PDF Summary:Rebel Ideas, by Matthew Syed

Many people think of diversity in the context of social justice. But Matthew Syed takes a different approach, exploring how diversity boosts group performance and intelligence. In particular, he argues that we should foster diversity because diverse groups are more collectively intelligent than homogeneous ones.

In his 2019 book, Rebel Ideas, Syed explains why cognitively diverse groups are superior to homogeneous ones. Cognitively diverse groups, he argues, can harness the diverse knowledge of their members, leading to increased performance and innovative ideas. Homogeneous groups, by contrast, often suffer from collective blindness, as members simply reinforce each other’s views rather than challenge them.

In this guide, we’ll examine Syed’s arguments that diverse groups are higher-performing than homogeneous groups, while also discussing the communication styles that can either amplify or diminish the value of cognitive diversity. Throughout this guide, we’ll also discuss actionable advice for creating cognitively diverse groups and consider counterarguments to Syed’s views.

(continued)...

To illustrate, Syed examines Angela Bahns’s 2009 study of universities in Kansas. Among them was Kansas University (KU), which had a diverse population of 27,000, as well as five smaller, less diverse colleges with an average population of 1,000. Despite KU’s diverse population, the study revealed that KU students’ social networks were the most homogeneous.

(Shortform note: In addition to finding that students at large, diverse schools had the most homogeneous social networks, Bahns’s study reported that students in more diverse social networks reported having closer relationships. This finding opens the possibility that closer friendships develop among diverse students, rather than similar ones.)

In his discussion of the study, Syed notes that although KU is diverse, its size allowed students to find other like-minded individuals. So, although KU is the most diverse university, its social networks are the least diverse.

By contrast, although the smaller universities were less diverse on the whole, students had fewer opportunities to find peers exactly like themselves. Accordingly, they had to accept some degree of diversity in their social networks, however minimal. Thus, these smaller, less diverse universities actually had the most diverse social networks.

(Shortform note: Researchers have proposed and tested strategies for promoting diversity and inclusion at US universities. For example, they note that social norms messaging—messages that explicitly affirm the value of inclusion, whether in the form of posters or videos—promotes inclusive behavior. Additionally, they find that intergroup contact intervention—intentionally creating diverse classroom groups—leads to long-term increases in inclusive behavior.)

Example: The Case of Derek Black

To demonstrate this paradoxical result, Syed discusses the case of Derek Black, a formerly prominent white supremacist. Syed argues that, although Black grew up in an echo chamber—his family was heavily involved in the Ku Klux Klan (KKK)—his experience at a small university forced him to encounter diverse perspectives, eventually leading him to renounce white supremacy.

(Shortform note: In one respect, Black’s experience is representative of a broader trend: Research finds that experiencing higher education has a significant correlation with changes in students’ political views. However, because students who attend college themselves tend to be from higher-income families, researchers caution that there might be other variables impacting these changes. In other words, it’s not clear that college causes students’ political views to change, despite being correlated with a change in political views.)

In particular, Syed notes that Black quickly met Matthew Stevenson, an Orthodox Jew who also attended New College in Florida. Even after learning of Black’s association with white supremacy, Stevenson didn’t abandon him; rather, he invited Black over for weekly Shabbat dinners, where diverse groups of people shared a meal.

Although Black’s upbringing had eroded his trust in those outside of white supremacist circles, Syed claims that Black’s relationship with Stevenson slowly restored this trust. He became more receptive to alternate views and decided to engage seriously with arguments undermining white supremacy. Ultimately, this led Black to renounce his racist beliefs and with them the echo chamber of his upbringing.

(Shortform note: Black wrote an article, published by the Southern Poverty Law Center in 2013, detailing his decision to abandon white nationalism. In particular, Black asserted his newfound view that white nationalism was a movement primarily committed to retaining power for white Americans—and thus committed to oppressing racial and ethnic minorities—rather than one founded on legitimate science.)

Danger 3: Standardization

Finally, Syed discusses the dangers of homogeneity in the context of standardization—the practice of designing systems to fit “average” individuals rather than tailoring systems to individual differences. Syed argues that standardization forces us to conform to uniform molds, creating systems that are less effective. Although Syed considers various areas in which standardization has been harmful, we’ll examine two: airplane cockpits and dieting.

(Shortform note: Standardization is also prevalent in the US education system, with the Common Core State Standards outlining specific learning goals for students at each grade level. Although 46 states initially adopted Common Core in 2011, it came under criticism from experts who argued that it set unreasonable expectations for students in lower socioeconomic areas, who face unique challenges. So, Syed’s general criticism of standardization resonates with respect to Common Core: It doesn’t acknowledge the diverse needs of students.)

First, Syed explains that around 1950, the US Air Force experienced an unusually high rate of safety incidents. At the time, cockpits were standardized according to the dimensions of the average pilot, which determined chair height, separation between pedals, and so on.

Lieutenant Gilbert Daniels, however, realized that few pilots conformed to this “average” cockpit, because they aren’t average across every dimension. To demonstrate, Daniels measured pilots across 10 dimensions to see how many are within the average range for all dimensions. Out of more than 4,000 pilots, zero fell within the average for all 10 dimensions. So, cockpits didn’t fit any individual pilot perfectly.

After engineers redesigned the cockpit so pilots could make individual adjustments, the incident rate dropped significantly. In turn, Syed concludes that adjusting for diversity among pilots optimized the cockpit design.

(Shortform note: In The End of Average, Todd Rose argues that Daniels’s revelation about cockpit design suggests a broader conclusion—any system design around the “average individual” is bound to fail. Consequently, Rose claims that the notion of “average” in various domains, such as school grades and employee performance reviews, is counterproductive at best, and harmful at worst.)

Additionally, Syed tells the story of Eren Segal, a scientist puzzled by advice on dieting. Upon reviewing the science, Segal found contradictory results; some studies, for example, claimed high-carb diets were optimal, while others endorsed low-carb diets.

Segal, however, realized a possible explanation for this disparity: Diet science focused on average responses to diets, rather than more personalized metrics. Consequently, diet studies didn’t consider the possibility that different diets are optimal for different individuals.

(Shortform note: According to experts, another reason for the abundance of conflicting diet advice stems from the source of diet studies. In particular, experts caution that many diet studies are funded by the food industry itself. Consequently, many researchers try to deliver results that benefit the industry that funds them, which encourages misleading conclusions.)

To test this possibility, Segal designed a study in which one group ate white bread for a week, and one ate sourdough, after which they switched diets. The results showed that, on average, there was no difference in blood sugar levels whether participants ate white or sourdough bread. The individual results, however, revealed the opposite: Some participants had healthier blood sugar levels after consuming white bread, and others had healthier blood sugar levels after consuming sourdough. Relying on average data obscured this result.

Later studies, according to Syed, confirmed Segal’s suspicion: Individuals often respond poorly to standardized diets, because standardized diets don’t recognize that dieting results differ significantly by individual.

(Shortform note: In The Personalized Diet, Segal and co-author Eran Elinav outline their further research on the value of diets tailored to individuals. They also explain how people can test their own blood sugar responses to different foods, and create a diet best suited to their needs.)

The Advantages of Cognitive Diversity

Alongside the harms of homogeneity, Syed discusses the specific benefits of cognitively diverse groups. In this section, we’ll discuss two such benefits: the wisdom of crowds, and greater capacity for innovation.

How Cognitive Diversity Harnesses the Wisdom of Crowds

According to Syed, cognitively diverse groups benefit from the wisdom of crowds: They become more collectively intelligent as their diverse perspectives create greater collective knowledge.

(Shortform note: The wisdom of crowds is often applied to stock markets in particular, as experts observe that the collective decisions of all market participants—which stocks to buy, and which to sell—create stock prices that reflect our collective knowledge. Consequently, market prices capitalize on the collective intelligence of individual investors.)

To illustrate, Syed cites Jack Soll’s study examining the accuracy of 28,000 economic forecasts. This study, Syed relates, found that top economists are about 5% more accurate than average economists.

However, the study also averaged the top six economists’ predictions, as if they were members of a single team. These results, Syed notes, were impressive: The average prediction of the top six economists was 15% more accurate than that of the top individual economist.

(Shortform note: In The Wisdom of Crowds, James Surowiecki discusses various other cases in which averaging group members’ estimates yields more accurate predictions than those of the best individuals. For example, he tells the story of Francis Galton, a British scientist who visited a fair where 800 people guessed the weight of an ox; the average prediction was 1,197 pounds, while the ox itself weighed 1,198 pounds. According to Surowiecki, this represents a broader trend: The aggregate judgment of non-experts is often more accurate than individual judgments of experts.)

Explaining these results, Syed argues that each economist’s model—like each of our perspectives—suffers from blind spots. Consequently, no individual model can capture the intricacies of the entire economy. However, when top models are combined, Syed says the result better captures these intricacies. In turn, they provide a more holistically accurate view of the economy.

Nonetheless, Syed concedes that cognitively diverse groups aren’t guaranteed to enjoy group wisdom—a group of cognitively diverse but incompetent individuals, for example, is unlikely to be collectively intelligent. But, when competent individuals without overlapping blind spots form a group, Syed asserts that their collective intelligence will rise above the level of any individual.

Are Crowds Always Wiser?

Although crowds are typically wiser when their members have some information about the relevant topic, the quality of the information matters too. Crowds actually become less collectively intelligent when their members are poorly informed.

To show as much, consider Condorcet’s Jury Theorem, which explains how the wisdom of crowds arises in the first place. This theorem states that, if every group member has above a 50% chance of correctly answering a yes-or-no question, then the larger the group, the more likely that the majority answer is correct. Although Condorcet’s theorem originally applied to yes-or-no questions, the logic also applies to questions with multiple answers: If group members are more likely to correctly choose the right answer than any wrong answer, then as the group gets larger, the plurality’s answer is more likely to be correct.

Consequently, groups with some knowledge about the question at hand become more likely to arrive at the correct answer as they add knowledgeable members. However, the converse is also true: If individual members are more likely to be wrong, then as the group grows larger, the chance that the plurality’s response is correct gets increasingly lower. For instance, because US citizens had little information about President George W. Bush’s preferred nominee to the Supreme Court, they failed to correctly predict John Roberts’ nomination.

How Cognitive Diversity Fosters Recombinant Innovation

Syed also discusses the impact of cognitive diversity on innovation. In particular, he argues that cognitive diversity—both in individuals and in institutions—leads to recombinant innovation, because it brings ideas from disparate fields together.

Syed explains that there are two types of innovation: incremental and recombinant. Incremental innovation refers to small improvements that occur within a field. For example, incremental innovation occurs when Apple makes gradual improvements to the iPhone, like changing its dimensions. By contrast, recombinant innovation occurs when two ideas from different fields come together. For instance, the creation of the iPhone represented recombinant innovation, as Apple combined ideas about telephones with those about computers.

(Shortform note: Although recombinant innovation can result in broader change, experts argue that incremental innovation also offers crucial benefits. For instance, while recombinant innovations carry a high risk of failure, incremental innovations are much lower risk, delivering more secure returns. Moreover, because recombinant innovation often takes longer to implement, incremental innovation can help keep your customers engaged in the meantime.)

Though both forms are important, Syed asserts that recombinant innovation is the driving force behind dramatic change, and he focuses on how individuals and institutions can foster recombinant innovation.

As for individuals, Syed points out that immigrants are inclined toward innovation: For example, over half of US-based Nobel Prize recipients were born abroad. According to Syed, this occurs because immigrants’ exposure to different cultures helps them question assumptions and fuse different ideas. In turn, they’re more likely to come up with recombinant innovations.

(Shortform note: In addition to being more innovative themselves, a 2023 study finds that immigrants’ presence also makes native-born inventors more innovative. In other words, native inventors who collaborated with immigrants had higher innovation output, on average.)

As for institutions, Syed argues that they must allow an open flow of information among diverse individuals to cultivate recombinant innovation. In particular, Syed argues that sharing ideas leads to information spillover: Innovative ideas become stronger as they’re exposed to more people, who further refine and improve them. Moreover, because diverse groups propose more distinct improvements to innovative ideas, they enjoy greater benefits from information spillover. So, Syed concludes that we shouldn’t view innovation only as the product of individual creative thinkers, but also of institutions that encourage these thinkers to interact.

(Shortform note: To foster information spillover in the workplace, experts recommend building a transparent environment that encourages collaboration. This could include rewarding employees who freely share their ideas, whether in performance reviews or even through a simple shout-out. Additionally, because sharing ideas in large groups can feel intimidating, others recommend holding smaller meetings with an emphasis on open conversation.)

To illustrate this point, Syed compares two technological hubs in the 1970s: Massachusetts’s Route 128 and California’s Santa Clara Valley, later called Silicon Valley. Syed observes that, in 1975, Route 128 had many advantages over Silicon Valley, such as cheaper land and lower wages. But, geographical distance and a tendency toward secrecy made Route 128’s companies insular, and ideas only traveled within individual companies. By contrast, Silicon Valley’s companies were densely concentrated and socially interconnected; engineers from different companies often spent time together, fostering cognitive diversity and widespread information spillover.

(Shortform note: Experts observe that the practice of “knowledge sharing” is widespread in Silicon Valley: Engineers have high turnover rates as they frequently switch companies, often bringing new ideas and product development plans with them. Consequently, Silicon Valley’s technology companies regularly develop similar products, leading to widespread success when any individual product succeeds.)

Syed attributes Silicon Valley’s later success and Route 128’s later failure to the information spillover that only Silicon Valley benefited from. The upshot of this difference is clear: Today, Silicon Valley is the home of tech giants like Apple, Google, and eBay, while many of Route 128’s tech giants have since shuttered.

Communication Within Diverse Groups

Finally, Syed discusses how the communication structure of a group can either undermine or enhance the value of cognitive diversity. In this section, we’ll discuss how dominance hierarchies lead to the suppression of diverse voices, while prestige hierarchies ensure that diverse voices are heard.

How Dominance Hierarchies Negate Cognitive Diversity

According to Syed, dominance hierarchies in which leaders rule via fear and threats are a common feature of human societies. He argues that these hierarchies suppress views that differ from the leaders’, thus reducing the collective intelligence of cognitively diverse groups.

Syed says dominance hierarchies played an important evolutionary role, as groups with dominant leaders were more likely to survive. Because prehistoric societies were typically faced with simple decisions, having a dominant leader make decisions was effective and more efficient. By contrast, groups that squabbled over straightforward decisions were less efficient and thus less likely to survive.

(Shortform note: Beyond bolstering group survival rates, researchers argue that dominance hierarchies led directly to humans’ greater cognitive capacities. Participating in a dominance hierarchy required us to distinguish between different ranks, understand the behaviors permitted for our own rank, and decide whether to change our rank. So, the individuals who succeeded in dominance hierarchies possessed the cognitive capacity needed to make these judgments, which they then passed on to future generations.)

However, Syed argues that dominance hierarchies are actually harmful in multifaceted situations, in which dominant leaders alone can’t know all the information in the problem space. In such situations, Syed says dominant leaders view diverse opinions as threatening, so they intimidate subordinates with diverse views into silence.

Diverse groups become no more effective than homogeneous ones, as subordinates don’t voice their opinions. Instead, Syed notes that they seek to please the leader by mirroring whatever the leader says. For example, Syed says that in business meetings, dominant individuals’ opinions quickly rise to the forefront, and different opinions of other team members are rarely voiced; rather, team members simply echo the leaders’ opinions.

Steps to Prevent the Groupthink Created by Dominance Hierarchies

Groups with dominance hierarchies often suffer from groupthink—they make irrational decisions because group members would rather conform than offer dissenting opinions. In Super Thinking: The Big Book of Mental Models, Gabriel Weinberg and Lauren McCann offer various strategies for companies to mitigate groupthink, including:

• Appoint a devil’s advocate—someone who argues against proposed ideas to make sure they’re being critically evaluated. The devil’s advocate can often find flaws with the idea in question.

• Divide meetings into independent subgroups, which pitch their ideas to the entire team afterward. These subgroups ensure that various ideas and perspectives are represented.

• Seek out employees with different opinions, both within the company and when hiring new candidates. These employees help you establish a more diverse pool of ideas to pull from.

How Prestige Hierarchies Harness Cognitive Diversity

As an alternative to dominance hierarchies, Syed examines prestige hierarchies, whose leaders are followed freely out of respect, rather than fear. According to Syed, leaders of prestige hierarchies listen carefully to diverse perspectives, maximizing collective intelligence.

(Shortform note: In a similar vein, experts also argue that groups need psychological safety—an environment where members don’t fear embarrassment or punishment for offering “wrong” ideas—to maximize collective intelligence and success. In such groups, members don’t blame each other for mistakes, but view mistakes with curiosity and an open mind. Therefore, psychologically safe groups actually benefit from discussing flawed ideas, as they learn from them.)

Syed cites anthropological findings that certain tribes had leaders who earned respect from peers through their wisdom and who sought to uplift these peers rather than strike fear into them. Such leaders, he notes, freely shared their knowledge with others and didn’t feel threatened by opposing voices.

Syed argues that, from an evolutionary perspective, prestigious leaders who shared their knowledge created groups where generosity was prized. Consequently, leaders themselves benefited from the open flow of information. And because prestigious leaders share their knowledge, multiple prestigious leaders could exist peacefully within a group.

(Shortform note: Because individuals attempt to imitate prestigious leaders, individuals have to quickly identify these leaders in the first place. To do so, researchers suggest that they often rely on social proxies for prestige, such as academic credentials or other awards. In other words, these researchers say that prestige is socially transmitted.)

As a result, Syed says prestigious hierarchies are better at harnessing the power of cognitive diversity during the decision-making process. However, he concedes that dominance hierarchies are useful once a decision is made, at which point dissent impedes the execution of the decision.

(Shortform note: Because dominance hierarchies and prestige hierarchies are best suited for different situations, experts say superior bosses alternate between dominant and prestigious leadership styles, depending on the context. For example, bosses might adopt a dominant approach when a clear business strategy is in place—one that requires a unified team. By contrast, during the discussion stage bosses might adopt a prestigious style to convey that they value subordinates and their input.)