Rebel Ideas- the Power of Diverse Thinking

by Matthew Syed

  The dubious power of the ad hominem has been revealed through extensive psychological research. One recent paper published in the Public Library of Science surveyed 39 college students and 199 adults. They found that when you attack someone’s character it undermines faith in their conclusions as powerfully as when you identify actual evidence questioning the basis of those claims. Playing the person rather than the ball works. In this sense, the ad hominem represents what economists call a free rider problem. All citizens benefit from the trust that is central to the functioning of democratic institutions, but this offers politicians an incentive to impugn the integrity of opponents, thus benefiting personally in electoral terms, but weakening the epistemic fabric upon which the collective intelligence of any democracy depends. Trust itself, as a consequence, starts to polarise.

  The ad hominem is not always fallacious, of course. If someone has consistently lied, or has a conflict of interest, then drawing attention to that is legitimate. The problem is when a person’s character is attacked not because of wrongdoing, but because she is an opponent; when a viewpoint is taken as prima facie evidence of bad faith. This kind of epistemological tribalism is not an impoverished form of informed debate, it is the antithesis of informed debate.

  Many Ancient Greek philosophers, not least Socrates himself, argued that the good functioning of a democracy is inextricably linked to the quality of its deliberations. It is only by testing ideas, by examining evidence, that we reach reasoned decisions. This is the lesson that emerges from our analysis of diversity science, too, as well as from formal theories that probe the conditions under which democracy leads to wise outcomes.FN2 And this is precisely why Socrates believed that it was crucial for citizens to be able to detect and punish fallacious reasoning. For Matthew Stevenson, one of America’s most fascinating modern citizens, this offers a ray of hope. ‘If public figures knew that constantly impugning the character of opponents would lose them credibility with their own side, they might engage on the evidence instead,’ he told me. ‘That would improve the tone of debate, and the quality of deliberations. If someone automatically attacks the trustworthiness of opponents, he should forfeit trust in himself.’

  As for Derek Black, after the initial stand-off with his parents, there has been a partial reconciliation. ‘We text, we call every now and again, I have been back three or four times for a day or two in the last five or six years,’ he has said. ‘Being able to communicate was more important than our political differences . . . My parents should take some credit for that.’ It is difficult to predict where the relationship will end, but it is not impossible that the guitar-playing student whose epistemic walls were dismantled by Matthew Stevenson will, in turn, help to dismantle the walls of his father. After all, there is not merely trust between Derek and Don, but love. Indeed, their discussion might one day lead to the most dramatic conversion of all: that of one of the highest profile white supremacists in modern political history.

  ‘They say that distrust is contagious,’ Matthew says. ‘Sometimes trust can be contagious, too.’

  6

  Beyond Average

  I

  So far in the book, we have examined the dangers of homophily, dominance dynamics and echo chambers. We have looked at the power of an outsider mindset and recombinant innovation. We have also seen how an understanding of diversity shines a light on everything from the failures of the CIA to benefits of constructive dissent. In this chapter, we are going to look at diversity from a fresh angle. We will examine how a conceptual flaw in the way we think about ourselves as human beings can undermine the influence of diversity, and prevent institutions and societies reaching their full potential. Indeed, we will see that this conceptual flaw has infiltrated many branches of science itself.

  To get to grips with this fallacy, and why it matters, will take a bit of digging. We will start with one of its most confounding manifestations, one that we are all familiar with; namely, the changing and often contradictory advice on diet and nutrition. This may seem to have little to do with the science of diversity. It certainly appears to be a long way away from what we have examined so far. Yet we will see that the tensions in dietary advice shine a light on an important facet of our world. And one with vast implications for our future.

  *

  Eran Segal was confused. In fact, it is probably more accurate to say that he was befuddled. For a brilliant scientist, who earned his PhD from Stanford University, this was not an agreeable feeling. The source of his confusion will be familiar to anyone who has spent more than a few moments contemplating diet and nutrition. What we eat is important to health, and can have a huge impact upon longevity, and yet the evidence remains perplexing.

  As an undergraduate, Segal played lots of handball and ate healthily, but was 40–50 pounds overweight. When he got together with Keren, his wife, whom he met at a party at twenty-three, his confusion heightened. Keren would soon become a clinical dietician, armed with the latest science. She cooked healthy meals, with plenty of fresh vegetables, largely following the guidelines of the American Dietetic Association, but Eran wasn’t shifting any pounds.

  ‘I decided to stop following the guidelines and look at the scientific evidence upon which they were based,’ Eran says when I chat with him and Keren. ‘It wasn’t what I expected to find. A lot of studies were based upon small samples of people. And a lot had been funded by food corporations, which made one sceptical about the results. It wasn’t as rigorous as I expected.’1

  As Eran is talking, Keren’s face breaks into a smile. ‘He is an easy-going person most of the time, but he cares a lot about data. This really got to him.’

  Perhaps Eran’s biggest surprise was that the evidence seemed contradictory. Some studies called for low fat diets. Others for high fat diets. Popular books eulogised a Paleo diet, or a Mediterranean diet, or an Asian diet, or some combination of the three, or some other latest craze, which would grip consumers, and then fade away, perhaps to return in another form, with a few twists.

  Or take carbohydrates. Some evidence suggests that low-carb, high fat diets can improve health, while others suggest that the best diets are low-fat, high-carb. There is evidence for both sides, which, in a sense, is evidence for neither. This isn’t just confusing; it is downright mysterious. It caused Segal’s frustration to harden into a deeper kind of fascination.

  Eran has a quizzical face, and his eyes flash as he talks about his intellectual journey. In many respects, he is like anyone else who has been befuddled by dietary advice. He lives with Keren, three kids, a dog called Snow and a cat called Blue, and he does his best to live responsibly. But he had one sizeable advantage in getting to the bottom of the conundrum: a world-class background in computation (in his twenties, he was the winner of the prestigious Overton Prize, awarded annually for outstanding accomplishment by a young scientist), and a post at the Weizmann Institute, one of the top academic institutions in the world. Segal says:

  It is no surprise that people are confused. In 2012, the American Heart Association and the American Diabetes Association suggested that people should drink diet soda for weight loss and health. There was then a huge rise in diet soda consumption, even though further research showed the opposite. In 1977, the US government said that fat was bad and fibre good, so people reduced fat and increased fibre. At much the same time, obesity trebled in men and doubled in women.

  This last point shows that Segal’s concerns are of more than theoretical interest. Diet has become a huge public health issue. If you live in the United States, you have a nearly 70 per cent chance of being overweight and 40 per cent chance of being obese. In the UK, the stats are similar. Obesity rates around the world have more than doubled since 1980 and in 2014 more than 1.9 billion adults, 39 per cent of the world’s population, were overweight, with 600 million obese. Segal says: ‘The obesity epidemic can’t have been helped by the fact that so many people are baffled by the advice. People who go on diets often put their weight back
on. In fact, there is a lot of evidence to suggest that dieting is associated with piling pounds on rather than taking them off.’

  A study conducted on The Biggest Loser, a dieting programme with huge ratings on US television, showed that contestants lost huge amounts of weight through exercise and calorie counting. And yet this drastic loss caused their metabolic rates to plummet to such an extent that, six years later, their metabolisms were so slow that they were not able to eat the same calories as people of the same weight who had never been obese. Scientists call this ‘persistent metabolic adaptation’. And yet this was just one of dozens of anomalies that Segal found as he probed the science. ‘There were certain agreed facts, such as that any diet should include fat, salt, protein, fibre, vitamins and minerals,’ he says, ‘but beyond that, almost anything seemed to go.’

  Another frustration emerged when he got into marathon running in his thirties and looked to see if diet could help him to improve his times. Sure enough, the dietary advice offered to runners is every bit as contradictory as that given to everyone else. He says:

  When I first started running, the big thing was ‘carb loading’ the night before a big race. It was an axiom of marathon running. I would typically eat three bowls of pasta the evening before a race, and then a few dates or energy bars thirty minutes before a run. At first, I didn’t question this advice, but after a while, I decided to take a closer look.

  The more he looked, the more perplexed he became. Some studies treated carbs as all the same; others suggested that there were ‘good’ carbs and ‘bad’ carbs. One study claimed that eating dates thirty to sixty minutes before running energised some, while others felt so exhausted that they had to stop a few minutes into their runs.

  ‘I decided to conduct an experiment on myself,’ Segal says. ‘One night, instead of eating multiple bowls of pasta, I ate a salad with lots of fat sources like avocado and nuts and tahini. The following morning, I did a twenty-mile run without eating anything at all.’ This contradicted mainstream advice. Indeed, many dieticians would have described it as self-sabotage. Yet he felt stronger, fitter and more primed than ever. He went on: ‘My energy levels were higher than with carbs, and my post-run hunger pangs disappeared. I inferred that my body had switched from burning carbs to burning fat, altering my energy and hunger.’

  In his thirties, he achieved his ambition of running a marathon in under three hours in Paris. In 2017, he ducked under three hours again, this time in Vienna. And yet he still hadn’t cracked his ultimate ambition of making sense of dietary science. ‘I couldn’t let it go,’ he says. ‘This was a mystery that had to be solved.’

  II

  In the late 1940s, the US Air Force confronted a mystery of its own. It was the early era of jet-powered flight, when engineering was supposed to have reached unprecedented levels of reliability, but the Air Force was enduring incident after incident. There were crash landings, unplanned dives, and much else besides. ‘The problems were so frequent and involved so many different aircraft that the Air Force had an alarming life-or-death mystery on its hands,’ Todd Rose, the Harvard academic, has written. ‘It was a difficult time flying,’ one retired airman said. ‘You never knew if you were going to end up in the dirt.’2

  To get a sense of the scale of the problem, consider US Air Force official records for February 1950. On the first day of the month, safety incidents were reported for Charles L. Ferguson in a C-82 Packet, a twin-engined, twin-boom cargo aircraft; Medford Travers in a P-51 Mustang, a long-range, single-seat fighter; Malcolm W. Hannah in a T-6 Texan, a trainer aircraft; and Herman L. Smith in a Boeing B-29. Others who suffered incidents included Harry L. McGraw, William K. Hook and George T. Shuster. Just to be clear, these all happened on the same day. But 1 February 1950 wasn’t an outlier, it was a typical twenty-four hours. On the second day of that same month, there were four safety incidents, on the third there were seven, and on the fourth there were four more. On the 14th of that month, there were sixteen separate incidents. In total in February 1950 there were 172 incidents.3

  What was going on? The problem didn’t seem to be with the mechanical or electronic systems of the planes. These had been thoroughly tested by engineers and found to be in good working order. Yet the incidents didn’t appear to be caused by a sudden deterioration in the skill of pilots, either. These were well-trained professionals, highly regarded within the industry.

  But if it wasn’t engineering or skill, what else could it be? Into the heart of this mystery stepped a Harvard graduate who specialised in physical anthropology. Lieutenant Gilbert S. Daniels was not a conventional airman: he was quiet and softly spoken, methodical and scientific, and listed gardening among his hobbies. In persona and interests he was not dissimilar to Eran Segal. And Daniels had a strong hunch. He believed that the problem wasn’t with engineering or pilot judgement, but with the design of the cockpit itself.

  The background is instructive. The cockpit design had been standardised in 1926 by the US Air Force after tabulating the dimensions of hundreds of airmen. This process uncovered the average physical traits of pilots, which were then used to determine the height of the chair, the distance to the pedals and stick, the height of the windscreen, the shape of the helmets, and so on. Some within the Air Force initially pondered if pilots had become bigger since 1926, making it difficult to operate the controls. Might that be the explanation for all the crashes? Daniels had a different hunch. He reckoned the problem wasn’t that the average airman had grown, but that the notion of the ‘average airman’ was itself defective. Perhaps there was no such thing as an average airman.

  In 1952, Daniels had the opportunity to test his intuition. He led a project at Wright-Patterson Air Force Base, which set out to measure the physical dimensions of pilots. Daniels threw himself into the task, carefully tabulating 4,063 pilots on 140 dimensions of size. These included ‘thumb length, crotch height, and the distance from a pilot’s eye to his ear’.4 He then calculated the average for the ten dimensions that he deemed most important when it came to cockpit design. In other words, he was tabulating the dimensions of the average pilot.

  But how many airmen would conform to this average? Daniels was quite generous in the way he calculated the latitude. If the measurement of a given pilot was within the middle 30 per cent of the range of values for a given dimension, he was deemed to be average. For example, the average height of the airmen was five-foot nine. So Daniels considered any particular pilot to be average in this dimension if he fell within the range of five-foot seven to five-foot eleven.

  Now, most experts in the military assumed that the majority of pilots would be in the average range across most of these ten dimensions. Indeed, this conclusion seemed obligatory. After all, the average was calculated from that very sample of airmen! Moreover (as Rose points out), these airmen had been preselected precisely because they conformed to the basic specification demanded by the Air Force. They would never have hired a five-foot four pilot in the first place.

  But what happened? How many pilots were within the average range across the ten dimensions? Zero. Not a single airman. From a cohort of more than 4,000, none were average. Daniels’s hunch was emphatically confirmed. The problem wasn’t that the average pilot had got bigger since 1926. The problem was that there was no such thing as an average pilot. As Todd Rose explains: ‘One pilot might have a longer-than-average arm length, but a shorter-than-average leg length. Another pilot might have a big chest but small hips.’ Even when Daniels picked out just three of the ten dimensions of size – say, neck, thigh and wrist circumference – less than 3.5 per cent of pilots were average on all three dimensions.5

  How is this possible? On the surface, the finding that no airman conformed to the average seems not just confusing but contradictory. If you take a group of people and calculate the average of some trait, it surely must tell you something about the individuals within the group. After all the average has been calculated from that sample of individuals! Yet it turns
out that focusing on averages can be misleading. Take the body length of weaver ants. There are two types of weaver ants – some very large, others very small. This means that if you take the average across all such ants, this average depicts none of the individual ants. The average is, in a certain sense, non-representative. This is called a ‘multimodal distribution’.

  Male height in humans, by contrast, conforms to a different type of distribution: the classic bell-curve. This means that most people do indeed cluster around the average. There are many men who are five-foot nine, but very few who are four-foot nine or six-foot nine. If you take a given person, you can be pretty confident he will be somewhat near the mean. But the design of a cockpit is not just about height. It is about human size across multiple dimensions. There is chest circumference, arm length, leg length, torso circumference, and so on. It is easy to assume that if you are high on one dimension (say, neck circumference) you will be high in another (waist size). But these correlations are, in fact, weak. This means that any metric that averages across these dimensions will obscure the diversity.

  Take the two men in Figure 9. If you took the average of all nine dimensions, they would be pretty much identical. And yet the man on the left is heavier, shorter, has narrower shoulders, a thicker neck, a shorter reach, a wider waist and narrower shoulders, while the man on the left is taller, lighter and so on. Summed across dimensions, they are both close to average. On any given dimension, however, they are often far from average.

  We can see this point in a different way by taking another metric: IQ. It is easy to assume that if you take two people with an IQ of, say, 105 they will have similar scores on each of the various components of IQ, such as vocabulary, problem solving, etc. In fact, as Rose has noted, the correlations are weak. You could be high on matrix reasoning, low on knowledge, medium on block design, high on symbol searching and low on encoding – or vice versa. The single IQ metric is not expressing this variation, it is concealing it. And in most areas of performance, it is the variation that matters.6