WEBVTT - Fritterin' Away Genius (Classic) 0:00:15.410 --> 0:00:24.650 Pushkin. Hello everyone, Happy New Year. The Cautionary Tales team 0:00:24.730 --> 0:00:28.530 is busy putting together a set of brand new cautionary 0:00:28.570 --> 0:00:32.450 tales for the year ahead. Happiness Cults, a race around 0:00:32.450 --> 0:00:35.410 the world, and some lessons from the front lines of 0:00:35.490 --> 0:00:39.770 finding love and much more, all on its way. In 0:00:39.810 --> 0:00:43.050 the meantime, don't forget We've got a Cautionary Club over 0:00:43.090 --> 0:00:46.370 on Patreon, where you can find loads of bonus content, 0:00:46.570 --> 0:00:49.570 extra cautionary tales, interviews with me and the team, and 0:00:49.770 --> 0:00:54.290 an exclusive newsletter. Now, while you're setting your sights on 0:00:54.410 --> 0:00:58.330 what twenty twenty six could look like, here's a cautionary 0:00:58.370 --> 0:01:06.810 tale from the archives which might help enjoy. It would 0:01:06.850 --> 0:01:09.170 be hard to think of a better example of a 0:01:09.290 --> 0:01:15.090 game of chance than Roulette. Beneath the romantic French terminology 0:01:15.250 --> 0:01:18.130 and the myriad rules of etiquette, each spin of the 0:01:18.170 --> 0:01:22.890 roulette wheel is utterly random. The casino's advantage is small 0:01:23.530 --> 0:01:29.170 that it cannot be overcome. The game is remorseless over 0:01:29.210 --> 0:01:32.530 the long haul. The only way to win is not 0:01:32.650 --> 0:01:37.810 to play, or is it. One day in August nineteen 0:01:37.890 --> 0:01:41.650 sixty one, Claude and Betty Shannon stroll up to a 0:01:41.730 --> 0:01:45.330 roulette table in Las Vegas, pretending not to know their companions, 0:01:45.890 --> 0:01:49.610 Ed and Vivian Thorpe. Claude and the ladies are nervous, 0:01:50.010 --> 0:01:54.250 but they don't show it. Ed Thorpe isn't nervous, he's excited. 0:01:54.930 --> 0:01:57.530 He's still in his twenties, but he's an old hand 0:01:57.530 --> 0:02:01.930 and the casinos Claude. Shannon stands right by the wheel. 0:02:02.530 --> 0:02:05.490 He's forty five years old, slim and good looking, with 0:02:05.610 --> 0:02:10.490 fine cheekbones and dark eyebrows. He's directing the attention of 0:02:10.530 --> 0:02:13.810 the floor manager by scribbling down numbers. He looks like 0:02:13.850 --> 0:02:17.810 he's got some crazy system that will inevitably bankrupt him. 0:02:18.210 --> 0:02:20.610 Thorpe is at the other end of the table, far 0:02:20.650 --> 0:02:24.050 from the wheel and far from Shannon. He has dark hair, 0:02:24.290 --> 0:02:28.570 a round face, and a smile. He's having fun placing 0:02:28.570 --> 0:02:31.010 his bets with the confidence of a man who knows 0:02:31.050 --> 0:02:35.970 the unbeatable game is about to be beaten. This is 0:02:36.010 --> 0:02:39.170 a defining moment in a project that has been quietly 0:02:39.250 --> 0:02:42.850 ticking over for a year. When it began, Thorpe and 0:02:42.930 --> 0:02:46.770 Shannon didn't know each other. Edward O. Thorpe was a 0:02:46.850 --> 0:02:52.330 junior mathematics instructor at MIT. Claude, Shannon was the greatest 0:02:52.330 --> 0:02:56.410 computer scientist in the world. Ed Thorpe had a plan 0:02:56.530 --> 0:03:01.330 to beat Roulette, and he needed Shannon to help him. 0:03:01.570 --> 0:03:05.170 Systems to beat Roulette are like blueprints for perpetual motion 0:03:05.330 --> 0:03:10.010 machines or formulas to turn lead into gold. Third, the 0:03:10.010 --> 0:03:14.570 pseudo scientific obsessions of Cranks and Claude. Shannon's secretary had 0:03:14.610 --> 0:03:18.690 already warned Thorpe that Professor Shannon doesn't spend time on 0:03:18.770 --> 0:03:22.530 topics or people that don't interest him. Shannon was a 0:03:22.650 --> 0:03:26.530 legendary figure. People in his field talked about Shannon the 0:03:26.570 --> 0:03:30.930 way physicists talk about Albert Einstein. What Ed Thorpe was 0:03:30.970 --> 0:03:35.170 doing was much like buttonholing Einstein and saying, Hey, Albert, 0:03:35.330 --> 0:03:37.650 I've got a surefire scheme for beating the bookies at 0:03:37.690 --> 0:03:43.290 the racetrack. An unknown young mathematician a patently futile goal. 0:03:43.810 --> 0:03:50.050 Claude Shannon, the computing legend, didn't hesitate. Take a seat. 0:03:50.490 --> 0:03:53.370 He said to Ed Thorpe, we have a lot to 0:03:53.450 --> 0:04:24.450 talk about. I'm Tim Harford, and you're listening to cautionary tales. Repeat. 0:04:24.490 --> 0:04:26.730 Please please send floor for the present. 0:04:27.010 --> 0:04:29.890 How do you receive send flower? Please see if you 0:04:29.930 --> 0:04:30.530 can read this? 0:04:30.930 --> 0:04:34.010 Can you read this, Yes, how are signals? 0:04:34.410 --> 0:04:38.210 Do you receive? Please send something? Please send thees and bees? 0:04:38.530 --> 0:04:39.570 How are signals? 0:04:40.690 --> 0:04:45.930 Those messages from eighteen fifty eight represent a full day 0:04:46.290 --> 0:04:50.410 of attempted conversation via Morse code. Who are cable? Lying 0:04:50.570 --> 0:04:54.370 three miles under the surface of the Atlantic Ocean. The 0:04:54.450 --> 0:04:58.850 cable had been enormously expensive, and as you might guess, 0:04:59.370 --> 0:05:04.050 it wasn't really working. In an attempt to boost the signal, 0:05:04.290 --> 0:05:08.570 the project's engineer, a man called Wildman white House, cranked 0:05:08.650 --> 0:05:14.370 up the vault. The cable melted. It had survived only 0:05:14.490 --> 0:05:19.690 twenty eight days. Over the years, telegraph engineers figured out 0:05:19.810 --> 0:05:22.410 how to work around the problem of noise on the line. 0:05:22.570 --> 0:05:26.250 They built stronger cables with better insulation and more sensitive 0:05:26.290 --> 0:05:30.250 detectures at the far end. But nobody fully solved the 0:05:30.250 --> 0:05:34.850 problem of noise. Nobody even fully understood it. Not until 0:05:35.290 --> 0:05:41.010 nearly a century later, along came Clawed Shannon. Shannon's career 0:05:41.170 --> 0:05:44.930 was defined by two thunderbolts of insight. When he was 0:05:44.970 --> 0:05:48.970 twenty one in nineteen thirty eight, his master's thesis showed 0:05:49.010 --> 0:05:52.850 that any logical statement could be evaluated by a machine, 0:05:53.290 --> 0:05:57.210 with true or false being represented by switches being open 0:05:57.490 --> 0:06:01.490 or closed those dots and dashes of Morse code, which 0:06:01.530 --> 0:06:04.770 as to hint at the possibilities. Armed only with open 0:06:04.970 --> 0:06:08.810 or closed on or off dot or dash zero or 0:06:08.850 --> 0:06:15.650 war machines could perform any operation in mathematics or logic, and, 0:06:15.770 --> 0:06:19.610 rather than merely proving the point in abstract, Shannon, who 0:06:19.690 --> 0:06:22.970 was barely old enough to buy a beer, showed electrical 0:06:23.010 --> 0:06:28.290 engineers how to efficiently build a logic machine. Claude Shannon 0:06:28.410 --> 0:06:33.010 had bridged the vast gap between electrical wiring diagrams and 0:06:33.210 --> 0:06:41.650 thought itself unlocking the age of the digital computer. Shannon's 0:06:41.690 --> 0:06:44.650 second Thunderbolt was published in nineteen forty eight, when he 0:06:44.730 --> 0:06:49.810 was working at Bell Labs alongside several future Nobel Prize winners, 0:06:49.890 --> 0:06:53.890 including the team that invented the transistor. Shannon returned to 0:06:53.930 --> 0:06:59.890 the deep problem underlying the Transatlantic cable fiasco. He created 0:06:59.970 --> 0:07:05.210 a unified mathematical theory of transmitting information. Some of that 0:07:05.290 --> 0:07:08.890 theory seems obvious from the viewpoint of the twenty first century. 0:07:09.290 --> 0:07:12.810 We now take it for granted that information bits and 0:07:12.850 --> 0:07:17.330 bytes and gigabytes might represent anything a computer game or 0:07:17.330 --> 0:07:22.770 a spreadsheet, or music or pornography. But that idea started 0:07:23.090 --> 0:07:28.290 with Shannon. Before him, researchers only dimly grasped the distinction 0:07:28.410 --> 0:07:31.290 between the meaning of a message and the quantity of 0:07:31.330 --> 0:07:36.490 information it contained. The idea of compressing a file so 0:07:36.530 --> 0:07:39.970 that it took up less space was Shannon's, and so 0:07:40.130 --> 0:07:43.690 too was the utterly radical idea that any amount of 0:07:43.730 --> 0:07:47.250 noise on a line could be overcome. We didn't do 0:07:47.370 --> 0:07:50.690 that by cranking up the voltage and melting the undersea cable, 0:07:51.050 --> 0:07:53.410 nor did you need to build a better listening device 0:07:53.530 --> 0:07:57.570 or a thicker cable. No matter how much distortion there was, 0:07:58.530 --> 0:08:02.330 you could convey any message if you had enough patience. 0:08:03.050 --> 0:08:06.370 All you had to do was add redundancy to the data. 0:08:06.610 --> 0:08:10.330 It's the inverse of compressing a file. You add extra 0:08:10.410 --> 0:08:13.330 data to make the message more likely to be recoverable. 0:08:13.770 --> 0:08:18.730 Even in the presence of interference. That idea was unthinkable, 0:08:19.250 --> 0:08:22.130 right up to the point that Claude Shannon proved how 0:08:22.130 --> 0:08:29.490 to do it. This new theory of information was revolutionary 0:08:29.850 --> 0:08:32.610 and so elegant and general that it could be applied 0:08:32.650 --> 0:08:36.250 to anything from the Internet to genetic information in DNA, 0:08:36.850 --> 0:08:39.530 even though the Internet did not then exist, and the 0:08:39.570 --> 0:08:42.970 double helix structure of DNA had not yet been discovered. 0:08:43.770 --> 0:08:47.330 Shannon wasn't merely ahead of his time. He was the 0:08:47.410 --> 0:08:50.410 one who had wound the clock and set it running. 0:08:51.410 --> 0:08:56.410 All this and he'd barely turned thirty. So what did 0:08:56.450 --> 0:09:00.650 Shannon do for an encore? Is a description from his biographers, 0:09:01.010 --> 0:09:05.930 Jimmy Sony and Rob Goodman of Shannon's work ethic. Shannon 0:09:06.010 --> 0:09:09.250 arrived late, if at all, and often spend the day 0:09:09.290 --> 0:09:12.370 absorbed in games of chess and hex in the common areas. 0:09:12.970 --> 0:09:15.810 When not besting his colleagues at board games, he would 0:09:15.810 --> 0:09:20.170 be found piloting a unicycle through Bell Labs's narrow passageways, 0:09:20.530 --> 0:09:24.330 occasionally while juggling. Sometimes he would pogo stick his way 0:09:24.330 --> 0:09:27.530 around the Bell Labs campus, much to the consternation we 0:09:27.610 --> 0:09:32.530 imagine of the people who signed his paychecks. Shannon wasn't 0:09:32.530 --> 0:09:36.770 goofing off completely. He often worked hard, but the projects 0:09:36.770 --> 0:09:40.850 he worked on seemed whimsical. For example, he spent many 0:09:40.890 --> 0:09:44.930 hours at home playing with a colossal erector set. He 0:09:44.970 --> 0:09:48.090 built a robot mouse that could explore a maze and, 0:09:48.290 --> 0:09:51.130 by trial and error, on the first attempt, learn how 0:09:51.170 --> 0:09:54.770 to reach its target flawlessly. On the second run, the 0:09:54.890 --> 0:09:58.050 robot mouse was clever and thought provoking, and it might 0:09:58.090 --> 0:10:03.250 have represented real progress towards artificial intelligence if Shannon had 0:10:03.290 --> 0:10:08.450 persisted with it. But he didn't. Shannon built perhaps the 0:10:08.610 --> 0:10:13.170 first chess playing computer, albeit one that could play only 0:10:13.210 --> 0:10:16.850 a radically simplified setup the end game with six pieces. 0:10:17.610 --> 0:10:21.250 He published a theoretical paper on computer chess. It could 0:10:21.250 --> 0:10:24.890 have been the start of something, but again he lost interest. 0:10:26.210 --> 0:10:29.090 It seemed a shame. If anyone could make progress with 0:10:29.130 --> 0:10:34.090 computer chess, surely it was Shannon. He was good. Shannon 0:10:34.170 --> 0:10:37.210 once traveled to Moscow and played chess with three time 0:10:37.290 --> 0:10:42.450 world champion Mikhail Botvinick, and he made Botvinick sweat. When 0:10:42.450 --> 0:10:45.570 it wasn't chess, it was juggling. Shannon tried to figure 0:10:45.570 --> 0:10:48.370 out how to juggle upside down by hanging from the 0:10:48.410 --> 0:10:51.410 ceiling and bouncing the balls off the floor. He built 0:10:51.490 --> 0:10:55.370 juggling robots too, and a variety of machines designed to 0:10:55.410 --> 0:10:59.610 play abstract games, such as Hex and a Rubik's Cube 0:10:59.610 --> 0:11:04.170 solving robot, and the Juggleometer, and a flame throwing trumpet 0:11:04.770 --> 0:11:09.290 and the Ultimate Machine. The Ultimate Machine is a box 0:11:09.330 --> 0:11:12.130 with a switch and a trapdoor, and you flick the 0:11:12.170 --> 0:11:15.730 switch to turn it on. A robot finger pops out 0:11:15.730 --> 0:11:18.610 of the trapdoor and flips the switch back again to 0:11:18.690 --> 0:11:23.770 turn itself off. Shannon made giant styrofoam shoes so he 0:11:23.770 --> 0:11:27.410 could walk on water at a nearby lake. After Shannon 0:11:27.530 --> 0:11:30.810 learned to juggle, ride a unicycle and walk a tightrope, 0:11:31.290 --> 0:11:34.770 he formulated the aim of juggling on a unicycle on 0:11:34.890 --> 0:11:38.850 a tightrope Alas he never got further than two out 0:11:38.850 --> 0:11:47.250 of three, Claude Shannon's boss, Henry Pollock, said, Shannon has 0:11:47.370 --> 0:11:51.650 earned the right to be non productive, and of course 0:11:51.690 --> 0:11:57.690 he had. But come on, you're a genius, Claude. You're 0:11:57.770 --> 0:12:01.370 thirty three years old, You're the Einstein of computer science, 0:12:01.810 --> 0:12:07.170 and you're unicycling, poe going and playing board games. Shannon 0:12:07.610 --> 0:12:12.250 never again published anything like his theory of information. He 0:12:12.370 --> 0:12:15.970 never even came close. Once he promised the editor of 0:12:16.050 --> 0:12:20.650 Scientific American an article on the physics of juggling. If 0:12:20.650 --> 0:12:24.130 that didn't seem trivial enough, he followed it up with 0:12:24.210 --> 0:12:28.490 an unapologetic letter. You probably think I've been frittering, I say, 0:12:28.530 --> 0:12:31.530 frittering away my time while my juggling paper is languishing 0:12:31.530 --> 0:12:34.370 on the shelf. This is only half true. I have 0:12:34.490 --> 0:12:37.770 come to two conclusions recently. One, I am a better 0:12:37.810 --> 0:12:42.210 poet than scientist. Two scientific American should have a poetry 0:12:42.290 --> 0:12:47.970 column instead of his juggling research. Shannon enclosed a seventy 0:12:48.050 --> 0:12:52.690 line poem about Rubik's Cubes, to be sung to the 0:12:52.770 --> 0:12:58.850 tune of Tarara Bundier. He added, I'm still working on 0:12:58.890 --> 0:13:05.010 the juggling paper. Shannon never finished it. Not only was 0:13:05.050 --> 0:13:09.450 he not producing thunderbolts, he wasn't even producing a study ofuggling. 0:13:10.570 --> 0:13:14.210 Perhaps we should not be surprised that Claude Shannon was 0:13:14.250 --> 0:13:18.290 happy to put aside serious research when the young mathematician 0:13:18.530 --> 0:13:21.610 Ed Thorpe approached him for help in hacking the roulette 0:13:21.690 --> 0:13:28.890 table in Vegas. Cautionary tales will be back in a moment. 0:13:36.810 --> 0:13:39.570 If we know anything, we know we're supposed to stick 0:13:39.650 --> 0:13:44.330 to a task. Psychologists have developed some attractive ideas about 0:13:44.370 --> 0:13:49.970 how success depends on practice and determination. There's Angela Duckworth 0:13:50.130 --> 0:13:54.690 who's popularized the idea of grit, Carol Dwex's research on 0:13:54.930 --> 0:13:59.210 the growth mindset and the late and as Ericsson, the 0:13:59.250 --> 0:14:02.530 source of the ten thousand hour rule made famous by 0:14:02.570 --> 0:14:07.010 Malcolm Gladwell. There are subtleties to each of these research programs, 0:14:07.290 --> 0:14:10.170 but the versions that have broken into pop culture are 0:14:10.210 --> 0:14:12.890 simple enough, like some motivational poster. 0:14:14.050 --> 0:14:17.450 Nothing in this world can take the place of persistence. 0:14:18.330 --> 0:14:22.410 Talent will not. Nothing is more common than unsuccessful people 0:14:22.450 --> 0:14:28.010 with talent. Genius will not unrewarded genius is almost a proverb, 0:14:28.930 --> 0:14:33.330 the slogan press iron has solved and always will solve 0:14:33.410 --> 0:14:34.970 the problems of the human race. 0:14:36.330 --> 0:14:41.090 Isn't that great? It's often attributed to President Calvin Coolidge, 0:14:41.130 --> 0:14:45.450 but it's older than that. Claude Shannon, however, seems not 0:14:45.530 --> 0:14:49.570 to have gotten the message. He achieved so much, but 0:14:49.650 --> 0:14:52.330 if it's stuck to a task, couldn't he have achieved 0:14:52.370 --> 0:14:57.050 so much more? Instead, he was playing with flame throwing trumpets, 0:14:57.130 --> 0:15:01.530 juggling robots, and silly poems, oh, and the impossible task 0:15:01.570 --> 0:15:08.130 of beating the casino at roulette. For a junior academic 0:15:08.690 --> 0:15:12.330 ed thought, Thorpe spent a surprising amount of time in casinos. 0:15:13.170 --> 0:15:16.810 Using some ferocious mathematics and the best computers he could 0:15:16.850 --> 0:15:19.850 access at MIT, Thorpe had figured out that it was 0:15:19.930 --> 0:15:24.050 possible to beat the dealer at the casino staple blackjack 0:15:24.610 --> 0:15:26.970 by keeping track of the cards that had been played 0:15:27.570 --> 0:15:31.130 in placing bets when the deck was offering favorable odds. 0:15:31.890 --> 0:15:35.530 Card counting is a familiar idea these days. It all 0:15:35.650 --> 0:15:41.050 started with Ed Thorpe. Thorpe's ideas were sophisticated enough to 0:15:41.050 --> 0:15:44.250 be worth publishing as an academic paper, which he did, 0:15:45.130 --> 0:15:48.650 but he wasn't content with that. He wanted to beat 0:15:48.730 --> 0:15:53.250 the casino too. To do that, Thorpe had to learn 0:15:53.290 --> 0:15:58.170 to spot crooked dealing, where a disguise count cards unobtrusively 0:15:58.250 --> 0:16:01.490 late into the night, and above all, make sure he 0:16:01.530 --> 0:16:06.970 didn't get killed. That was no idle worry. One day, 0:16:07.130 --> 0:16:10.490 Thorpe made a little too much money and the casino 0:16:10.690 --> 0:16:14.890 spiked his coffee with something mysterious that blurred his vision 0:16:14.930 --> 0:16:18.610 for hours. He came back the next day and the 0:16:18.650 --> 0:16:23.850 casino tried it again, but Thorpe wasn't scared. His idea 0:16:23.890 --> 0:16:27.410 to beat roulette was the boldest of all. He didn't 0:16:27.450 --> 0:16:30.530 have in mind a clever mathematical system. There are loads 0:16:30.570 --> 0:16:34.330 of them, and he knew that none of them work. Instead, 0:16:34.810 --> 0:16:37.810 he planned to build a computer that could predict where 0:16:37.850 --> 0:16:41.410 the ball would land. That would be hard even today, 0:16:42.050 --> 0:16:45.130 but at a time when computers were the size of pianos. 0:16:45.890 --> 0:16:49.090 This computer needed to be one that you could conceal 0:16:49.290 --> 0:16:55.570 inside your clothes, the world's first wearable computer, decades before 0:16:55.610 --> 0:17:01.170 the fitbit, Google Glass or the Apple Watch. Thorpe had 0:17:01.210 --> 0:17:04.010 done some experiments on the timing of a roulette wheel 0:17:04.050 --> 0:17:07.330 with his wife, Vivian, a woman who was both intelligent 0:17:07.650 --> 0:17:10.730 and indulgent, as you'd need to be if you were 0:17:10.770 --> 0:17:14.530 marriage to Ed Thorpe. But to crack the problem he 0:17:14.610 --> 0:17:18.330 needed to team up with perhaps the best gadgeteer in 0:17:18.410 --> 0:17:23.770 the world, Claude Shannon. Thorpe spent twenty hours a week 0:17:23.810 --> 0:17:26.210 at Shannon's house. He was in heaven. 0:17:26.890 --> 0:17:35.970 The basement was a gadgeteer's paradise, molders, transistors, switches, pulleys, gears, condensers, transformers. 0:17:36.650 --> 0:17:40.210 I was now happily working with the ultimate gadgeteer. 0:17:41.130 --> 0:17:43.650 Shannon and Thorpe were able to time the spinning of 0:17:43.650 --> 0:17:46.330 the ball around an upper loop and the contrary motion 0:17:46.370 --> 0:17:48.970 of the wheel itself. With practice, they were able to 0:17:49.010 --> 0:17:52.050 start a clock within one hundredth of a second and 0:17:52.090 --> 0:17:55.330 then stop the clock after ten revolutions. That gave them 0:17:55.370 --> 0:17:57.890 both the speed and the position of the ball. Relative 0:17:57.930 --> 0:18:00.650 to the wheel, and Newtonian physics could do the rest. 0:18:02.450 --> 0:18:06.090 The result of months of experimentation taught them that using 0:18:06.130 --> 0:18:08.650 their computer to compute the path of the ball, they 0:18:08.650 --> 0:18:11.210 could put that it would fall into one of five 0:18:11.290 --> 0:18:15.250 numbers just over one eighth of the wheel and expect 0:18:15.330 --> 0:18:18.330 to be right twenty percent of the time. It seems 0:18:18.370 --> 0:18:23.250 a modest advantage, but the potential profits were enormous. All 0:18:23.290 --> 0:18:25.730 they had to do was to figure out how to 0:18:25.810 --> 0:18:29.850 miniaturize that computer, making it small enough to slip into 0:18:29.890 --> 0:18:33.890 a pocket and carry into the casino undetected. It was 0:18:34.010 --> 0:18:38.850 an astonishingly audacious project and a huge effort. For the 0:18:38.890 --> 0:18:42.650 final three weeks, Thorpe was effectively living at Shannon's house, 0:18:43.290 --> 0:18:47.010 but by August nineteen sixty one, the device was ready 0:18:47.850 --> 0:18:51.770 with their accomplices, Vivian Thorpe and Claude's wife, the mathematician 0:18:51.810 --> 0:18:55.970 Betty Shannon. The two gadgeteers then took it to the casinos. 0:18:56.330 --> 0:19:04.010 The Einstein of computer science was going to Las Vegas. 0:19:05.250 --> 0:19:08.730 Looking at Claude Shannon's career from age thirty three onwards, 0:19:09.170 --> 0:19:11.970 it's hard to escape the conclusion that he might have 0:19:12.010 --> 0:19:15.490 achieved more, much more, if not for his habit of 0:19:15.690 --> 0:19:19.530 flitting between whimsical projects and typically setting them aside before 0:19:19.570 --> 0:19:23.610 they were finished. But some very smart people would disagree. 0:19:24.050 --> 0:19:28.090 Vanevar Bush arguably knew more than anyone about the way 0:19:28.210 --> 0:19:32.650 scientific progress occurred. He guided science policy for the United 0:19:32.690 --> 0:19:36.490 States during the Second World War, coordinating the efforts of 0:19:36.730 --> 0:19:41.650 six thousand researchers. Bush said that great scientists should range 0:19:41.690 --> 0:19:45.250 widely and keep changing things up. In a speech to 0:19:45.290 --> 0:19:50.930 professors at MIT, Bush advocated breadth rather than depth. It 0:19:51.050 --> 0:19:55.330 is unfortunate when a brilliant and creative mind insists upon 0:19:55.370 --> 0:19:59.690 living in a modern monastic cell. Bush's idea was later 0:19:59.730 --> 0:20:06.450 backed up by scientific investigation of scientists themselves. In nineteen 0:20:06.490 --> 0:20:10.050 fifty eight, a remarkable study was launched by a young 0:20:10.130 --> 0:20:14.850 psychologist named Bernice Agison. The study followed a group of 0:20:14.890 --> 0:20:19.850 promising researchers as their careers unfolded, periodically interviewing them and 0:20:19.930 --> 0:20:23.610 continuing even after Agison herself died in nineteen eighty five. 0:20:24.450 --> 0:20:29.010 Four of the scientists eventually won Nobel Prizes. The findings 0:20:29.050 --> 0:20:32.810 of the Agison study support Shannon's habit of flipping from 0:20:32.810 --> 0:20:36.850 one project to another. The scientists who'd most flourished over 0:20:36.890 --> 0:20:40.730 the decades, had switched back and forth dozens of times. 0:20:41.450 --> 0:20:44.650 Once you start looking for this pattern, you see it everywhere. 0:20:45.250 --> 0:20:48.650 Isaac Newton is most famous for formulating the law of gravity, 0:20:49.090 --> 0:20:53.050 but made huge advances in mathematics and optics. He was 0:20:53.050 --> 0:20:56.690 the master of the Royal Mint and was fascinated by economics, 0:20:57.130 --> 0:21:00.370 and devoted as much attention to alchemy as to anything else. 0:21:01.090 --> 0:21:06.090 Einstein published four astonishing scientific papers on four different topics, 0:21:06.290 --> 0:21:10.450 all in the same year nineteen oh five. Charles Darwin 0:21:10.570 --> 0:21:14.810 worked simultaneously on the theory of evolution. The definitive two 0:21:14.890 --> 0:21:18.890 volume work on barnacles and a book about the human infant, 0:21:19.290 --> 0:21:23.130 began while his son William was a baby, and published 0:21:23.290 --> 0:21:27.730 just in time for William Darwin's thirty eighth birthday. Multiple 0:21:27.770 --> 0:21:31.890 projects aren't unusual at the highest level of science. They're 0:21:31.890 --> 0:21:36.090 the norm. Not only that, high achieving scientists often have 0:21:36.250 --> 0:21:41.050 time consuming side interests, pursuing photography, fine art, or music 0:21:41.410 --> 0:21:46.490 at or near a professional level. Nobel Prize winning scientists 0:21:46.610 --> 0:21:50.290 are substantially more likely to have serious hobbies than other 0:21:50.410 --> 0:21:53.650 leading scientists, who in turn are more likely to have 0:21:53.730 --> 0:21:57.410 them than the rest of us. The later part of 0:21:57.490 --> 0:22:02.090 Shannon's career fits right into this highly diverse pattern, but 0:22:02.130 --> 0:22:06.290 then so does the early part. Back in nineteen thirty nine, 0:22:06.570 --> 0:22:10.930 shortly after his first thunderbolt, he wrote a note to 0:22:11.010 --> 0:22:16.490 an academic mentor, dear doctor Bush, Yes Vanovar Bush, the 0:22:16.530 --> 0:22:20.330 man who knew everyone who mattered in mid century American science. 0:22:20.810 --> 0:22:23.970 Of course, he was there to support the young Claude Shannon. 0:22:24.890 --> 0:22:29.370 Dear doctor Bush, I've been working on three different ideas simultaneously, 0:22:29.410 --> 0:22:32.530 and strangely enough, it seems a more productive method than 0:22:32.610 --> 0:22:36.570 sticking to one problem. When Shannon wrote to Vanovar Bush, 0:22:36.970 --> 0:22:40.450 he wasn't working on engineering or logic. He was working 0:22:40.530 --> 0:22:44.570 on genetics. He knew nothing about the subject, but swiftly 0:22:44.610 --> 0:22:48.410 produced a completely new kind of algebra to describe and 0:22:48.650 --> 0:22:53.490 analyze genetic inheritance. The work was intriguing and wholly original, 0:22:53.770 --> 0:22:57.730 but needed developing. Did Shannon develop it He did not. 0:22:58.690 --> 0:23:02.370 In fact, he never even bothered publishing it. Neither did 0:23:02.370 --> 0:23:06.450 he ever return to genetics. Later scholars lament the loss. 0:23:06.730 --> 0:23:09.810 His new algebra might really have advance on the field, 0:23:10.410 --> 0:23:13.370 but sticking with genetics might also have meant he never 0:23:13.450 --> 0:23:19.410 had his second thunderbolt on information theory. Between those two thunderbolts, 0:23:19.610 --> 0:23:23.010 Shannon didn't just switch fields. He lit a rich and 0:23:23.130 --> 0:23:27.010 complicated life. He married and then divorced within a year. 0:23:27.410 --> 0:23:30.290 He moved to Manhattan to spice things up. It played 0:23:30.370 --> 0:23:33.810 chess in Washington Square Park. It played clarinet. He loved 0:23:33.850 --> 0:23:37.610 the jazz scene in New York. He swam, played tennis, 0:23:37.930 --> 0:23:40.410 stayed up too late, and played his music too loud. 0:23:41.170 --> 0:23:43.770 All this was happening when Shannon was at the peak 0:23:43.930 --> 0:23:48.010 of his intellectual powers. Shannon didn't just hit thirty five 0:23:48.330 --> 0:23:52.450 then abandoned serious thinking in favor of playing around. Shannon 0:23:53.010 --> 0:23:57.370 was playing around all along. Maybe Shannon's love of frittering, 0:23:57.490 --> 0:24:02.090 I say, frittering away his time on juggling or unicycling, 0:24:02.330 --> 0:24:05.850 or music or chess. Maybe that's not the reason he 0:24:05.930 --> 0:24:10.810 produced only two truly brilliant ideas. Maybe it's the reason 0:24:11.010 --> 0:24:15.050 he produced two truly brilliant ideas in the first place. 0:24:19.330 --> 0:24:28.610 Cautionary tales will be back in a moment. I try 0:24:28.650 --> 0:24:31.050 hard to answer all the people who write to me. 0:24:31.730 --> 0:24:35.610 I get anxious knowing that the task is unfinished. Claude 0:24:35.610 --> 0:24:39.330 Shannon didn't feel that same compulsion to clear his inbox 0:24:39.850 --> 0:24:44.290 He often left correspondence unanswered, then eventually cleared the decks 0:24:44.330 --> 0:24:47.130 to the use of a trash can marked letters I've 0:24:47.210 --> 0:24:51.570 procrastinated on for too long. That might seem a trivial thing, 0:24:51.650 --> 0:24:55.130 but I think it points to something deeper. Psychologists have 0:24:55.210 --> 0:25:00.490 identified a tendency called completion bias. If you've ever assembled 0:25:00.490 --> 0:25:03.050 a list of things to do, then ticked off all 0:25:03.090 --> 0:25:07.410 the easy ones while ignoring the important stuff, you've demonstrated 0:25:07.450 --> 0:25:13.530 completion bias. That apparently admirable tendency persistence the determination to 0:25:13.570 --> 0:25:17.730 finish what we start well, it could be twisted and perverted. 0:25:18.210 --> 0:25:21.490 If we feel compelled to reach the finish line, we 0:25:21.610 --> 0:25:25.450 also feel tempted to choose a short racetrack. Is more 0:25:25.490 --> 0:25:28.770 at stake here than making ourselves feel better by cheating 0:25:28.810 --> 0:25:33.170 with our own to do lists. Psychologists recently studied completion 0:25:33.290 --> 0:25:37.170 by us in a high stake setting, a hospital emergency department. 0:25:37.730 --> 0:25:40.970 They found that the busier the emergency room becomes, the 0:25:41.010 --> 0:25:44.490 more the doctors look for quick wins the patients who 0:25:44.570 --> 0:25:47.690 aren't really very ill and can therefore be treated swiftly 0:25:48.050 --> 0:25:51.810 and ticked off the list, and this behavior is counterproductive. 0:25:52.450 --> 0:25:56.170 The more seriously ill patients wait longer, of course, and 0:25:56.290 --> 0:25:58.930 the doctors start to slow down after working through a 0:25:58.970 --> 0:26:02.210 lot of fairly trivial cases. I expect we all know 0:26:02.290 --> 0:26:05.930 the feeling, but in their subconscious desire to see some 0:26:06.170 --> 0:26:10.690 work through to completion, doctors were harming the patients who 0:26:10.690 --> 0:26:17.010 were in greatest need. Claude Shannon's willingness to set aside 0:26:17.090 --> 0:26:21.210 projects starts to look like a strength rather than a weakness. 0:26:21.770 --> 0:26:26.290 Shannon certainly could focus, whether building information theory from scratch 0:26:26.810 --> 0:26:31.050 or building a wearable computer to be Roulette. Yet Shannon 0:26:31.170 --> 0:26:34.210 also seemed to have an inner confidence that allowed him 0:26:34.250 --> 0:26:37.410 to declare victory at any point that suited him. If 0:26:37.450 --> 0:26:40.490 a piece of work was not good enough to publish, fine, 0:26:40.770 --> 0:26:44.410 he was happy to leave it unpublished. That juggling paper 0:26:44.450 --> 0:26:47.450 is an example, but so too was his early work 0:26:47.490 --> 0:26:51.890 on genetic algebra. One of Claude Shannon's colleagues at Bell 0:26:52.010 --> 0:26:56.690 Labs praised him as a man of infinite courage. He 0:26:56.810 --> 0:27:01.810 was talking about Shannon's intellectual daring, a willingness to march 0:27:01.970 --> 0:27:05.610 into unknown territory to begin the search for solutions to 0:27:05.690 --> 0:27:10.490 problems that seemed as unbeatable as Roulette. But perhaps Courage 0:27:10.530 --> 0:27:13.490 is not quite the right word to describe Shannon's approach. 0:27:14.290 --> 0:27:20.050 I prefer in soussiens Claude. Shannon just wasn't worried. He 0:27:20.090 --> 0:27:22.610 didn't feel completion by us the way you and I 0:27:22.730 --> 0:27:25.810 feel it. He would walk away from any project at 0:27:25.890 --> 0:27:29.730 any time without regret. And if he was willing to 0:27:29.810 --> 0:27:34.050 abandon a stalled project, where was the risk? And if 0:27:34.090 --> 0:27:38.490 there was little risk, why talk about courage. Shannon didn't 0:27:38.530 --> 0:27:42.330 need courage, He just needed the ability to move on. 0:27:45.090 --> 0:27:49.050 In August nineteen sixty one, Claude and Betty Shannon met 0:27:49.250 --> 0:27:52.570 Ed and Vivian Thorpe in a hotel room in Las Vegas. 0:27:53.330 --> 0:27:57.170 Claude and Ed prepared the wearable computer system, which required 0:27:57.170 --> 0:28:01.370 both of them to operate. Shannon controlled the computer itself, 0:28:01.810 --> 0:28:04.890 the size of a cigarette packet, with twelve transistors in it. 0:28:05.730 --> 0:28:09.650 He used his toes to trigger silent mercury switches hidden 0:28:09.690 --> 0:28:13.890 in his shoes. Thorpe, whose research into blackjack had given 0:28:13.970 --> 0:28:17.290 him plenty of experience hanging around in casinos, was the 0:28:17.330 --> 0:28:20.050 one who would place the bets. He had a radio 0:28:20.090 --> 0:28:23.290 receiver and an ear piece connected to a hair thin 0:28:23.410 --> 0:28:28.530 steel wire. The ear piece played an ascending musical scale. 0:28:28.570 --> 0:28:31.770 Shannon would use the toe switches to time a rotation 0:28:31.890 --> 0:28:34.530 of the wheel, and then the counter rotation of the 0:28:34.570 --> 0:28:38.650 ball from the moment it passed a reference mark. Thorpe 0:28:38.810 --> 0:28:42.410 would hear the musical scale stop on a continuous note 0:28:42.850 --> 0:28:46.050 at the moment that Shannon finished timing the rotation, and 0:28:46.090 --> 0:28:49.330 the pitch of that continuous note would indicate in which 0:28:49.370 --> 0:28:51.930 part of the wheel the ball was likely to drop. 0:28:52.930 --> 0:28:56.250 Thorpe still had a few seconds to place bets and 0:28:56.330 --> 0:29:01.130