WEBVTT 00:00:22.239 --> 00:00:25.519 You are walking home without an umbrella. 00:00:25.519 --> 00:00:31.359 The forecast said 60%, and you did the math the way everybody does the math. 00:00:31.359 --> 00:00:33.840 Which is to say that you didn't. 00:00:33.840 --> 00:00:38.240 You looked at the sky and you decided that it would hold. 00:00:38.240 --> 00:00:41.280 So now you are hurrying. 00:00:41.280 --> 00:00:47.600 Shoulders up, and the rain is coming down in big, scattered droplets. 00:00:47.600 --> 00:00:50.960 You watch it land on the sidewalk. 00:00:50.960 --> 00:00:52.640 One drop. 00:00:52.640 --> 00:00:53.200 Then a dozen. 00:00:54.079 --> 00:00:58.960 Dark coins appearing on the pale concrete in no particular order. 00:00:58.960 --> 00:01:03.759 You could not say where the next one will fall. 00:01:03.759 --> 00:01:05.760 Nobody could. 00:01:05.760 --> 00:01:07.280 Not you. 00:01:07.280 --> 00:01:10.400 Not whoever wrote the forecast. 00:01:10.400 --> 00:01:13.280 Not anyone who has ever lived. 00:01:13.280 --> 00:01:16.079 And that, right there. 00:01:16.079 --> 00:01:18.480 The randomness. 00:01:18.480 --> 00:01:22.079 That is one of the last wild things. 00:01:22.079 --> 00:01:24.079 And it's everywhere. 00:01:24.079 --> 00:01:27.200 It is the exact fork of the lightning. 00:01:27.200 --> 00:01:30.719 Drawn once across the dark and never again. 00:01:30.719 --> 00:01:33.519 It is the freckles on your arm. 00:01:33.519 --> 00:01:37.439 It is the hiss between two radio stations. 00:01:37.439 --> 00:01:42.879 The world throws off randomness the way it throws off heat. 00:01:42.879 --> 00:01:44.719 Constantly. 00:01:44.719 --> 00:01:46.879 For free. 00:01:46.879 --> 00:01:49.280 You cannot bottle it. 00:01:49.280 --> 00:01:52.719 You cannot ask the storm to wait. 00:01:52.719 --> 00:01:56.319 It falls and the randomness is gone. 00:01:56.319 --> 00:02:00.719 Which would not matter at all if we didn't need it. 00:02:00.719 --> 00:02:04.000 But we do need it. 00:02:04.000 --> 00:02:08.159 We need it because it is how we try to see the future. 00:02:08.159 --> 00:02:14.639 To determine whether the river comes over its banks tonight while you sleep. 00:02:14.639 --> 00:02:20.639 Whether the funnel touches down on your street or one in the next town over. 00:02:20.639 --> 00:02:26.080 To see the future, we build a little sky inside a machine. 00:02:26.080 --> 00:02:32.639 We tell the sky everything we know about the air right now and about physics. 00:02:32.639 --> 00:02:37.280 And then we let it run the sky forward, out ahead of the real weather. 00:02:37.280 --> 00:02:40.639 But the air will not sit still to be copied. 00:02:40.639 --> 00:02:50.319 Miss one gust of wind, one pocket of warmth too faint to measure, and your little digital sky evolves in the completely wrong direction. 00:02:50.319 --> 00:02:54.159 So we do not run the simulation once. 00:02:54.159 --> 00:02:56.479 We run it 50 times. 00:02:56.479 --> 00:03:04.240 And each run starts with a tiny random nudge to stand in for everything we don't know. 00:03:04.240 --> 00:03:08.080 Each nudge creates a slightly different tomorrow. 00:03:08.080 --> 00:03:12.479 Then we count how many of those tomorrows end in rain. 00:03:12.479 --> 00:03:17.120 And here is the thing about the machine dreaming up those tomorrows. 00:03:17.120 --> 00:03:22.800 It cannot make a single one of those random nudges on its own. 00:03:22.800 --> 00:03:27.280 We had to find the randomness somewhere and pour it in. 00:03:27.280 --> 00:03:36.159 And the randomness we pour into nearly every forecast, every simulated flood, every imagined storm. 00:03:36.159 --> 00:03:38.960 It is not real. 00:03:38.960 --> 00:03:41.840 It's a counterfeit. 00:03:41.840 --> 00:03:44.000 We faked it. 00:03:44.000 --> 00:03:48.960 I'm Daina Bouquin, and this is Found in the M achine. 00:03:49.840 --> 00:04:02.240 To you understand why anyone would fake randomness you have to understand how hard the real thing is to catch. 00:04:02.240 --> 00:04:04.639 Go back to 1947. 00:04:04.639 --> 00:04:11.680 To a low building in Santa Monica, close enough to the ocean that you could smell it if you opened a window. 00:04:11.680 --> 00:04:17.199 Where a new outfit called RAND was doing mathematics for the Air Force. 00:04:17.199 --> 00:04:20.959 Someone there needed random numbers. 00:04:20.959 --> 00:04:23.199 Not a handful. 00:04:23.199 --> 00:04:25.040 A million. 00:04:25.040 --> 00:04:27.279 Clean ones. 00:04:27.279 --> 00:04:29.839 With no human thumb on the scale. 00:04:29.839 --> 00:04:34.319 Because you could never get a million random numbers out of a person. 00:04:34.319 --> 00:04:40.879 Ask anyone to rattle off random digits and they will lean on some without realizing it. 00:04:40.879 --> 00:04:44.160 They'll make little patterns without meaning to. 00:04:44.160 --> 00:04:46.800 They'll say seven too much. 00:04:46.800 --> 00:04:50.480 People are pattern machines. 00:04:50.480 --> 00:04:54.160 It is one of the things we cannot stop ourselves from being. 00:04:54.160 --> 00:04:56.879 And the people at RAND knew that. 00:04:56.879 --> 00:05:00.319 So they built a chaos engine instead. 00:05:00.319 --> 00:05:05.839 It was, more or less, an electronic roulette wheel. 00:05:05.839 --> 00:05:10.959 A wheel with 32 slots spun by raw electrical noise. 00:05:10.959 --> 00:05:20.079 The frantic static inside a gas tube, whipping through thousands of revolutions and landing on one number every second. 00:05:20.079 --> 00:05:22.240 It clattered. 00:05:22.240 --> 00:05:23.519 It clicked. 00:05:23.519 --> 00:05:30.160 It punched its chaos into stiff paper cards hour after hour, day after day. 00:05:30.160 --> 00:05:31.839 And it worked. 00:05:31.839 --> 00:05:33.439 Sort of. 00:05:33.439 --> 00:05:38.720 Because it turns out that building a thing with no habits at all is almost impossible. 00:05:38.720 --> 00:05:40.879 The tubes degraded. 00:05:40.879 --> 00:05:43.920 They were sensitive to the room's temperature. 00:05:43.920 --> 00:05:48.480 They felt the grid's voltage flutter when the building's power dipped. 00:05:48.480 --> 00:05:55.439 And slowly, quietly, a faint shadow of a pattern would creep back in. 00:05:55.439 --> 00:06:00.879 So you would have to shut the machine down and tune it, and then it would behave for a while. 00:06:00.879 --> 00:06:03.439 And then it would drift again. 00:06:03.439 --> 00:06:06.160 Which meant someone had to check. 00:06:06.160 --> 00:06:11.439 Someone had to take a million numbers and hunt through them for ghosts of a pattern. 00:06:11.439 --> 00:06:15.839 Her name was Bernice Brown. 00:06:15.839 --> 00:06:24.800 Before Rand, she had been in Ames, Iowa, at the statistics program that became the first statistical laboratory in the country. 00:06:24.800 --> 00:06:31.120 In 1932, she became the second person ever to earn a master's degree in statistics there. 00:06:31.120 --> 00:06:35.920 At RAND, once the machine had spat out its million digits. 00:06:35.920 --> 00:06:40.800 Her job was to make sure they were completely meaningless. 00:06:40.800 --> 00:06:45.360 So she counted how often each digit appeared. 00:06:45.360 --> 00:07:04.560 Whether pairs showed up too often, whether runs ran too long, test after test, by hand and on IBM tabulating machines, feeding in tray after heavy tray of punched cards, day after day, for years. 00:07:04.560 --> 00:07:09.279 Tedious, physical, invisible work. 00:07:09.279 --> 00:07:22.160 And at the end of it all, after tests and retests and one final mathematical shuffle to scrub out the machine's last stubborn leanings, Bernice Brown signed off. 00:07:22.160 --> 00:07:28.879 She wrote that nothing in her tests argued against calling these numbers random. 00:07:28.879 --> 00:07:32.079 Not these are random. 00:07:32.079 --> 00:07:34.560 No one can ever prove that. 00:07:34.560 --> 00:07:42.000 Only I looked as hard as I know how to look and I could not catch them being anything else. 00:07:42.000 --> 00:07:47.360 Her certified numbers were eventually printed and bound into a book. 00:07:47.360 --> 00:07:53.360 A real book published in 1955 that you could order in the mail. 00:07:53.360 --> 00:07:59.199 It is called A Million Random Digits with 100,000 Normal Deviates. 00:07:59.199 --> 00:08:02.560 You can still buy a copy today. 00:08:02.560 --> 00:08:04.800 And people did buy it. 00:08:04.800 --> 00:08:08.959 Scientists, pollsters, lottery designers. 00:08:08.959 --> 00:08:13.759 For years, if you needed real randomness, you did not make it. 00:08:13.759 --> 00:08:15.439 You looked it up. 00:08:15.439 --> 00:08:18.720 The book even came with instructions. 00:08:18.720 --> 00:08:21.600 You were not supposed to start on page one. 00:08:21.600 --> 00:08:28.079 You were told to open the book somewhere unpredictable and read off in an unpredictable direction. 00:08:28.079 --> 00:08:33.279 Even the Book of Chaos needed you to bring a little chaos of your own. 00:08:33.279 --> 00:08:37.919 So this is what randomness cost in 1955. 00:08:37.919 --> 00:08:40.480 A machine that fought you. 00:08:40.480 --> 00:08:43.759 A woman who spent years checking every digit. 00:08:43.759 --> 00:08:47.840 A book of certified nonsense that you kept on a shelf. 00:08:47.840 --> 00:08:53.919 Now, around that same time, a man was lying in a hospital bed playing cards. 00:08:53.919 --> 00:09:00.559 His name was Stanisław Ulam, and he was a Polish-American mathematician. 00:09:00.559 --> 00:09:04.240 In 1946, he had nearly died. 00:09:04.240 --> 00:09:08.480 A sudden encephalitis, an emergency brain surgery. 00:09:08.480 --> 00:09:12.799 And as he recovered, he played solitaire. 00:09:12.799 --> 00:09:16.480 And somewhere in there he got curious. 00:09:16.480 --> 00:09:20.639 What are the odds that a given hand of solitaire can be won at all? 00:09:20.639 --> 00:09:25.200 He tried to work it out properly with pure mathematics. 00:09:25.200 --> 00:09:28.879 And the calculations just ballooned beyond reach. 00:09:28.879 --> 00:09:32.399 So he gave up on elegance and just played. 00:09:32.399 --> 00:09:34.320 Lots of hands. 00:09:34.320 --> 00:09:37.200 And he counted how often he won. 00:09:37.200 --> 00:09:40.080 He figured, don't solve the problem. 00:09:40.080 --> 00:09:42.240 Simulate it. 00:09:42.240 --> 00:09:48.480 Run it over and over with the deck shuffled differently each time, and let the count become your answer. 00:09:48.480 --> 00:10:11.360 Ulam brought that idea back to his colleagues and one of them saw immediately what it could do. His name was John von Neumann, a mathematician of terrifying range who played German marching music too loud in his office, and who was, at that moment, trying to understand what happens inside an exploding bomb. 00:10:11.360 --> 00:10:14.480 You could write that problem down in theory. 00:10:14.480 --> 00:10:20.799 Every neutron, every trajectory through a chunk of metal the size of a grapefruit. 00:10:20.799 --> 00:10:25.360 But solving it would take longer than the universe has been alive. 00:10:25.360 --> 00:10:28.879 Or you could do what Ulam had done with the cards. 00:10:28.879 --> 00:10:31.759 Follow one imaginary neutron. 00:10:31.759 --> 00:10:34.799 When it strikes a nucleus, roll the dice. 00:10:34.799 --> 00:10:36.720 Does it bounce? 00:10:36.720 --> 00:10:38.480 Is it absorbed? 00:10:38.480 --> 00:10:40.559 Does it split the atom? 00:10:40.559 --> 00:10:44.720 That random roll of the dice would determine the outcome. 00:10:44.720 --> 00:10:48.720 And von Neumann needed millions of rolls. 00:10:48.720 --> 00:10:51.120 More than any book could hold. 00:10:51.120 --> 00:10:53.279 So he came up with a trick. 00:10:53.279 --> 00:11:00.080 Take a number, square it, then reach into the middle of the answer and pull out a few digits. 00:11:00.080 --> 00:11:02.399 That's your next number. 00:11:02.399 --> 00:11:05.440 Now do it again and again. 00:11:05.440 --> 00:11:09.200 Plain, fast, repeatable arithmetic. 00:11:09.200 --> 00:11:13.279 And if you squint, the numbers it throws off look random. 00:11:13.279 --> 00:11:16.320 They jump around and feel unpredictable. 00:11:16.320 --> 00:11:18.399 They are not. 00:11:18.399 --> 00:11:20.639 Of course they are not. 00:11:20.639 --> 00:11:24.960 Feed the trick the same starting number, the same seed. 00:11:24.960 --> 00:11:27.919 And it hands you the exact same sequence. 00:11:27.919 --> 00:11:31.360 There is no more chance in it than in a times table. 00:11:31.360 --> 00:11:35.200 And nobody knew that better than von Neumann. 00:11:35.200 --> 00:11:37.360 He said it himself. 00:11:37.360 --> 00:11:45.279 Anyone who uses arithmetic to produce random digits is, of course, in a state of sin. 00:11:45.279 --> 00:11:49.919 For decades, the fakes were good enough. 00:11:49.919 --> 00:11:54.879 Because for most of what we ask randomness to do, a good fake is fine. 00:11:54.879 --> 00:12:04.080 For weather, for bombs, for the slow patient work of science, von Neumann's sin was pretty much good enough. 00:12:04.080 --> 00:12:08.320 But then we started asking randomness to do something new. 00:12:08.320 --> 00:12:14.080 Not to imagine the future, but to keep our secrets. 00:12:14.080 --> 00:12:25.279 You see, when you send a message you do not want read, or when money moves, the locks that protect all of it are built from random numbers. 00:12:25.279 --> 00:12:30.480 And a secret built on a fake is just a secret waiting for the right person to notice. 00:12:30.480 --> 00:12:36.480 Like in 1995, two graduate students at Berkeley noticed. 00:12:36.480 --> 00:12:49.919 They pulled apart the web browser Netscape, which was at the time the front door to the internet, and found that the quote, random numbers securing its traffic, were seeded with things like the time of day. 00:12:49.919 --> 00:12:53.120 Not chaos, just a clock. 00:12:53.120 --> 00:12:56.240 And so in minutes the locks came open. 00:12:56.240 --> 00:13:04.399 And so, after half a century of getting away with it, we discovered that we still needed the real thing. 00:13:04.399 --> 00:13:09.360 Genuine unpredictability, caught from the physical world. 00:13:09.360 --> 00:13:11.519 Bernice Brown's kind. 00:13:11.519 --> 00:13:15.360 Computer science has a word for it. 00:13:15.360 --> 00:13:18.159 They call it entropy. 00:13:18.159 --> 00:13:21.840 And we were sent back out to hunt for it. 00:13:21.840 --> 00:13:25.360 Here is what we worked out. 00:13:25.360 --> 00:13:30.639 To keep our secrets, we do not need a million real random numbers. 00:13:30.639 --> 00:13:34.960 We only need the starting points, the seeds. 00:13:34.960 --> 00:13:41.840 Once you have a truly unpredictable seed, the fast, cheap arithmetic can take it from there. 00:13:41.840 --> 00:13:44.879 So the hunt became a hunt for seeds. 00:13:44.879 --> 00:14:01.600 And once you go looking the mess is everywhere, the thermal noise inside a resistor, the decay of a radioactive atom, the jitter of the exact microsecond you happen to move your mouse. 00:14:01.600 --> 00:14:10.960 Your computer is scavenging entropy off of you right now, gathering up your tiny accidents and pooling them like rainwater. 00:14:10.960 --> 00:14:20.559 And in one building in San Francisco, a company called Cloudflare keeps a whole wall of lava lamps. 00:14:20.559 --> 00:14:26.799 Blobs rising and folding and splitting and never taking the exact same shape twice. 00:14:26.799 --> 00:14:39.200 A camera watches the wall, and a computer turns every pixel, its position, brightness, and color, into numbers. 00:14:39.200 --> 00:14:49.279 The exact posture of every slow molten blob becomes a seed, feeding the cryptographic keys that secure web traffic for millions of sites. 00:14:49.279 --> 00:14:55.679 If you walked in front of that wall, your shadow would shape the randomness too. 00:14:55.679 --> 00:15:04.240 Some tiny fraction of the internet's locks would be forged in part from the outline of you. 00:15:04.240 --> 00:15:12.879 We built machines that can do almost anything. 00:15:12.879 --> 00:15:17.200 They can play out fifty tomorrows no problem. 00:15:17.200 --> 00:15:21.279 They can follow a neutron through a bomb. 00:15:21.279 --> 00:15:24.399 And lately they can talk. 00:15:24.399 --> 00:15:27.840 You may have had a conversation with one recently. 00:15:27.840 --> 00:15:34.240 Every time one of those machines chooses its next word, it rolls dice. 00:15:34.240 --> 00:15:37.440 And they are von Neumann's dice. 00:15:37.440 --> 00:15:44.639 Fed at the very start by a seed of real entropy scavenged from the physical world. 00:15:44.639 --> 00:15:48.399 Because that is the one thing no machine can do. 00:15:48.399 --> 00:15:53.840 The most powerful things we have ever built come to the physical world like beggars. 00:15:53.840 --> 00:15:55.759 To the static. 00:15:55.759 --> 00:15:57.840 To the lava. 00:15:57.840 --> 00:16:02.240 The machines cannot make it. 00:16:02.240 --> 00:16:06.799 Bernice Brown could spend years checking and never prove it. 00:16:06.799 --> 00:16:12.799 Von Neumann could only fake it, and the sky is giving it away for free. 00:16:12.799 --> 00:16:18.000 You feel it on the back of your neck as you hurry home. 00:16:18.000 --> 00:16:25.440 I'm Daina Bouquin, and this is Found in the Machine. 00:16:25.440 --> 00:16:33.440 If you enjoy these stories, please rate and review this podcast wherever you listen.