Cascades
Page 9
Katz and Lazarsfeld concluded that the real reason mass media was effective in forming opinions wasn’t simply because it reached a lot of people, but because of its ability to influence the influencers. Their work eventually became a core principle for generations of marketers and political consultants: reach those whose opinions can affect the opinions of others and you can sell just about anything. Inspiring word of mouth became a fixture of every political and promotional campaign.
Subsequent studies revealed three characteristics of opinion leaders: the personification of values (who one is); competence (what one knows); and strategic social location (whom one knows). We’ll look at each in turn:
1. Personification of values (who one is). The studies found, not surprisingly, that influence isn’t a universal quality. While a middle-aged mother might wield influence in such areas as what products to buy at the market (remember, these studies were done in the 1950s), her daughter may be more persuasive in areas such as fashion and what to watch at the movies.
2. Competence (what one knows). It was also found that people who were seen to be more competent had more influence, but this was highly situational. For example, doctors who were seen to be more “scientific” were often able to form the opinions of their peers, while young people who go to the movies often and can rattle off the works of acclaimed directors are seen to be knowledgeable and influential in that realm.
3. Strategic social location (whom one knows). Social connections play a role in two ways. First, people tend to trust others they feel are part of their group, a principle that seems to be fully grasped by politicians whose accent changes slightly depending on which part of the country they happen to be in at any given time. This phenomenon is often referred to as homophily. Second, those who travel outside the group are seen as being privy to information others don’t have and can therefore leverage the knowledge they gain to persuade others within their group. So a person who is seen as part of a tight-knit group, but also has links to the outside world, can be particularly influential.
All of this seems perfectly plausible and in accordance with our daily lives. We all have people we go to for information in particular spheres of our lives: The niece who is a veritable encyclopedia of the best mobile apps to use, the uncle who is glued to CNN and knows all there is to know about politics, the friend from college who knows just where to take a particularly promising date. So it’s easy to believe that those same people can lead to large-scale change. However, that’s not how the world really works.
DEBUNKING THE INFLUENTIALS MYTH
* * *
In a very large sense, Malcolm Gladwell’s Law of the Few can be seen as an update of Katz and Lazarsfeld’s decades of research. We can easily see how Connectors, Mavens, and Salesmen have “rare qualities” that make them adept at influencing others. However, in ascribing these personal qualities to large-scale social phenomena, Gladwell misunderstood the data.
The problem with his idea is that to create a significant impact, you need a lot of people to buy into an idea. So, while a Connector, a Maven, or a Salesman may be able to influence those around them, those people would have to influence others, and those people, in turn, would have to influence others still. To create something like the Orange Revolution, it takes literally millions of people to join in. So, if the Law of the Few were true, we would expect people with “rare social gifts” not only to be able to persuade those around them, but to create long chains of influence more reliably than a normal person would. That’s a very testable hypothesis, and researchers who have studied it have found it wanting.
One e-mail study that incorporated 60,000 respondents to replicate Milgram’s small-world research found that, while the principle of six degrees of separation held true, messages didn’t need to go through “hubs” (highly connected people) to reach their recipients.6 Another study of 1.6 million Twitter users found that, while highly connected users were slightly more likely to initiate large viral cascades, the difference between them and normal users was so small that it’s probably not economically feasible to waste resources seeking them out.7
So the empirical evidence debunks Gladwell’s Law of the Few. Not only are people with “rare social gifts” not sufficient to initiate a social epidemic, they aren’t even necessary. Just because someone has a lot of connections to other people, specialized knowledge, or ample powers of persuasion doesn’t mean that he or she is more likely to initiate messages that get passed on by long chains of people.
Beyond the scientific evidence, there is a more compelling reason to discount Gladwell’s Law of the Few. Does it stand to reason that Slobodan Milošević in Serbia or Viktor Yanukovych in Ukraine lacked influence or access to influential people? That certainly doesn’t seem right.
On the other side of the ledger, if those who played a part in the Orange Revolution and the Arab Spring had magical powers of influence, why did these powers fail them in the aftermath? The protestors that overthrew Mubarak in Cairo were overrun by the Muslim Brotherhood in the elections that followed the uprising. What happened to their “rare set of social gifts” then? The members of the Pora student movement, which played such a central role in the Orange Revolution, also played little, if any, further part in public affairs in Ukraine. For that matter, the leaders of the Boston tech firms were highly influential within the tech industry until they were steamrolled by the mostly anonymous (at the time), but highly interdependent, Silicon Valley entrepreneurs. Clearly, if we’re going to understand how ideas spread and create change, we’re going to need a different model based on the best available empirical evidence.
To understand why people who are influential personally often fail to create widespread change, let’s go back to my friend Robbie. Sure, he could do or say something goofy in the locker room, and the next thing you knew it seemed like everybody was repeating it, but we were all deeply connected in a school environment. The kids in the next town over were oblivious to Robbie’s charms and therefore completely unaffected by the memes floating around our school.
What belies the theory of “influentials” or “opinion leaders” or whatever you want to call them is that change is a matter of networks and not nodes. While it is easier and more cognitively pleasing to single out individuals, the truth is that it is the connections between people that are decisive. True power does not lie at the top of hierarchies, but emanates from the center of networks. Moreover, as digital technology enables connections to form faster and more pervasively, networks are becoming even more salient and essential to create transformational change.
So while we always need to be careful when considering the explanatory power of any theory,8 I do believe that there is a body of research that much better fits the data than Gladwell’s Law of the Few which can help us understand how cascades form, how they can result in transformational change, and how we can harness many of these same forces, albeit imperfectly, to bring change about.
THE INFLUENCE OF LOCAL MAJORITIES
* * *
When you go to a new country, or even just enter a new business environment, one of the first things you’ll notice is that people do things much differently than you’re used to. Sometimes these differences are startling and you find yourself objecting vigorously, but when challenged you can’t really say why you have a problem with the seemingly strange custom. It just feels wrong to you. After a while, though, what once seemed alien and exotic starts to feel normal. You begin to conform to your local environment.
Solomon Asch noticed this phenomenon when he was a small boy at a Passover Seder. When he asked his uncle why an extra glass was being left on the table, he was told that it was for the prophet Elijah. When the time came and the door was opened for the legendary prophet, the young boy actually thought he saw the level of wine in the glass drop ever so slightly.9
Later, when he became a prominent social psychologist, Asch wanted to study the effect of suggestion and expectation. So he designed an ingenious
experiment. A group of seven to nine subjects were brought into a room and told that they would be evaluating the length of lines. They were then shown pairs of cards like the ones in Figure 3.1.
FIGURE 3.1 Cards from Asch’s Experiment
* * *
The subjects were asked to choose which line on the right matched the line on the left. The answer was made to be deliberately obvious, but as the administrator went around the room, every person answered that “A” was clearly the same length as the sample line. The last person in the group (in fact, the only real subject, the rest were collaborating with the researchers) paused nervously and then slowly stuttered out that he too agreed that “A” was the correct answer.
The experiment was run 123 times, with confederates purposely giving wrong answers to 12 out of 18 trials. Subjects gave incorrect answers 36.8 percent of the time compared to only 1 percent of the time under ordinary circumstances.10 The experiment was replicated many times by Asch as well as others, and while the precise conditions and results would change slightly, the conclusion was always the same—we have a natural tendency to conform to a local majority.
In a sense, this shouldn’t be surprising. The idea that people have a tendency toward conformity is nothing new. But that people would give obviously wrong answers to simple and unambiguous questions was indeed shocking.
Asch’s experiment should give us some pause in how we think about influence. While much discussion has focused on influential individuals, groups exert considerable sway that should not be ignored. In fact, in their study of opinion leaders, Katz and Lazarsfeld found a large degree of “political homogeneity” among groups of family and friends, but attributed it to an individual’s influence rather than a group dynamic. They also reported that many people who changed their opinions during the research were actually just coming back into line with their peers.11 So while the studies gave rise to the concept of “opinion leaders” and, indirectly, to Gladwell’s Law of the Few, it appears that a group dynamic would fit the data at least as well. While the evidence was sound, the conclusions were not.
THE THRESHOLD MODEL OF COLLECTIVE BEHAVIOR
* * *
One important aspect of Asch’s conformity studies was that the results were far from uniform. A quarter of the subjects never conformed, some always did, and others were somewhere in the middle. We all have different thresholds for conformity based on our own personal characteristics, confidence in our knowledge of the subject at hand, our opinions about other people in the group, and so on.
We are always influenced by people around us, but not in a uniform manner. It varies. Maybe we are slaves to fashion in the clothes we wear, but more independent in the books we choose to read. If everyone we know reads a certain novel, we might succumb, but it would have to be an overwhelming majority. Most of the time, we are keeping an eye on both our inner compass and our peer group, considering both our own preferences and those around us.
Mark Granovetter, whom we met in the last chapter in connection with his theory of “the strength of weak ties,” proposed a model for understanding group dynamics as a series of thresholds of resistance. As a thought experiment, he asks us to imagine a diverse group of people milling around in a square. Some are natural deviants, always ready to start trouble, most are susceptible to provocation in varying degrees, and the remainder is made up of unusually solid citizens, almost never engaging in antisocial behavior.12
A graphic representation of two possibilities of Granovetter’s scheme can be found in Figure 3.2. In the example on the left, a troublemaker with a zero threshold decides he wants to throw a rock and break a window. His friend with a slightly higher threshold sees him, and that’s enough for him to throw a rock too. As more and more people join in, the threshold of others in the group is met, and before you know it, everybody is engaging in a full-scale riot.
FIGURE 3.2 Two Possibilities of Granovetter’s Threshold Model
* * *
The example on the right is slightly different. After the first two troublemakers start, there is no one around with a low enough threshold to join in. Rather than the contagion spreading, it fizzles out, the three miscreants are isolated, and little note is made about the incident. In the first case, news reports the next day would be likely to attribute the full-scale riot to a sociological phenomenon, such as race, class, political discord, or some other external factor. However, the real cause had more to do with a slight difference in the distribution of resistance thresholds.
Granovetter pointed out that the model can be generalized to a variety of social situations, such as whether to adopt a new technology, leave a party, go to college, join a strike, and so on. We naturally take social cues from those around us, whether they have “rare qualities” or not. Majorities don’t just rule, they also influence.
HOW CASCADES FORM
* * *
In the last chapter, we discussed the issue of cascades. Local problems tend not to stay local, but spread through a network of connections and interdependences. Just as it would be nonsensical to describe the first power line that failed in the 1996 blackout as having “rare qualities,” it is unreasonable to attribute informational cascades, such as business fads, financial contagions, and political ideas, solely to individuals. The structure of the connections within the network matters—a lot.
We can, in fact, take Granovetter’s threshold model further by combining it with his principle of the “strength of weak ties” to understand how the threshold model can lead to viral cascades. Take a look at Figure 3.3 on the next page. Now, instead of one group of people, we have three. The one at the top, marked “A,” is identical to the group in the previous diagram that resulted in a riot. Now, however, we add two new groups, “B” and “C,” that are weakly connected to the first group. As the contagion infects the entire first group, it quickly spreads to highly susceptible members of the other two groups, and before you know it, a large cascade ensues. Physicists have a name for this type of phenomenon—percolation—and a configuration like the one in Figure 3.3 is called a percolating cluster.
FIGURE 3.3 A Percolation Model with Weak Ties
* * *
Notice that no individual in this model has any more influence than anybody else. Small groups, loosely connected, but united by a common purpose give rise to a cascade regardless of the personal qualities of the individuals involved. Once again, it is not the nodes, but the network that drives transformational change. That may not be as simple and intuitively pleasing as Gladwell’s “Law of the Few,” but it is the explanation that empirical research supports.13
The threshold model of cascades also provides a helpful explanation of the events I witnessed during the Orange Revolution. The tent city phenomenon got started at one university in the form of student meetings, soon spread to others in Kyiv, and eventually spread across the country through participants’ personal connections such as their parents, aunts, uncles, and neighbors. As the idea spread, others with higher thresholds—like my fiancée, our friends, and me—began to join in. We weren’t convinced to act by people with special qualities, but by those close to us. They were friends, coworkers, neighbors, and family members. In Otpor’s Serbian revolution, a crucial component of the movement was the mothers who saw their children arrested for taking part in peaceful protest. Again, it is not special qualities of individuals that spread an idea, but vulnerable clusters influencing other (often slightly less) vulnerable clusters that drives a cascade.
Another interesting aspect of the threshold model of cascades is that there is no reason that thresholds need to be static, intrinsic qualities of individuals. Instead, thresholds can change over time and according to environmental factors as well as to differences in how the idea is presented.14 In the heightened environment of the Orange Revolution, the lowering of thresholds became palpable and contributed to the abundance of viral memes discussed in Chapter 1—the horn protest outside of the Central Election Commission, the no-drink
ing rule, and the Razom Nas Bahato chant being three of the most salient examples. With revolution in the air and millions of people in the streets, it didn’t take much for something to catch on. Everybody felt connected and wanted to be a part of what was happening.
So now that we understand in theory how cascades effect change, let’s take a more practical look at how they play out in the real world.
THE DIFFUSION OF INNOVATIONS: A NETWORK VIEW
* * *
Let’s put ourselves in John Antioco’s place in 2000. He had been a phenomenally successful retail executive, pulling off turnarounds of major brands like Circle K and Taco Bell, and he had thrived in his role of CEO at Blockbuster, leading it to market dominance and serious profitability. When the Netflix team flew to Dallas to propose a partnership to his team, their company was little more than a start-up renting movies through the mail. The deal that Netflix offered, to basically license Blockbuster’s brand name for the web, was not attractive. In fact, when Toys “R” Us entered a similar deal with Amazon, it proved to be a disastrous mistake. At the time, Netflix had little to offer him and his company.
Seemingly more questionable was his dismissal of Netflix as a niche product. After all, every product begins with a small following—the question is whether it has the potential to grow beyond that. Yet if Antioco had reacted to every potential disruption, he would have had to consider not only Netflix, but every niche competitor, as a serious threat. That too would have been problematic. After all, if Antioco rang alarm bells every time anybody came up with a new way to rent videos, he wouldn’t have been able to run his company in the orderly, disciplined way that had made him successful in the first place.