Feynman Cracked Safes At Los Alamos — And What It Taught Him About Physics

Ajouté : 2026-04-29

[00:00:00]It's 2:00 in the morning at Los Alamos and I'm sitting on the floor of an office that doesn't belong to me in front of a green steel filing cabinet that holds behind its little nickelplated dial. Every secret about how we built the atomic bomb. The lights are out. Everybody has gone home. I have my ear pressed to the metal. My fingers are turning the dial slowly, slowly. And I'm about to open it. I have no business opening it. I'm a theoretical physicist.

[00:00:31]I'm here to do calculations about neutrons and shock waves and uranium.

[00:00:36]But I'm about to crack this safe anyway.

[00:00:38]And I'm going to crack the eight other safes in this room before sunrise. And what I'm going to learn from these spinning little numbers is going to teach me more about how physics actually works than any equation I ever wrote down. Let me back up. I don't know exactly when locks first got their hooks into me. Maybe it was my sister's diary, the little brass clasp with its little brass key and the fact that she had a key and I did not. And that seemed to me a problem worth solving. Maybe it was the padlocks at school. Maybe it was just the general feeling that anytime somebody puts a closed thing in front of me and tells me I cannot open it, my hands begin twitching. There's something in there. Why shouldn't I know what it is? I was the kind of boy who took apart radios. Not because I wanted to fix them exactly, but because there was a back.

[00:01:29]There was an inside. There was a thing the radio knew that I did not. I would sit on the floor of my room with the cover off, and I would trace each wire with my finger, and I would say to myself, "All right, this thing goes here, and that thing goes there, and now what does this little glass tube actually do?" I drove my mother crazy.

[00:01:51]The radios came back together, more or less. The neighbors started bringing me their broken ones. By 12 years old, I had a reputation in the neighborhood as the boy who fixes radios by thinking, which sounds impressive until you realize that thinking in this case mostly meant looking inside. This habit followed me into physics, and it followed me into Los Alamos. By the time I got there in 1943, this habit of mine had become something more than a hobby. It had become almost a discipline. I was a young man recently arrived at the most secret laboratory in the world, surrounded by some of the smartest people who had ever lived. All of us working on a weapon that could end a war and possibly civilization. The work was incredible. The mountains were beautiful. The food was bad. And the locks, the locks were everywhere. Every office had a filing cabinet. Every filing cabinet had a combination dial.

[00:02:49]Behind each dial was a stack of papers that if photographed and sent to Berlin could have changed the course of human history. The papers contain calculations of critical mass. The papers contain diagrams of implosion lenses. The papers contain the names of people and the schedules of trains and the locations of facilities that were not officially supposed to exist. You'd think the people in charge would have thought hard about those dials. You'd think they would have asked themselves, "How good are these locks really?" You'd think they would have hired a man to break in and check. They did not. They had bought the cabinets from a company called Mosler. The cabinets had little dials with numbers from 0 to 99. You said a combination of three numbers. The cabinet was steel. The cabinet was heavy. The cabinet looked secure. And that for the United States military and the Manhattan project was apparently enough. Now, I want you to imagine being 25 years old, walking down a hallway at this laboratory and noticing that almost every door has one of these cabinets behind it. And imagine that you have over the course of your young life developed a certain itch about closed things. And imagine that the closed things in this case contain the actual blueprints for the atomic bomb. What would you do? I'll tell you what I did.

[00:04:10]I started looking. The first thing I learned, and this is going to sound stupid, but I want you to remember it because everything else flows from this.

[00:04:18]The first thing I learned is that almost nobody had bothered to change the combination from the factory. You know what the factory combination was? 25 0 25 or sometimes 50 25 50. The Mosler company when they shipped the safe set it to one of these little default numbers and the people who received the safe were supposed to change it and they didn't. They didn't because nobody told them to. They didn't because they figured the dial was on the safe and the safe was made of steel and what difference did it make? They didn't because changing the combination on a Mosler is fiddly and you have to remember the new combination. And these were physicists. And physicists already have too many things to remember. So I would walk into an office late at night when the office was empty. I would crouch down in front of the cabinet. I would dial 25 0 25 and try to open it.

[00:05:19]About one time in five, the cabinet opened. The first time it happened, I almost laughed out loud. There I was, a young theoretical physicist, looking into a filing cabinet full of classified documents about uranium enrichment, and the only thing standing between me and those documents had been a number that the Mosler company writes on a card and sticks in every box they ship. I felt embarrassed, not for myself, for the laboratory. I closed the cabinet. I spun the dial. I left, but I started keeping a list. In the back of my head, I had a kind of running tally of which cabinets I had opened, which I had not opened, which combinations I knew. It became a quiet game, a nighttime hobby. I was not stealing anything. I was not photographing anything. I was just satisfying that itch. And I was learning. And as I learned, the game got more interesting because I started to notice something very strange. Most of the cabinets were not on the factory default. Most of the cabinets had been changed, so those were closed to me. And I started thinking about how to get into the changed ones. This is where things got beautiful. I should explain how a combination lock works. You've got a dial. Behind the dial, inside the safe, there are three little discs stacked on top of each other, each one with a notch cut into its edge. When you spin the dial, you spin the first disc. After a certain number of full turns, the first disc picks up the second disc. After another set of turns, the second disc picks up the third. To open the safe, you have to line up all three notches in a row. When all three are lined up, a little metal bar drops into the notches and the bolt slides back and the door opens. That's the whole machine. Three discs, three notches, one bar. That is the entire mechanism that was supposed to protect the secrets of the Manhattan project. So in principle, the combination is three numbers between 0ero and 99, which gives you 1 million possible combinations. If you tried one combination per second, it would take you 11 and a half days working around the clock to try them all. That sounds pretty good. That sounds in fact secure.

[00:07:41]But it's not. And here's the first crack in the security. The first thing I figured out by sitting in front of these cabinets and just watching what happened when I turned the dial. The mechanism doesn't actually require you to land exactly on the number. The notch in each disc has a width. The metal bar that drops into the notch has a width. You can be off by one, sometimes by two. The lock will still open. Mosler designed it this way on purpose because if the tolerance were too tight, then a tiny bit of wear or a tiny bit of dust or a tiny shift in temperature would cause people to be unable to open their own safes. So they built in a margin, a friendly little margin of plus or minus two. Now do the math with me. If you only have to get within two of the right number, then instead of 100 possible numbers per dial, you really only have 100 divided by 520 possible numbers per dial. Three dials, 20 possibilities each. 20 * 20 * 20. That's 8,000 combinations. Not a million. 8,000. At 5 seconds per try, that's about 11 hours.

[00:08:55]11 hours sounds like a lot. But you have weekends and evenings and lunch breaks and nobody's going to walk in on you if you choose your moment. So already just by realizing that the lock is sloppy on purpose, the security has gone from 11 days to two long afternoons. That's a huge change. That's the kind of change that in physics when you find a factor of 100 hiding in a problem nobody's looking at, you should pay attention because something interesting is going on. But I had not even gotten to the good part yet. The good part was this.

[00:09:31]One day I was fooling around with a cabinet. I had legitimate access to my own cabinet in my own office. The cabinet was open. The drawer was pulled out. I was just turning the dial idly.

[00:09:44]The way you fidget with anything when you're thinking about something else.

[00:09:48]And I noticed something. When the cabinet is already open, the bolt is retracted and the discs behind the dial are sitting there unconstrained.

[00:09:57]And as you turn the dial, you can feel them. You can feel the little click when the first disc picks up the second. You can feel the little click when the second picks up the third. And this was the magic moment. If you pull gently on the dial while you turn it, you can feel the dial catch in a particular spot.

[00:10:17]That spot tells you the position of the third disc's notch, which is the third number in the combination. So when a cabinet is open, which during the workday every cabinet at Los Alamos was, you can casually walk by, lean against the cabinet, fiddle with the dial, and walk away knowing the third number of the combination. Stop and think about that for a second. Every single safe at Los Alamos, every single afternoon, was advertising its own combination to anyone who knew how to listen. The third number, anyway. And that was just the beginning, because once I had the third number, I could go further. With the cabinet still open, I could carefully turn the dial back, pulling lightly, and find the second number, too. I just had to be patient. I just had to be willing to spend a few minutes pretending to be lost in thought next to somebody's filing cabinet. I started doing this not constantly, not creepily, but when I was visiting an office on legitimate business, when somebody had stepped out for a moment to fetch coffee or argue with a colleague down the hall, I would lean on their cabinet, and I would learn the last two numbers of their combination, and I would write them down, not where anybody could see, in my own private little notebook of mischief.

[00:11:35]Now, you might be asking, "What good is the last two numbers if you don't know the first?" Here's the answer, and this is where it gets really pretty. If you know the last two numbers, then to open the safe, you only need to find the first number. That's a 100 possibilities. With the tolerance, 20 possibilities. So, you sit down at the safe and you dial the first number, let's say zero, and then you spin to the known second number. And then you spin to the known third number and you try the handle. If it doesn't open, you back the dial up by five and you try again.

[00:12:12]20 tries. Each try takes about 10 seconds. 200 seconds. 3 minutes and 20 seconds. That's how long it takes to crack a Los Alamos safe whose owner has sometime in the last day had the cabinet open while you were in the room. 3 minutes. Now, this discovery, let's pause on this for a second because it's important. This discovery did not come from some genius insight. It did not come from an equation. It came from sitting in front of a cabinet, fiddling with the dial, and paying attention to what my fingers were telling me. The lock was talking. Most people just don't listen. This, by the way, is the closest thing to a religion that I have ever had. The conviction that things will tell you what they are if you stop trusting what people say about them and just look. Just look. Just listen. Just turn the dial and feel where it catches.

[00:13:07]But back to Los Alamos. I had this knowledge. I had this technique. And like any good technique, it was useless unless I tried it on something that mattered. So one day, and I want you to picture this because the scene is delicious. One day I'm in the office of a colleague. He was a young physicist, very sharp, very high up, and he had nine filing cabinets in his office.

[00:13:31]Nine. They contained almost everything important about the most secret weapons project in human history. And he was out of town, and I had been asked to retrieve a particular document from one of his cabinets. The document was, naturally, in a cabinet I could not open.

[00:13:47]A normal person in this situation calls a locksmith or waits. I am not a normal person. I went to my notebook. I checked. I had been in his office before. I had noted some of his last two numbers. So I knew that on at least a couple of his cabinets I had a head start. I sat down on the floor. I started with one of the cabinets where I had the last two numbers. I dialed in zero, then his second, then his third.

[00:14:17]Tried the handle, nothing. Backed up by five, tried again, nothing. Backed up, tried, nothing. Now, you have to understand the feeling here, sitting on the floor in another man's office, in the middle of the most secret laboratory in the world, in a war, slowly turning the dial of a safe. Each click, each pull, each silence, there's a kind of trance that comes over you. The world shrinks. You become the dial. The pencil in your hand goes still. You can hear the wind outside through the thin desert walls. You can smell the metal of the cabinet. The faint old paper smell leaking out of the seam in the door.

[00:15:01]About 13 minutes in, the handle moved. I opened the cabinet. The papers I needed were not in this cabinet. So I closed it, spun the dial, moved to the next, sat down, started over, new combination, new little ritual of dial and listen and pull. About 10 minutes later, that one opened, too. I went through nine cabinets that afternoon. Some I had the last two numbers for, some I didn't, and I had to do the longer brute force version. By the time I was done, I had every single one of his filing cabinets standing open. The accumulated paperwork of the Manhattan project basically open, sitting there like an anatomy lecture. I found the document I needed. Then I did something which looking back I cannot fully justify but which felt right at the time. I took a piece of paper. I wrote on it. The note said something like, "This was an easy one, borrowed by a friend." and I left the note inside one of the cabinets where the owner would find it when he came back. Then I closed the cabinets I had opened. I spun every dial. I left the office tidier than I found it. And I want to tell you about the reaction because the reaction is the lesson. The reaction was not, "Oh, this is interesting. Our security has a flaw."

[00:16:24]The reaction was not, "Well, let's call Mosler and tell them their tolerances are too generous." The reaction was not, "Let's bring in this young fineman and have him show us systematically what he can do because clearly we have been thinking about this wrong." The reaction was, "That crazy man got into my safe.

[00:16:44]Lock him out of my office. Tell him to stop, make a memo, issue a directive, inform the security officers, do not let Richard Fineman near classified material." And the laboratory eventually did exactly that. Memos went around.

[00:17:00]Combinations were changed. Doors that had been open to me began to close. And here's the funniest thing of all. When people changed their combinations, what did they change them to? They changed them to their wedding anniversaries, to their children's birthdays, to the date they got their doctorate, to meaningful, memorable, deeply personal numbers. Numbers that given a few minutes of conversation with a man over lunch, you could guess. I'm not going to tell you whether I tried, but I want you to sit with that for a moment. I want you to sit with the fact that the response to a discovered security flaw by some of the smartest people on Earth was not to fix the flaw.

[00:17:40]The response was to perform the appearance of security harder. Now, here's the part of the story I really want you to pay attention to, because this is where it stops being a funny anecdote about a young man's mischief and starts being something deeper. One day a man came to Los Alamos to give a lecture about safes. A real safe expert.

[00:18:01]A man whose entire profession was the design and the cracking of safes. This was buil as a security training. We were all supposed to attend and I attended because of course I attended. I had been waiting my whole career for this lecture. He stood up. He talked for an hour. He told stories. He showed pictures of famous safes. He talked about hardened steel and relockers and time delays and all sorts of impressive sounding things. And I was sitting in the front row taking notes, my heart pounding, because I thought, finally, finally, I am going to learn the things I have been figuring out on my own. The professional is going to tell me the secrets of his craft. The factory defaults, the last two numbers, the tolerances. He's going to tell me things I have not figured out yet. After the lecture, I went up to him. I said, "That was wonderful, sir. May I ask you a question? When you have to open a Mosler combination safe and you don't know the combination and you don't have the manufacturer's records, how do you actually open it? What is your method?"

[00:19:10]He looked at me. He thought for a moment, and then he said, "Well, you know, I just try a lot of combinations."

[00:19:16]That was his answer. That was the answer of the master safecracker. He didn't know about the factory default. He didn't know about the last two numbers.

[00:19:26]He didn't know about the tolerance. He had given a 1-hour lecture full of impressive vocabulary about safes. And when I asked him the most basic possible question about his profession, he had no method. He just tried things. He brute forced. He used patience and luck and a vague sense for which numbers people pick. I want you to feel the disappointment I felt in that moment because that disappointment was the beginning of an education that has lasted my whole life. The expert, the man whose job it was to know these locks. He did not know how they worked.

[00:20:00]He had a position. He had a salary. He had a title. He had been flown across the country to teach the Manhattan Project about security. And he did not know. I wish I could tell you this is rare. It is not rare. I have met more experts who do not understand their own subject than experts who do. The astronomer who does not really know how a telescope mirror is figured. The biologist who does not really know how a centrifuge separates molecules. The economist who does not really know how the bank he writes about clears a check.

[00:20:35]The physicist, and I include myself here, I have made this mistake hundreds of times. the physicist who uses an equation for years before suddenly realizing he never quite figured out where it comes from. So what does any of this have to do with physics? I promised you at the beginning that the safes were going to teach me how physics actually works. Let me deliver on that promise.

[00:20:58]In physics, we work with two kinds of things. We work with the appearances of things and we work with the mechanisms of things. The appearance is what the thing looks like. what it sounds like, what people say about it. The mechanism is what's actually going on inside, the gears and the springs and the discs turning behind the dial. Most people, including most scientists, including some of the best scientists in the world, spend most of their time on appearances. They take the textbook's word for it. They take the famous physicist's word for it. They take the equation and they shuffle the symbols around and they get an answer. and the answer agrees with what they were told and they move on. But every now and then somebody sits down in front of the safe and turns the dial slowly and listens.

[00:21:48]And what they discover is that the appearance and the mechanism are not the same thing. The appearance was a million combinations. The mechanism was 8,000.

[00:21:58]The appearance was secret blueprints behind locked doors. The mechanism was a default factory setting plus a tolerance of two plus a dial that whispered its own secrets to anybody patient enough to lean against it. And once you see that gap between appearance and mechanism, you cannot unsee it. You see it everywhere. Take an example from physics. For hundreds of years, people thought heavy things fell faster than light things. That was the appearance. A feather falls slow. A stone falls fast.

[00:22:31]End of story. And people taught it that way and authoritative books said so and the appearance held up until somebody Galileo or one of his rivals or all of them together. The history is a little messy. Until somebody actually sat down and did the experiment with proper care and the appearance broke. The mechanism turned out to be different. The mechanism was in a vacuum. The feather and the stone fall the same. The heaviness causes speed thing was air resistance dressed up in fancy clothes.

[00:23:06]It was a factory default combination on the universe and Galileo cracked it. Now ask yourself, how many other things do we currently believe because of appearances and not because we sat down with a dial and listened? Take another example, the atom. For a long time, people thought atoms were like tiny solar systems, little electron planets orbiting a nucleus sun. That was the picture in every textbook. And the picture made sense, and people drew it on blackboards. It's a beautiful picture. It's also wrong. The actual mechanism is something much stranger with electrons that don't have positions in the way planets do and clouds of probability and quantum behavior that breaks every analogy you bring to it.

[00:23:50]The appearance was a solar system. The mechanism was well, we are still arguing about exactly what the mechanism is. And I am not going to pretend we have it all figured out. But it is not a solar system. The solar system was the factory default combination.

[00:24:08]Convenient, memorable, and basically wrong. Take a third example. Heat. For a long time, people thought heat was a fluid. a magical weightless fluid called caloric which flowed from hot things to cold things. That was the appearance.

[00:24:26]Heat moves. Therefore, there's a stuff that's moving. Therefore, there's a fluid. The whole 18th century built theories on this. They were wrong. The mechanism was heat is just the jiggling of atoms. There's no fluid. There's just motion, statistical motion. Billions of little hits and bumps. And the moment somebody sat down with the dial and listened, the whole appearance crumbled and the mechanism replaced it. This pattern, which I learned by sitting in front of a Mosler safe in the middle of the night, is the pattern of physics. It is the pattern of every real discovery I have ever seen. It is the pattern of every scientific revolution I have ever read about. The appearance is the cover story. The mechanism is the secret. And the entire job of a physicist is to crack the safe. And here is the deepest thing. The thing the safes really taught me, the thing I want you to walk away with. Nature is sloppy. Nature, like the Mosler company, has built tolerances into her mechanisms. Nature has factory defaults. Nature has back doors that nobody is using because nobody is leaning on the cabinet long enough to feel where the dial catches. Every time we think we have figured everything out, somebody comes along and finds a back door that has been there the whole time.

[00:25:48]Quantum mechanics is a backdoor.

[00:25:50]Relativity is a backdoor. The discovery that the continents move was a backdoor.

[00:25:55]The discovery that the same equations describe an electron in a copper wire and a star 10,000 lighty years away was a back door. They were sitting there in plain view, advertising themselves to anyone who was patient enough to listen.

[00:26:10]We just hadn't listened. Why is nature sloppy? Why does she have these back doors? Honestly, I don't know. I have a guess. My guess is that the universe runs on a small number of simple rules.

[00:26:23]And any system that runs on simple rules has fewer hiding places than it appears to have. The space of possibilities looks vast, but the space of actual possibilities given the rules is small.

[00:26:36]The Mosler safe looks like a million combinations. It's 8,000.

[00:26:41]The universe looks like infinite mystery. It's a few dozen good ideas applied to many situations. I'll tell you something else. After I had become known around the lab as the man who could open any safe, and let me say plainly, this reputation got out of hand. I was not as good as the legend made me sound. After that, people would sometimes ask me to open a safe whose owner had gone away or forgotten his combination.

[00:27:08]And one day somebody asked me to open a safe whose owner had been working on something extremely sensitive and the owner was not available and the lab needed the contents. I sat down. I tried my tricks. None of them worked. The combination was not the factory default.

[00:27:26]I could not pull the last two numbers off the dial. And I did not have the time to brute force 8,000 possibilities.

[00:27:34]I sat there for an hour beaten. I went home. I came back the next day and then I tried something stupid. I tried the man's birthday. The handle moved. This was the last lesson and maybe the most important. People are predictable. We are creatures of pattern even when we think we are being clever. We like our birthdays. We like our anniversaries. We like the years our children were born.

[00:28:01]We pick combinations that mean something to us because we don't want to forget them. And we forget that the people trying to crack our safes can read the same calendar we can. Nature, by the way, is also like this. Nature has favorite numbers. Pi shows up everywhere. So does the speed of light, 186,000 m/s.

[00:28:21]So does Planck's constant 6.6 26 * 10us 34 in units we won't worry about. The universe has anniversaries it can't help but use. If you find an equation and the same number keeps showing up, nature is telling you something. Lean on the cabinet. Listen, I want to leave you with one last image. The image is me, 25 years old, sitting on the floor of an office at Los Alamos in front of a green steel filing cabinet in the middle of the night. The lights are out. The cabinet contains the actual blueprints for the atomic bomb. My ear is pressed to the metal. My fingers are turning the dial slowly, slowly. Each click, each silence, each tiny resistance of the mechanism telling me something I am not supposed to know. And here is what I want you to understand. The position I was in in that office in front of that cabinet. That is the position every physicist is always in. That is the position every scientist is always in.

[00:29:27]That is the position every honest, curious person is always in. The world is the safe. The dial is right there in front of you. The combination is unknown. Most people walk past the cabinet and assume it is locked. Most people accept the appearance and never question the mechanism. Most people, when somebody else cracks the safe, prefer to ban the cracker rather than admit the lock was bad. Most people when an expert tells them how something works write it down without checking. Most people when offered the factory default never consider that the factory might be lazy. Don't be most people. Sit down.

[00:30:07]Press your ear against the cold metal.

[00:30:09]Turn the dial slowly. Listen for the click. The whole universe is waiting to tell you its combination. And most of it is just the last two numbers and a little patience. So tell me, the next time you find yourself in front of a closed thing in your own life, a closed problem, a closed idea, an institution everyone says is too solid to question.

[00:30:32]What is the first number you're going to try?