tech problem, but it was 00:18:00.480 --> 00:18:02.400 critical for preventing accidental disasters. 00:18:02.880 --> 00:18:04.700 And then there is this feature that really saved 00:18:04.700 --> 00:18:07.480 the day on Apollo 11, the kill and recompute. 00:18:07.869 --> 00:18:10.690 This sounds aggressive. It is the ultimate fail 00:18:10.690 --> 00:18:12.750 -safe. Think about your laptop. If it freezes, 00:18:12.789 --> 00:18:14.769 what do you do? I hold the power button down. 00:18:14.890 --> 00:18:17.849 It shuts off, I swear a bit, and I probably lose 00:18:17.849 --> 00:18:20.509 the document I was working on. Right. In space, 00:18:20.730 --> 00:18:23.089 you cannot lose the document. You cannot lose 00:18:23.089 --> 00:18:25.369 your navigation alignment. If you restart from 00:18:25.369 --> 00:18:27.529 zero, you don't know where you are and you tumble. 00:18:27.769 --> 00:18:30.950 So, Hamilton's team... built a restart protection 00:18:30.950 --> 00:18:34.269 system. How did it work? The software was constantly 00:18:34.269 --> 00:18:37.170 taking snapshots of critical data, your position, 00:18:37.369 --> 00:18:40.049 your velocity, your orientation. It was saving 00:18:40.049 --> 00:18:42.369 these to protected areas of memory, like a series 00:18:42.369 --> 00:18:44.890 of safe checkpoints. Like a save game in a video 00:18:44.890 --> 00:18:47.190 game. That's a perfect analogy. It's a save game 00:18:47.190 --> 00:18:49.869 checkpoint. If the software ever realized it 00:18:49.869 --> 00:18:52.329 was confused or getting corrupt data or about 00:18:52.329 --> 00:18:55.029 to crash, it wouldn't just blue screen. It was 00:18:55.029 --> 00:18:57.730 programmed to detect its own failure. And then 00:18:57.730 --> 00:19:00.150 what? it would perform a soft reboot. It would 00:19:00.150 --> 00:19:02.289 flush all the currently running lower priority 00:19:02.289 --> 00:19:06.029 tasks, reload the last known safe snapshot of 00:19:06.029 --> 00:19:08.589 critical data, and restart its processing from 00:19:08.589 --> 00:19:11.069 that safe point. And it could do all this in 00:19:11.069 --> 00:19:13.329 milliseconds. So it was designed to fail and 00:19:13.329 --> 00:19:15.829 recover so fast the human might not even notice. 00:19:16.109 --> 00:19:18.170 That was the goal. It was resilience by design. 00:19:18.329 --> 00:19:20.750 The system assumed failure could happen and was 00:19:20.750 --> 00:19:22.789 built to survive it. And that brings us to the 00:19:22.789 --> 00:19:27.019 climax. July 20th, 1969. The ultimate test of 00:19:27.019 --> 00:19:29.359 that resilience. All those years of philosophical 00:19:29.359 --> 00:19:32.680 thinking and design choices are about to be put 00:19:32.680 --> 00:19:35.920 on trial. We are back in the Eagle. Three minutes 00:19:35.920 --> 00:19:39.339 to landing. Ah! And the alarms start. Yeah. 1201, 00:19:39.440 --> 00:19:43.160 1202. Right. And to understand why this happened, 00:19:43.259 --> 00:19:45.240 we have to look at a piece of hardware called 00:19:45.240 --> 00:19:47.559 the rendezvous radar. This is the radar they 00:19:47.559 --> 00:19:49.400 would use to find the mothership, the command 00:19:49.400 --> 00:19:52.059 module, if they had to abort the landing and 00:19:52.059 --> 00:19:54.559 fly back up, right? Correct. It wasn't needed 00:19:54.559 --> 00:19:57.240 for the landing itself. But, and there is some 00:19:57.240 --> 00:19:59.579 debate here among historians and even the astronauts, 00:19:59.859 --> 00:20:02.279 but the prevailing theory from the engineers 00:20:02.279 --> 00:20:04.940 points to the switch position. The rendezvous 00:20:04.940 --> 00:20:07.220 radar switch was likely left in a mode that it 00:20:07.220 --> 00:20:09.480 shouldn't have been. So it was on when it should 00:20:09.480 --> 00:20:11.279 have been off. It's a little more complex than 00:20:11.279 --> 00:20:13.460 that. It was likely about the phasing of the 00:20:13.460 --> 00:20:16.019 power supply to the radar. Due to the switch 00:20:16.019 --> 00:20:18.279 position, the radar was receiving power that 00:20:18.279 --> 00:20:20.539 was out of phase with the computer's power supply. 00:20:20.700 --> 00:20:23.119 This created electrical interference. Like static. 00:20:23.930 --> 00:20:27.329 Just noise. Massive noise. The radar started 00:20:27.329 --> 00:20:29.849 sending thousands of phantom signals interrupts 00:20:29.849 --> 00:20:32.430 to the computer every second. It was essentially 00:20:32.430 --> 00:20:36.289 screaming, read me, read me, read me at the CPU, 00:20:36.430 --> 00:20:38.569 demanding attention for a job that wasn't even 00:20:38.569 --> 00:20:40.769 important at that moment. So the computer is 00:20:40.769 --> 00:20:42.869 trying to do the incredibly complex math for 00:20:42.869 --> 00:20:45.269 the descent, which is already hard. And suddenly 00:20:45.269 --> 00:20:47.529 it has this screaming child in the backseat kicking 00:20:47.529 --> 00:20:49.910 the chair over and over. An excellent analogy. 00:20:50.089 --> 00:20:52.650 And then the straw that broke the camel's back. 00:20:53.549 --> 00:20:56.289 Buzz Aldrin, following his training, enters a 00:20:56.289 --> 00:21:00.210 command, 1668. And what does command 1668 do? 00:21:00.450 --> 00:21:02.849 It asks the computer to calculate and display 00:21:02.849 --> 00:21:05.710 the difference in altitude delta. Buzz wanted 00:21:05.710 --> 00:21:07.390 to double check their height against what he 00:21:07.390 --> 00:21:09.569 was seeing out the window. A totally reasonable 00:21:09.569 --> 00:21:12.170 request. He had done it in simulations a hundred 00:21:12.170 --> 00:21:14.289 times. But the simulations didn't have the radar 00:21:14.289 --> 00:21:17.259 screaming in the background. Exactly. The system 00:21:17.259 --> 00:21:19.619 was designed to handle a workload of about 85 00:21:19.619 --> 00:21:22.339 % of its total capacity to leave a safety margin. 00:21:22.819 --> 00:21:25.599 With the landing software running and now this 00:21:25.599 --> 00:21:28.299 constant barrage of radar interrupts, it was 00:21:28.299 --> 00:21:30.700 already pushed near its limit. When Aldrin asked 00:21:30.700 --> 00:21:32.920 for that extra display calculation, that was 00:21:32.920 --> 00:21:35.359 the eighth major task on a system built to handle 00:21:35.359 --> 00:21:38.359 seven, the demand on the CPU hit something like 00:21:38.359 --> 00:21:42.400 115%. Overload. Total saturation. In any other 00:21:42.400 --> 00:21:44.779 computer of that era, probably any other computer 00:21:44.779 --> 00:21:47.000 on Earth at that moment, the system would have 00:21:47.000 --> 00:21:49.759 frozen. It would have crashed. The thrusters 00:21:49.759 --> 00:21:52.039 would have locked in their last position. The 00:21:52.039 --> 00:21:53.660 eagle would have tumbled and crashed into the 00:21:53.660 --> 00:21:56.759 moon. But it didn't freeze. The 1201 alarm flashed. 00:21:56.759 --> 00:21:58.579 And that alarm was the software talking. It was 00:21:58.579 --> 00:22:01.660 the system communicating its state. 1201 meant 00:22:01.660 --> 00:22:04.539 executive overflow. The asynchronous executive, 00:22:04.940 --> 00:22:07.220 the ER doctor, looked at the list of tasks and 00:22:07.220 --> 00:22:09.549 said. I have too much to do. I cannot process 00:22:09.549 --> 00:22:11.950 all of this in the time I have. And here is where 00:22:11.950 --> 00:22:14.170 the genius of the design, the genius of Margaret 00:22:14.170 --> 00:22:17.250 Hamilton's philosophy kicks in. It didn't quit. 00:22:17.490 --> 00:22:20.569 No, it made a management decision. It looked 00:22:20.569 --> 00:22:23.730 at its prioritized list of jobs and said. updating 00:22:23.730 --> 00:22:26.349 Buzz Aldrin's display. That is low priority. 00:22:26.650 --> 00:22:29.329 Kill it. Servicing those screaming radar interrupts, 00:22:29.329 --> 00:22:32.250 not critical for landing. Kill it. Firing the 00:22:32.250 --> 00:22:34.369 descent engine to keep us from dying. Highest 00:22:34.369 --> 00:22:36.710 priority. Keep that running at all costs. It 00:22:36.710 --> 00:22:39.430 effectively told the astronauts, shut up, I'm 00:22:39.430 --> 00:22:42.029 busy saving your life. That is exactly what it 00:22:42.029 --> 00:22:44.690 did. It shed the load. It performed one of those 00:22:44.690 --> 00:22:47.710 soft reboots, a kill and recompute, cleared the 00:22:47.710 --> 00:22:50.390 queue of low -priority junk, reloaded the last 00:22:50.390 --> 00:22:53.250 safe data, and kept the engines firing. And it 00:22:53.250 --> 00:22:55.369 did this multiple times during the final descent. 00:22:55.630 --> 00:22:57.910 And in Houston, this is where the human element 00:22:57.910 --> 00:23:00.269 on the ground becomes critical. Jack Garman. 00:23:00.369 --> 00:23:02.349 Jack Garman, a young NASA computer engineer, 00:23:02.450 --> 00:23:04.470 one of the backroom guys, he was looking at his 00:23:04.470 --> 00:23:07.049 console. He had a cheat sheet of all the possible 00:23:07.049 --> 00:23:09.230 error codes taped onto the glass on his desk. 00:23:09.710 --> 00:23:12.150 Because Margaret Hamilton had insisted on rigorous 00:23:12.150 --> 00:23:14.390 testing and creating these very specific error 00:23:14.390 --> 00:23:18.269 definitions, Garmin knew what 1201 meant. He 00:23:18.269 --> 00:23:20.210 knew it wasn't a death rattle. It wasn't the 00:23:20.210 --> 00:23:22.309 computer dying. He knew it was the opposite. 00:23:22.569 --> 00:23:24.789 He knew it meant the computer was working perfectly. 00:23:24.930 --> 00:23:27.829 It was doing its job, shedding tasks to survive. 00:23:28.150 --> 00:23:30.349 So he yelled into the intercom to the flight 00:23:30.349 --> 00:23:33.789 director, go, go. He gave them the go. He gave 00:23:33.789 --> 00:23:36.829 them the go. If he hadn't understood that, if 00:23:36.829 --> 00:23:39.890 they had hesitated for even a few seconds, the 00:23:39.890 --> 00:23:42.089 mission rules were clear. They would have had 00:23:42.089 --> 00:23:44.589 to abort. So Margaret Hamilton's code didn't 00:23:44.589 --> 00:23:47.250 just fly the ship. It gave mission control the 00:23:47.250 --> 00:23:49.470 information and the confidence to stay in the 00:23:49.470 --> 00:23:51.900 fight. Hamilton wrote about this later in a letter 00:23:51.900 --> 00:23:54.900 to Datamation magazine. She said, and I'm paraphrasing, 00:23:55.079 --> 00:23:57.599 if the computer hadn't recognized this problem 00:23:57.599 --> 00:24:00.599 and taken recovery action, I doubt if Apollo 00:24:00.599 --> 00:24:02.799 11 would have been the successful moon landing 00:24:02.799 --> 00:24:05.640 it was. It's just chilling to think about. The 00:24:05.640 --> 00:24:08.000 margin between a ticker tape parade and a global 00:24:08.000 --> 00:24:10.920 memorial service was a few lines of priority 00:24:10.920 --> 00:24:14.200 logic in a 72 kilobyte computer. It really was. 00:24:14.299 --> 00:24:16.339 The whole mission balanced on that knife's edge. 00:24:16.539 --> 00:24:20.569 So Apollo 11 lands. History is made. Margaret 00:24:20.569 --> 00:24:24.309 Hamilton is a legend inside the lab. But her 00:24:24.309 --> 00:24:27.730 story doesn't end in 1969. And this is the part 00:24:27.730 --> 00:24:29.829 that I think is most critical for our listeners, 00:24:29.990 --> 00:24:31.769 especially anyone working in tech or systems 00:24:31.769 --> 00:24:35.650 today. Section 6. The Post -Apollo Methodology. 00:24:35.920 --> 00:24:38.359 This is where she pivots from being a heroic 00:24:38.359 --> 00:24:41.759 firefighter to being a fire marshal. She looked 00:24:41.759 --> 00:24:43.460 back at the Apollo project and realized that 00:24:43.460 --> 00:24:45.940 what they did was amazing, but it was also exhausting 00:24:45.940 --> 00:24:49.019 and, in a way, inefficient. They spent thousands 00:24:49.019 --> 00:24:51.660 and thousands of hours testing, finding bugs, 00:24:51.799 --> 00:24:54.900 and fixing them. finding the mistakes after they've 00:24:54.900 --> 00:24:56.880 been made. Right. But she asked a deeper, more 00:24:56.880 --> 00:24:59.220 philosophical question. Why do we allow errors 00:24:59.220 --> 00:25:01.140 to exist in the first place? It sounds like a 00:25:01.140 --> 00:25:03.460 utopian question. I mean, come on. Code always 00:25:03.460 --> 00:25:05.119 has bugs. That's just the rule of the universe. 00:25:05.240 --> 00:25:07.640 Or do we just accept that it does? Because that's 00:25:07.640 --> 00:25:10.690 how we've always done it. In 1976, she co -founded 00:25:10.690 --> 00:25:13.329 a company called Higher Order Software, or HOS. 00:25:13.529 --> 00:25:17.170 And later, in 1986, she founded Hamilton Technologies. 00:25:17.509 --> 00:25:19.750 And through these companies, she developed a 00:25:19.750 --> 00:25:22.049 concept she called Universal Systems Language, 00:25:22.269 --> 00:25:25.630 or USL. Okay, let's unpack USL. Yeah. What is 00:25:25.630 --> 00:25:28.589 the core idea here? It sounds very abstract. 00:25:28.970 --> 00:25:31.569 The core philosophy is development before the 00:25:31.569 --> 00:25:34.509 fact. Development before the fact. Explain that 00:25:34.509 --> 00:25:37.329 to me. Think about how we build almost all software 00:25:37.329 --> 00:25:40.470 today. You write some code, you compile it, you 00:25:40.470 --> 00:25:43.450 run it, it crashes, you say, oops, you go into 00:25:43.450 --> 00:25:46.029 the debugger, you find the mistake, and you patch 00:25:46.029 --> 00:25:48.670 it. We call it agile. We iterate. We iterate 00:25:48.670 --> 00:25:51.609 on failure. We test after the fact to find the 00:25:51.609 --> 00:25:55.170 errors we built in. So it's like building a house 00:25:55.170 --> 00:25:57.289 with a blueprinting software that physically 00:25:57.289 --> 00:25:59.170 prevents you from putting a door where a load 00:25:59.170 --> 00:26:01.369 -bearing wall needs to be. It just won't let 00:26:01.369 --> 00:26:03.990 you draw it. That is a perfect analogy. If the 00:26:03.990 --> 00:26:06.069 design is mathematically sound and complete, 00:26:06.329 --> 00:26:08.950 the code that is automatically generated from 00:26:08.950 --> 00:26:11.950 it will be reliable by definition. It moves the 00:26:11.950 --> 00:26:14.529 focus from software reliability, which is asking, 00:26:14.569 --> 00:26:17.890 is the code good, to system reliability, which 00:26:17.890 --> 00:26:20.819 asks, is the logic sound. And when you look at 00:26:20.819 --> 00:26:23.900 the world today, I mean, you look at major software 00:26:23.900 --> 00:26:26.539 outages that take down airlines and banks or 00:26:26.539 --> 00:26:28.920 the constant stream of security patches for our 00:26:28.920 --> 00:26:31.359 phones because of vulnerabilities. It feels like 00:26:31.359 --> 00:26:33.099 we went in the complete opposite direction of 00:26:33.099 --> 00:26:35.680 Hamilton's philosophy. We absolutely did. The 00:26:35.680 --> 00:26:38.200 industry, by and large, chose speed over rigor. 00:26:38.359 --> 00:26:40.940 The mantra became move fast and break things. 00:26:41.440 --> 00:26:44.000 Hamilton's philosophy was move deliberately and 00:26:44.000 --> 00:26:46.539 break nothing. And in a world where software 00:26:46.539 --> 00:26:48.759 is now controlling self -driving cars, pacemakers 00:26:48.759 --> 00:26:51.319 and national. power grids, her approach of development 00:26:51.319 --> 00:26:54.079 before the fact is arguably more relevant now 00:26:54.079 --> 00:26:57.579 than it was in 1969. I believe so. It's a road 00:26:57.579 --> 00:27:00.240 not taken, or at least a road less traveled by 00:27:00.240 --> 00:27:02.759 the mainstream. But it's a standard we might 00:27:02.759 --> 00:27:04.839 need to return to as the stakes get higher and 00:27:04.839 --> 00:27:07.180 higher. Let's look at her legacy. Because for 00:27:07.180 --> 00:27:09.019 a long time, she was a bit of a hidden figure 00:27:09.019 --> 00:27:11.700 outside of engineering circles. But the last 00:27:11.700 --> 00:27:14.799 decade has really seen her get her flowers, so 00:27:14.799 --> 00:27:17.180 to speak. The recognition has been cascading, 00:27:17.180 --> 00:27:20.180 and it's been wonderful to see. In 2003, NASA 00:27:20.180 --> 00:27:23.140 gave her the Exceptional Space Act Award. And 00:27:23.140 --> 00:27:25.579 this is a great detail. It included a check for 00:27:25.579 --> 00:27:30.519 $37 ,200. That seems... Oddly specific. Why that 00:27:30.519 --> 00:27:32.799 amount? At the time, it was the largest individual 00:27:32.799 --> 00:27:35.400 financial award NASA had ever given to anyone 00:27:35.400 --> 00:27:37.319 in its history. It was their way of saying, you 00:27:37.319 --> 00:27:39.200 know, we know you and this is the biggest thank 00:27:39.200 --> 00:27:40.880 you we're allowed to give. That's incredible. 00:27:41.000 --> 00:27:44.220 And then the big one in 2016. The Presidential 00:27:44.220 --> 00:27:46.680 Medal of Freedom, awarded by President Obama, 00:27:47.000 --> 00:27:49.000 he draped it around her neck, and that is the 00:27:49.000 --> 00:27:50.900 highest civilian honor in the United States. 00:27:51.099 --> 00:27:53.720 And the citation explicitly credited her with 00:27:53.720 --> 00:27:55.799 creating the software concepts that allowed for 00:27:55.799 --> 00:27:58.000 the moon landing. She also got into the National 00:27:58.000 --> 00:28:01.000 Aviation Hall of Fame and became a Computer History 00:28:01.000 --> 00:28:03.740 Museum fellow. But I have to ask about the pop 00:28:03.740 --> 00:28:06.079 culture stuff. Because you know you've really 00:28:06.079 --> 00:28:07.940 made it when you're a Lego. Oh, absolutely. The 00:28:07.940 --> 00:28:10.339 Women of NASA Lego set. Yeah. It's fantastic. 00:28:10.460 --> 00:28:13.339 It features her in her 1960s outfit, and it actually 00:28:13.339 --> 00:28:16.319 includes a tiny Lego stack of the code books 00:28:16.319 --> 00:28:19.680 next to her. Recreating that iconic photo. Yes. 00:28:19.799 --> 00:28:22.240 And there was another beautiful tribute by Google 00:28:22.240 --> 00:28:25.440 back in 2019 for the 50th anniversary of the 00:28:25.440 --> 00:28:28.680 landing. They went to the Ivanpah Solar Power 00:28:28.680 --> 00:28:31.039 Facility, you know, that massive array of mirrors 00:28:31.039 --> 00:28:32.900 in the Mojave Desert. The one that looks like 00:28:32.900 --> 00:28:35.420 a sci -fi power plant. That's the one. They programmed 00:28:35.420 --> 00:28:38.579 over 100 ,000 of those mirrors to reflect the 00:28:38.579 --> 00:28:41.680 moonlight, literally moonlight, to create a giant 00:28:41.680 --> 00:28:43.819 portrait of Margaret Hamilton on the landscape 00:28:43.819 --> 00:28:46.460 visible from above. A portrait drawn in moonlight. 00:28:46.539 --> 00:28:48.859 That is poetry. It's perfect. And for the TV 00:28:48.859 --> 00:28:50.759 fans out there, if you watch the show For All 00:28:50.759 --> 00:28:53.460 Mankind on Apple TV plus Bink, the character 00:28:53.460 --> 00:28:55.779 of Margot Madison is heavily inspired by her. 00:28:55.839 --> 00:28:58.720 The glasses, the intense drive, the sheer brilliance. 00:28:59.140 --> 00:29:02.259 But beyond the Legos and the medals, her biggest 00:29:02.259 --> 00:29:04.990 victory, I think, has to be that name. software 00:29:04.990 --> 00:29:08.980 engineering. It really is. In 2018, the IE Software 00:29:08.980 --> 00:29:11.680 Journal celebrated the 50th anniversary of the 00:29:11.680 --> 00:29:14.700 discipline. They were celebrating the very field 00:29:14.700 --> 00:29:17.819 she named. It went from being a joke that hardware 00:29:17.819 --> 00:29:21.039 guys laughed at to being the dominant, most important 00:29:21.039 --> 00:29:23.980 industry of the 21st century. She hired the sheriffs 00:29:23.980 --> 00:29:27.099 for the Wild West. She did. She proved that code 00:29:27.099 --> 00:29:30.000 wasn't magic and it wasn't just typing. She proved 00:29:30.000 --> 00:29:31.680 it was an engineering discipline that could be 00:29:31.680 --> 00:29:34.119 trusted with human lives. So to wrap this all 00:29:34.119 --> 00:29:36.480 up. Margaret Hamilton didn't just write code. 00:29:36.700 --> 00:29:39.200 She created the entire environment, the philosophy, 00:29:39.480 --> 00:29:42.660 where code could survive chaos. From that trick 00:29:42.660 --> 00:29:44.900 Greek and Latin program where she outsmarted 00:29:44.900 --> 00:29:47.099 the hazing ritual to the moment she taught a 00:29:47.099 --> 00:29:49.380 computer to ignore a screaming radar and land 00:29:49.380 --> 00:29:52.319 on the moon, she consistently solved the impossible. 00:29:52.640 --> 00:29:55.559 And she did it by sticking to her guns, by trusting 00:29:55.559 --> 00:29:57.859 her logic and her rigor when the rest of the 00:29:57.859 --> 00:30:00.339 world told her she was just support staff. So 00:30:00.339 --> 00:30:03.259 here is a final thought for you to chew on. We 00:30:03.259 --> 00:30:06.039 live in a world of beta tests and move fast and 00:30:06.039 --> 00:30:08.480 break things. We accept that our apps will crash. 00:30:08.700 --> 00:30:11.380 We accept that our smart fridges might need a 00:30:11.380 --> 00:30:14.140 reboot. Margaret Hamilton's philosophy was that 00:30:14.140 --> 00:30:17.039 errors are not inevitable. They are a failure 00:30:17.039 --> 00:30:21.619 of design. As we build ever more complex AI and 00:30:21.619 --> 00:30:23.940 autonomous systems that will hold our lives in 00:30:23.940 --> 00:30:26.960 their digital hands, why have we moved so far 00:30:26.960 --> 00:30:30.039 away from the failure is not an option? rigor 00:30:30.039 --> 00:30:32.759 of Apollo. Maybe it's time we stopped just fixing 00:30:32.759 --> 00:30:35.299 bugs and started, really started, preventing 00:30:35.299 --> 00:30:37.539 them from ever being born. Development before 00:30:37.539 --> 00:30:39.819 the fact. It's a powerful idea whose time may 00:30:39.819 --> 00:30:41.579 have come again. If you haven't seen that photo 00:30:41.579 --> 00:30:44.000 of her with a stack of code, go Google it right 00:30:44.000 --> 00:30:46.579 now. It is the definition of iconic. Thanks for 00:30:46.579 --> 00:30:48.319 taking this deep dive with us. It was a pleasure. 00:30:48.420 --> 00:30:48.940 See you next time.