A Sense of the Enemy

by Shore, Zachary

  National leaders are always in need of thoughtful approaches to prediction, especially when lives are on the line in matters of war and peace. We therefore need to have some sense of what scholars from a range of disciplines have learned about predictions in general and enemy assessments in particular. One of the most recent observers to find fault with the prediction business is Nate Silver, the election guru whose work I described in chapter 9.5 Silver is only the latest thinker to tackle the question of how we can enhance our predictive prowess. Much of this work has involved, in one form or another, the question of discerning signals amid noise. Writing in 1973, the economist Michael Spence asked how an employer can distinguish potential good employees from bad ones before hiring anyone.6 Spence proposed that good and bad employees signal to employers by dint of their educational credentials. More recently, the sociologist Diego Gambetta has used signaling theory to understand criminal networks.7 Gambetta observes, for example, that a mafia must be especially prudent before including members into its organization. If candidates are not carefully vetted, the mafia might enlist an undercover police officer. By asking new recruits to commit a murder, the mafia imposes a substantial cost upon the undercover agent, who, presumably, would be unwilling to pay that price. A genuine would-be mobster, in contrast, can signal his commitment with a single shot. In Colombia, youth gangs have even been known to require that prospective members first kill one of their closest relatives to prove their sincerity. Imposing high costs upon ourselves is one way of signaling what we value.

  Another economist who focused on asymmetric information, George Akerlof, examined the market for used cars. In 1970, he suggested that because used-car buyers have no easy means of knowing the quality of a particular car, they will pay only what they believe to be the price of an average used car of a given model and year. As a result, owners of high-quality used cars (ones that were barely driven and well-maintained) will refuse to sell because they will not get the price they deserve, thereby reducing the overall average quality of used cars on the market. Although assessing enemy behavior and buying used cars are dramatically different realms, Akerlof’s notions do suggest the dangers that result from assuming that the seller (or the enemy) is of low quality. In chapter 8 we saw what happens when statesmen and their advisers project negative qualities onto their opponents without an accurate understanding of the other side.

  One obvious difference between these studies in economics and sociology on the one hand and the history of international relations on the other is that much of the time foreign leaders do not want to signal their true commitments. The strategic empath must therefore locate ways of identifying an adversary’s drivers amidst conflicting signals. Nevertheless, the idea of costs is useful. At meaningful pattern breaks, statesmen make choices with significant costs to themselves and with likely long-term implications. These actions can be valuable signals to foreign statesmen, even though they are unintentionally transmitted.8

  The natural sciences have also aided our understanding of prediction through the development of information theory. John Archibald Wheeler, the physicist who coined the term “black hole,” is also famous for crafting the catchphrase “it from bit.” Wheeler was a leading light in the development of nuclear fission, having studied under Niels Bohr and later having taught Richard Feynman. When Wheeler uttered his pithy slogan in 1989, he intended to imply that all matter as well as energy, the whole of our universe, emerged from information. The bit is a particle that cannot be split. Everything, in the end, reduces to information. But it was Claude Shannon, father of information theory in computer science, who truly brought about the information turn in scientific study.

  Shannon recognized that not all information is created equal. To test this, he pulled a Raymond Chandler detective novel from his bookcase and read a random passage to his wife: “A small oblong reading lamp on the—.” He asked Betty to guess which word came next. She failed to guess correctly at first, but once he told her that the first letter was d, the rest was easy. It was more than mere pattern recognition that mattered here. Shannon wanted to show that the information that counted most came before the missing letters, whereas the letters that followed the d in desk were of lesser value. For Shannon, information equaled surprise. The binary digits, or “bits,” as they came to be known, that mattered in any message were the ones that gave us something new or unexpected.9 It was a valuable insight, and one with applicability across fields. The historical cases in this book bore this out. Each case examined the particular bits of surprising information on which leaders focused and why that focus helped or hindered them.

  Evolutionary biology, specifically regarding the literature on theory of mind, is equally important for historians of decision-making. The classic experiment on theory of mind involves researchers who placed a candy in front of two little girls. We’ll call them Sally and Jane. The researchers then covered the candy with a box so it could not be seen. While Jane exited the room, the researchers removed the candy from under the box and hid it elsewhere, while leaving the box in place. When Jane returned, they asked Sally where Jane thinks the candy is located. Below the age of four most children think that Jane, the girl who did not see the candy being removed, will somehow know that the candy is no longer under the box. Most children believe that everyone else knows what they themselves know. It turns out that only after children reach the age of four do they discover that each of us has a distinct perspective on the world, shaped by access to different information. Before that age, children do not possess “theory of mind.” They cannot imagine that someone else does not possess the same knowledge or perspective that they themselves do.

  Compelling as they are, these theories have real limits when it comes to understanding the kinds of complex decisions that statesmen face. Although the theory of mind shows how we develop a kind of mental empathy—the ability to see things from another’s point of view—this work was initially centered on primates and very small children. That said, there does exist work of relevance to statecraft. In a paper by Alison Gopnik and others, for example, researchers describe the differences between two common ways in which we predict the actions of others.10 Most people, it seems, assume that past behavior is the best indicator of future behavior. If someone lied in the past, for example, that person can be expected to lie again. But others take a different view. Some people place greater weight on the current context. They do not discount past behavior, but they ask how the present context is likely to affect another’s actions. In my own study of statecraft, I find that the leaders who succeeded most at anticipating enemy actions incorporated analysis of both prior patterns and current context, but they heavily weighted the information gleaned at certain moments.

  One other scientific contribution bears indirectly, though significantly, on this book. Ray Kurzweil, the scientist who developed speech recognition software (and who is now the Director of Engineering at Google), has advanced a theory of how our brains function. In his 2012 book, How to Create a Mind, Kurzweil proposes the pattern recognition theory of mind to explain how the neocortex functions.11 Kurzweil points out that the primary purpose of our brains is in fact to predict the future through pattern recognition. Whether we are trying to anticipate threats, locate food sources, catch a ball, or catch a train, our brains are constantly performing complex calculations of probability.

  Kurzweil asserts that the neocortex—the large frontal region of the brain where most such calculations are conducted—is composed of layers upon layers of hierarchical pattern recognizers. These pattern recognizers, he maintains, are constantly at work making and adjusting predictions. He offers the simple sentence:

  Consider that we see what we expect to—

  Most people will automatically complete that sentence based on their brain’s recognition of a familiar pattern of words. Yet the pattern recognizers extend far deeper than that. To recognize the word “apple,” for example, Kurzweil notes that our brains not only anticipate th
e letter “e” after having read a-p-p-l, the brain must also recognize the letter “a” by identifying familiar curves and line strokes. Even when an image of an object is smudged or partially obscured, our brains are often able to complete the pattern and recognize the letter, or word, or familiar face. Kurzweil believes that the brain’s most basic and indeed vital function is pattern recognition.

  This ability is exceptionally advanced in mammals and especially in humans. It is an area where, for the moment, we still have a limited advantage over computers, though the technology for pattern recognition is rapidly improving, as evidenced by the Apple iPhone’s use of Siri speech recognition software. For a quick example of your own brain’s gifts in this arena, try to place an ad on the website Craigslist. At the time of this writing, in order to prove that you are a human and not a nefarious robot, Craigslist requires users to input a random string of letters or numbers presented on the screen. The image, however, is intentionally blurred. Most likely, you will have no difficulty identifying the symbols correctly. Robots, in contrast, will be baffled, unable to make sense of these distorted shapes. For fun, try the option for an audio clue. Instead of typing the image, listen to the spoken representation of those letters and numbers. You will hear them spoken in a highly distorted manner amidst background noise. The word “three,” for example, might be elongated, stressed, or intoned in a very odd way. “Thaaaaaaaa-reeeeeeee.” It sounds like the speaker is either drunk, on drugs, or just being silly. The point is that a computer program attempting to access the site could not recognize the numbers and letters when they do not appear in their usual patterns. Our brains possess an amazing ability to detect patterns even under extremely confusing conditions. But before you start feeling too smug, Kurzweil predicts that we have until the year 2029, when computers will rival humans in this and other regards. So enjoy it while it lasts.

  Let me sum up this section: Kurzweil’s theory suggests that pattern recognition is the brain’s most crucial function, and our sophisticated development of this ability is what gives human beings the edge over other animals and, for now, over computers as well. I suggest that the best strategic empaths are those who focus not only on enemy patterns but also on meaningful pattern breaks and correctly interpret what they mean. Next, Claude Shannon’s information theory shows that it is the new and surprising information that is more valuable than other data. I observe that pattern breaks are, in fact, markers of new and surprising information, possessed of greater value to leaders than the enemy’s routine actions. Finally, the theory of mind scholarship provides ways of thinking about how we mentalize, or enter another’s mind, which I employed throughout this book, but especially when scrutinizing how Stalin tried to think like Hitler.

  Before we grow too enamored of all these theories, we should remember that theories are not always right. Often their proponents, in their well-intentioned enthusiasm, exaggerate the scope and significance of their discoveries. This is particularly true of some recent works in social science—studies that bear directly on the nature of prediction.

  How Silly Are We?

  One of the striking features infusing much of the recent social science scholarship on prediction is its tendency to expose alleged human silliness. Across fields as diverse as behavioral economics, cognitive psychology, and even the science of happiness or intuition, studies consistently show how poor we are at rational decision-making, particularly when those choices involve our expectations of the future. Yet too often these studies draw sweeping conclusions about human nature from exceedingly limited data. In the process, they typically imply that their subjects in the lab will respond the same way in real life. Before we can apply the lessons of cognitive science to history, we must first be clear on the limits of those exciting new fields. We should temper our enthusiasm and must not be seduced by science.

  Consider one daring experiment by the behavioral economist Dan Ariely. Ariely recruited male students at the University of California Berkeley to answer intimate questions about what they thought they might do under unusual sexual settings. After the subjects had completed the questionnaires, he then asked them to watch sexually arousing videos while masturbating—in the privacy of their dorm rooms, of course. The young men were then asked these intimate questions again, only this time their answers on average were strikingly different. Things that the subjects had previously thought they would not find appealing, such as having sex with a very fat person or slipping a date a drug to increase the chance of having sex with her, now seemed much more plausible in their excited state. Ariely concluded from these results that teenagers are not themselves when their emotions take control. “In every case, the participants in our experiment got it wrong,” Ariely explains. “Even the most brilliant and rational person, in the heat of passion seems to be absolutely and completely divorced from the person he thought he was.”12

  Ariely is one of America’s most intriguing and innovative investigators of behavioral psychology. His research has advanced our understanding of how poorly we all know ourselves. And yet there is a vast difference between what we imagine we would do in a situation as compared to what we would actually do if we found ourselves in that situation. In other words, just because a young man in an aroused state says that he would drug his date does not guarantee that he truly would do it. He might feel very differently if the context changed from masturbating alone in his dorm room to being present with a woman on the real date. Can we be so certain that he really would slip the drug from his pocket into her drink? Or would he truly have sex with a very overweight person if she were there before him? Would he have sex with a sixty-year-old woman or a twelve-year-old girl, or any of Ariely’s other scenarios, if he were presented with the opportunity in real life? Life is not only different from the lab; real life has a funny way of being rather different from the fantasy.

  A great many recent studies suffer from a similar shortcoming. They suggest profound real-world implications from remarkably limited laboratory findings. In his wide-ranging book on cognitive psychology, Nobel Laureate Daniel Kahneman describes the priming experiments conducted by Kathleen Vohs in which subjects were shown stacks of Monopoly money on a desk or computers with screen savers displaying dollar bills floating in water. With these symbols priming their subconscious minds, the subjects were given difficult tests. The true test, however, came when one of the experimenters “accidentally” dropped a bunch of pencils on the floor. Apparently, those who were primed to think about money helped the experimenter pick up fewer pencils than those who were not primed. Kahneman asserts that the implications of this and many similar studies are profound. They suggest that “. . . living in a culture that surrounds us with reminders of money may shape our behavior and our attitudes in ways that we do not know about and of which we may not be proud.”13

  If the implications of such studies mean that American society is more selfish than other societies, then we would have to explain why Americans typically donate more of their time and more of their income to charities than do those of nearly any other nation.14 We would also need to explain why some of the wealthiest Americans, such as Bill Gates, Warren Buffett, Mark Zuckerberg, and a host of billionaires, have pledged to donate half of their wealth within their lifetimes.15 Surely these people were thinking hard about their money before they chose to give it away. We simply cannot draw sweeping conclusions from snapshots of data.

  I want to mention one other curious study from psychology. Its underlying assumption has much to do with how we behave during pattern breaks. Gerd Gigerenzer is the highly sensible Director of the Max Planck Institute for Human Development and an expert on both risk and intuition. Some of his work, which he related in a book titled Gut Feelings, was popularized in Malcolm Gladwell’s Blink. Gigerenzer has never been shy to point out perceived weaknesses and shallow logic in his own field. He has written cogently on the flaws embedded in Daniel Kahnemann’s and Amos Tversky’s heuristics and biases project.16 Yet even Gigerenzer h
as occasionally fallen into the “how silly are we?” camp, though the following topic he certainly did not take lightly. Unfortunately, this particular study suggests that Americans behaved irrationally after 9/11, though their reactions may have been perfectly sound.

  Gigerenzer found that American fatalities from road accidents increased after 9/11.17 Because many Americans were afraid to fly in the year following the attacks, they drove instead. Presumably, the increased number of drivers increased the number of collisions, leading to roughly 1,500 more deaths than usual. Gigerenzer’s main aim is prudent and wise. Governments should anticipate likely shifts in behavior following terrorist attacks and should take steps to reduce indirect damage such as greater accidents from changed behavior. But the underlying assumption is that many Americans cannot think rationally about probability. Gigerenzer implies that the decision not to fly after 9/11 was based on irrational fears. Had they continued to fly instead of drive, fewer Americans would have died.

  The problem with such reasoning, as you’ve likely already guessed, is that it ignores the pattern-break problem. A statistician might argue that, despite the 9/11 hijackings, the odds of dying in a plane crash were still extremely low. But those odds are based on a prior pattern—prior to a meaningful and dramatic pattern break. After 9/11, Americans had to wonder whether other terrorist plots using airplanes were still to come. If the terrorists could defeat our security checks once, could they do it again? Given that these were the acts of an organization and not of a single, crazed individual, and given that the leader of that organization vowed to strike America again, it was wise to adopt a wait-and-see approach. The past odds of flying safely no longer mattered in light of a potentially ongoing threat. Without any means of determining how great that threat would be, driving was a perfectly rational alternative, even knowing that one’s odds of dying in a car crash might rise. Until a new pattern is established (or a prior one returns), the odds of dying in a hijacked plane might be even higher.