成功方程式——引言（成功的来源）

Intr od u c t i o n 3

M 15 Y CARE E R WAS L AUNCHE D by a trash can. 16 Like many seniors in college, I wasn’t sure what I wanted 17 to do for a living, but I knew that I needed a job. Drexel Burnham 18 ­Lambert, an investment bank that was hot at the time, came on cam- 19 pus to recruit students for a new training program. My interview 20 went well enough that I was called to the firm’s headquarters in 21 New York City. I put on my best suit and tie, polished my shoes, and 22 headed to the Big Apple. 23 Early the next morning, we candidates gathered in a vast confer- 24 ence room and listened intently as the leader of the program told 25 us what to expect for the day. “You will have full interviews with 26 six members of our staff,” she informed us, “and then each of you 27 will have ten minutes with the senior executive in charge of our 28 division.” When it was clear that she had everyone’s attention, she 29 added, “If you want the job, you’ll have to shine in that interview.” 30 My half dozen interviews went as well as could be expected. When 31 they were over, a member of the staff led me down a long corridor to 32 an office paneled in dark wood, with deep wall-to-wall carpeting and 33 a picture window overlooking a panorama of downtown Manhattan. 34

1 A sharp-eyed administrative assistant ushered me in, and the senior 2 executive greeted me warmly. Then I saw it.

3 Peeking out from underneath a huge desk was a trash can bearing 4 the logo of the Washington Redskins, a professional football team. 5 As a sports fan who had just spent four years in Washington, D.C., 6 and had attended a game or two, I complimented the executive on 7 his taste in trash cans. He beamed, and that led to a ­ten-minute inter- 8 view that stretched to fifteen minutes, during which I listened and 9 nodded intently as he talked about sports, his time in ­Washington, 10 and the virtues of athletics. His response to my opening was purely 11 emotional. Our discussion was not intellectual. It was about a 12 shared passion.

13 I got the job. My experience in the training program at Drexel 14 Burnham was critical in setting the trajectory of my career. But 15 after a few months in the program, one of the leaders couldn’t resist 16 pulling me aside. “Just to let you know,” he whispered, “on balance, 17 the six interviewers voted against hiring you.” I was stunned. How 18 could I have gotten the job? He went on: “But the head guy overrode 19 their assessment and insisted we bring you in. I don’t know what 20 you said to him, but it sure worked.” My career was launched by 21 a trash can. That was pure luck, and I wouldn’t be writing this if 22 I hadn’t benefited from it.

23 24 25 The Boundaries of Skill and Luck 26 27 Much of what we experience in life results from a combination of 28 skill and luck. A basketball player’s shot before the final buzzer 29 bounces out of the basket and his team loses the national champi- 30 onship. A pharmaceutical company develops a drug for hyperten- 31 sion that ends up as a blockbuster seller for erectile dysfunction. 32 An investor earns a windfall when he buys the stock of a company 33 shortly before it gets acquired at a premium. Different levels of skill 34 and of good and bad luck are the realities that shape our lives. And 35 yet we aren’t very good at distinguishing the two.

Part of the reason is that few of us are well versed in statistics. 1 But psychology exerts the most profound influence on our failure to 2 identify what is due to skill and what is just luck. The mechanisms 3 that our minds use to make sense of the world are not well suited 4 to accounting for the relative roles that skill and luck play in the 5 events we see taking shape around us. Let me start with some exam- 6 ples that are clearly controlled by either luck or skill. 7 The drawing for Powerball, a multistate lottery, went off unevent- 8 fully on the evening of Wednesday, March 30, 2005. The first five 9 balls came through the clear tubes: 28, 39, 22, 32, 33. The final ball, 10 which came from a separate machine, clicked into place: 42. The 11 whole process took less than a minute. 12 Sue Dooley, the staff member who was overseeing the drawing 13 that night, rolled the machines back into the vault and drove from 14 the television studio to the Powerball headquarters five miles away. 15 Based on the statistics, she expected that perhaps one ticket would 16 take home that day’s jackpot of $84 million and that three or four 17 people would have picked five of the six numbers correctly, winning 18 second place. 19 She turned on her computer and waited for the states to report 20 their results. The trickle of winners she had expected was actually a 21 torrent. In total, there were 110 second-place winners. The statisti- 22 cians employed by Powerball had warned that six or seven times the 23 predicted figure was well within the realm of chance, but an outcome 24 nearly thirty times the expectation appeared statistically impossi- 25

ble. Another oddity was that nearly all of the winning tickets had 26 the same sixth number, 40. Truth be told, the officials at Powerball 27 would have preferred it if the winners had picked all six numbers 28 correctly, because the jackpot is split evenly among them. No mat- 29 ter how many people win, it costs Powerball the same amount. But 30 each winner of the second prize receives a set amount, which meant 31 that in this case, Powerball had to pay out $19 million more than it 32 had anticipated. 33 Dooley called her boss and together they puzzled over possible 34 explanations, including numbers shown on television, pattern plays, 35

1 lottery columns, and even fraud. None of them checked out. The next 2 morning, they got their first inkling of what had happened. When 3 a staff member at a prize office in Tennessee asked a winner where 4 he had gotten his numbers, he answered, “From a fortune cookie.” 5 Later a winner in Idaho said the same thing, and shortly thereafter 6 winners in Minnesota and Wisconsin echoed the reply. Jennifer 8. 7 Lee, a reporter from the New York Times, jumped on the story and 8 traced the fortune cookies with the winning numbers back to the 9 factory of Wonton Food in Long Island City, New York. ­Derrick 10 Wong, a vice president at the company, explained that they had 11 put numbers in a bowl and randomly picked out six of them. Since 12 generating the number sequences takes time, the company printed 13 the same numbers on different fortunes so as to save labor on the 14 4 million cookies the factory produced each day.1 Each of those very 15 lucky winners took home between $100,000 and $500,000, according 16 to how much they had bet.

17 Marion Tinsley won a lot, too, but it wasn’t because he was lucky. 18 Tinsley was known as the greatest player of checkers (also known as 19 draughts) in the world. In 1948 he was crowned as the United States 20 champion; shortly before his death in 1994, he tied Don Lafferty 21 and a computer program named Chinook for first place. In the inter- 22 vening forty-five years, Tinsley lost only seven individual games 23 for a near-perfect record. In two of those games he was defeated by 24 ­Chinook. Despite the fact that he didn’t play for long periods of time 25 (he was a professor of mathematics at Florida State and Florida 26 A&M Universities), he reigned as world champion in three separate 27 decades.2 28 Tinsley’s success resulted from years of deliberate practice. In his 29 youth, Tinsley spent eight hours a day, five days a week, studying 30 checkers, and he continued to study the game, though less intensely, 31 throughout his life. He cultivated a prodigious memory that allowed 32 him to recall the flow of games he had played decades earlier. ­Tinsley 33 was fiercely competitive and claimed that he could beat all comers, 34 man or machine, as long as his health didn’t fail him.3

The Sources of Success 1 2 Both those who won the Powerball lottery and Marion Tinsley 3 enjoyed great success. But it’s easy to see that the causes of the two 4 types of success differed markedly. The lottery outcome that day 5 was a matter of pure good luck for the 110 winners and pure bad luck 6 for Powerball. But Tinsley’s success was almost entirely the result 7 of skill. With all the luck in the world, you would have almost no 8 chance of winning if Tinsley were across the table from you. For 9 practical purposes, we can regard Tinsley’s success as all skill. 10 Unfortunately, most things in life and business are not that clear. 11 Most of the successes and failures we see are a combination of skill 12 and luck that can prove maddeningly difficult to tease apart. 13 The purpose of this book is to show you how you can understand 14 the relative contributions of skill and luck and how to use that 15 understanding in interpreting past results as well as making better 16 decisions in the future. Ultimately, untangling skill and luck helps 17 with the challenging task of prediction, and better predictions lead 18 to greater success. 19 20 21 Skill, Luck, and Prediction 22 23 Shortly after winning the Nobel Prize in Economics in 2002, ­Daniel 24 Kahneman, a retired professor of psychology at Princeton, was asked 25 which of his 130-plus academic papers was his all-time ­favorite.4 He 26 chose “On the Psychology of Prediction,” a paper he cowrote with 27 the late Amos Tversky that was published in ­Psychological Review in 28 1973. The paper argues that intuitive judgments are often unreliable 29 because people base predictions on how well an event seems to fit a 30 story. They fail to consider either how reliable the story is or what 31 happened before in similar situations. More formally, ­Kahneman 32 and Tversky argue that three types of information are relevant to 33 statistical prediction. The first is prior information, or the base rate. 34

1 For example, if 85 percent of the taxicabs in a city are green, then 2 85 percent is the base rate. Absent any other information, you can 3 assume that whenever you see a taxicab there’s an 85 percent chance 4 that it will be green. The second type of information is the specific 5 evidence about an individual case. The third type of information is 6 the expected accuracy of the prediction, or how precise you expect it 7 to be given the information you have.5 8 I had a conversation with a doctor that illustrates these three 9 types of information. He mentioned that he had a treatment for 10 improving a specific ailment that succeeded about 50 percent of the 11 time (the base rate). But he added that he could induce almost any 12 patient to undergo the treatment if he simply told them, “The last 13 patient who was treated this way is doing great!” (specific evidence 14 about an individual case). For the patients who were evaluating the 15 treatment, the story of success swamped the statistics.

16 The key to statistical prediction is to figure out how much 17 weight you should assign to the base rate and specific case. If the 18 expected accuracy of the prediction is low, you should place most 19 of the weight on the base rate. If the expected accuracy is high, 20 you can rely more on the specific case. In this example, the doctor 21 gave the patient no reason to believe that the procedure had bet- 22 ter than a 50/50 chance of working for him. So the patient should 23 place almost no weight on the specific evidence that it worked for 24 one patient, and should rely instead on the base rate in making his 25 decision.

26 Here’s how the weighting of the base rate and the specific case 27 relate to skill and luck. When skill plays the prime role in deter- 28 mining what happens, you can rely on specific evidence. If you’re 29 playing checkers against Marion Tinsley, you can easily predict 30 the winner on the basis of your knowledge of Tinsley’s deadly skill. 31 In activities where luck is more important, the base rate should 32 guide your prediction. If you see someone win a million dollars, that 33 doesn’t change the odds of winning the lottery. Just because some- 34 one wins at roulette, it doesn’t help you to guess where the ball will 35 end up on the next spin.

Unfortunately, we don’t usually think this way. When we make 1 predictions, we often fail to recognize the existence of luck, and as a 2 consequence we dwell too much on the specific evidence, especially 3 recent evidence. This also makes it tougher to judge performance. 4 Once something has happened, our natural inclination is to come up 5 with a cause to explain the effect. The problem is that we commonly 6 twist, distort, or ignore the role that luck plays in our successes and 7 failures. Thinking explicitly about how luck influences our lives can 8 help offset that cognitive bias. 9 10 11 Quantifying Luck’s Role in the Success Equation 12 13 The starting place for this book is to go beyond grasping the gen- 14 eral idea that luck is important. Then we can begin to figure out the 15 extent to which luck contributes to our achievements, successes, 16 and failures. The ultimate goal is to determine how to deal with luck 17 in making decisions. 18 This book has three parts: 19 20

• Chapters 1 through 3 set up the foundation. I start with some 21 working definitions of skill and luck, examining the types 22 of interactions where luck is relevant and noting where our 23 methods to sort skill and luck may not work. I then turn to 24 why we have such a difficult time comprehending the influence 25 that luck exerts. The basic challenge is that we love stories 26 and have a yearning to understand the relationship between 27 cause and effect. As a result, statistical reasoning is hard, and 28 we start to view the past as something that was inevitable. 29 The section finishes by looking at the continuum from all-luck 30 to all-skill. I examine a basic model to help guide intuition.

31 These ideas include the paradox of skill and what determines 32 the rate of reversion to the mean.

33

• Chapters 4 through 7 develop the analytical tools necessary 34 to understand luck and skill. I open with methods for ­placing 35

1 activities on the luck-skill continuum. Where an activity 2 falls on that continuum provides a great deal of insight into 3 how to deal with it. I then look at how skill changes over time. 4 ­Simply put, skill tends to follow an arc: it improves for some 5 time, peaks, and then glides lower. Next, I turn ­attention 6 to the distributions—or the range of values—of luck. In 7 ­activities where the results are independent of one another, 8 simple models effectively explain what we see. But when a 9 past result affects a future result, predicting winners becomes 10 very difficult. The most skillful don’t always win. I close this 11 part by showing the difference between a useless statistic 12 and a ­useful one. Useful statistics are persistent (the past 13 ­correlates highly with the present) and predictive (doing well 14 or poorly ­correlates strongly with the desired goal). As we will 15 see, many statistics fail this basic test.

16

• Chapters 8 through 11 offer concrete suggestions about 17 how to take the findings from the first two parts of this book 18 and put them to work. I begin by outlining ways to improve 19 skill. Where little luck is involved, deliberate practice is 20 ­essential to developing skill. Where luck is rampant, we 21 must think of skill in terms of a process, because the results 22 don’t ­provide clear feedback. Checklists can also be of 23 great value because they improve execution and can guide 24 behavior under ­stressful circumstances. I then look at how 25 to cope with luck. When you are the favorite, for example, 26 you want to simplify the game so that you can overwhelm 27 your opponent. If you are the underdog, you want to inject 28 luck by making the game more complex. Because luck is in 29 part what remains ­unexplained, controlled tests allow for a 30 more accurate ­reading on causality. If you want to know if 31 an advertisement worked, for example, you need to consider 32 the purchasing behavior of those who saw the ad versus those 33 who didn’t. This part also includes an in-depth discussion of 34 reversion to the mean, an idea that most people believe they

understand, even though their behavior shows that they don’t. 1 The book finishes with ten concrete tips on how to overcome 2 the psychological, analytical, and procedural barriers in 3 untangling skill and luck. 4 5 This analysis of skill and luck will focus on business, sports, and 6 investing because these are the areas I know best. Naturally, these 7 realms are quite different. Sports are the easiest activities to ana- 8 lyze because the rules are relatively stable over time and there is 9 lots of data. Other social processes, including business, have fewer 10 rules and boundaries than sports and therefore tend to be more 11 c­omplex. Still, many of the same analytical methods are valid.6 12 ­Markets in general are the most difficult to analyze because prices 13 are ­established through the interaction of a large number of indi- 14 viduals. Here again, the nature of the problem may be somewhat 15 ­different from sports, but many of the tools for sorting out the 16 ­relative influence of skill and luck still apply.

17 Part of the fun and challenge of analyzing skill and luck is that 18 it’s a multidisciplinary endeavor. Statisticians, philosophers, psy- 19 chologists, sociologists, corporate strategists, professors of finance, 20 economists, and sabermetricians (those who apply statistical meth- 21 ods to the study of sports) all have something to contribute to the 22 discussion.7 Unfortunately, the people within these disciplines don’t 23 always reach outside their fields. You will see ideas from each of 24 these disciplines, and I’m hopeful that bringing them together will 25 lead to a sounder and more balanced approach to analyzing ­decisions 26 and interpreting the results.

27 28 Untangling skill and luck is an inherently tricky exercise, and there 29 are plenty of limitations, including the quality of the data, the sizes 30 of samples, and the fluidity of the activities under study. The argu- 31 ment here is not that you can precisely measure the contributions 32 of skill and luck to any success or failure. But if you take concrete 33 steps toward attempting to measure those relative contributions, 34 you will make better decisions than people who think improperly

1 about those issues or who don’t think about them at all. That will 2 give you an enormous advantage over them. Some statisticians, 3 especially in the world of sports, come across as know-it-alls who are 4 out of touch with the human side of things. This characterization is 5 unfair. Statisticians who are serious about their craft are acutely 6 aware of the limitations of analysis. Knowing what you can know 7 and knowing what you can’t know are both essential ingredients 8 of deciding well. Not everything that matters can be measured, and 9 not everything that can be measured matters.

10 While there are wide swaths of human activity where the ideas 11 in this book are hard to apply, the ideas have concrete applica- 12 tion in some important areas and should serve as a template for 13 ­thinking about decisions beyond the scope of this book. Luck may 14 explain that you met your future wife after your buddy lured you 15 out on a Thursday night, but this book will have little to directly say 16 about that or other issues of love, health, and happiness. We need to 17 define the activity we’re talking about and what measures we need 18 to use to evaluate that activity effectively.

19 In his book The Theory of Gambling and Statistical Logic, ­Richard 20 Epstein, a game theorist trained in physics, notes that there is no 21 way to assure that you’ll succeed if you participate in an ­activity 22 that combines skill and luck. But he does say, “It is gratifying 23 to rationalize that we would rather lose intelligently than win 24 ­ignorantly.”8 Luck may or may not smile on us, but if we stick to a 25 good process for making decisions, then we can learn to accept the 26 outcomes of our decisions with equanimity.