Reviewed by Cass R. Sunstein
The New Republic Online
A few years ago, Daniel Kahneman, David Schkade, and I were involved in several studies of punitive damage awards by juries. We began by asking one thousand or so demographically diverse people to register their judgments about misconduct by various wrongdoers. We asked them to rate their moral outrage on a scale of zero to six, where zero meant "not at all outrageous" and six meant "exceptionally outrageous." We also asked them to come up with an appropriate dollar award.
To study their answers, we used a simple technique, in which individual responses are pooled to produce "statistical juries," whose verdict is the judgment of the median member. Having done this, we found that small groups of six people, or statistical juries, will agree about outrageousness and appropriate punishment on the scale of zero to six. If one group says that the conduct deserves a five, other groups are highly likely to agree. By contrast, statistical juries show a high degree of variability with respect to dollar awards. The dollar judgment of one jury is not a good predictor of the dollar judgment of other juries.
This study suffered from a serious problem: it did not involve group deliberation. If the goal is to understand how juries really behave -- or more ambitiously, how outrage develops in the real world -- this is a major defect. To remedy it, we conducted a follow-up study, involving about three thousand juryeligible citizens and five hundred deliberating juries, each consisting of six people. Here is how the experiment worked. Every juror read about a personal injury case, including the arguments made by both sides. Jurors were also asked to record, in advance of deliberation, their individual judgments on a bounded numerical scale, and also in terms of dollars. Next they were asked to deliberate together to reach a verdict, both on the bounded scale and on the dollar scale. Our goal was to discover the relationship between people's individual judgments, in advance of deliberation, and the ultimate views and actions of group members who have discussed the matter.
You might predict (as I did) that deliberation would lead to compromise, and hence that the verdicts of juries would be pretty close to the median of punishment judgments of jurors; but your prediction would be badly wrong. It turned out that the effect of deliberation was to create a "severity shift." When people began with a lot of outrage, their interactions made them significantly more outraged than they were before they started to talk. And with dollar awards, the severity shift was especially large. The ultimate award of juries was usually higher than the award favored by the median juror in advance of deliberation. In many cases, the jury ended up with an award at least as high as the highest award favored, in advance, by any of the jury's members.
We may draw two conclusions from these studies. First, judgments about appropriate punishment, and the right response to bad acts, are rooted in moral outrage. Second, social interactions greatly increase outrage and people's desire to punish in response to it. If like-minded people share a degree of outrage and get together with one another, they are likely to move in extreme directions.
When juries impose high punishments, they are hardly engaged in acts of violence. But violent behavior, including terrorism, often seems to be a product of social interactions. We might even hypothesize that it is different in degree, but not really in kind, from the crimes involved in the punitive damage experiments. Is the hypothesis correct? When people shift from feeling moral outrage to committing outrageous acts, are group dynamics the key factor? A resoundingly affirmative answer is the central theme of Marc Sageman's fascinating effort to understand the emerging nature of Islamic terrorism.
Sageman is a forensic and clinical psychiatrist who has been studying terrorism for a number of years. Once employed by the Central Intelligence Agency and now a counterterrorism consultant with the United States government, he is especially interested in Islamic terrorism. Calling for "evidence-based terrorism research," he rejects the approach of those who invoke case histories to examine particular terrorists. He believes that such an approach is insufficiently scientific. Focusing on individuals "can never tell us how they are different from those who might have become terrorists but did not." Nor is Sageman enthusiastic about efforts to explore the causes of terrorism by investigating economic and social conditions. In his view, such purportedly causal explanations fail to come to terms with a crucial fact: the rate of terrorism varies dramatically even when social conditions remain constant. In certain nations, campaigns of violence begin slowly, spike rapidly, and then fade abruptly. Sageman believes that this observed pattern "cannot be explained through slow-moving societal forces and cultural templates."
Sageman prefers what he calls a middlerange analysis -- one that is genuinely scientific, in the sense that it is based on evidence rather than anecdotes or speculation, and that uses statistical techniques to learn how people "evolve into terrorists; how they interact with others (terrorists and non-terrorists); how they become motivated to commit their atrocities; how they are influenced by ideas; and how they follow orders from far-away leaders." To answer these questions, he used a number of public and private sources to compile a large database, now including more than five hundred Islamic terrorists and their various characteristics. Sageman typically proceeds by collecting trial transcripts, legal documents, notes on testimony, and published interviews with Islamic terrorists. (He acknowledges that Islamic terrorism has distinctive features and he does not purport to have studied terrorism in general.) He favors the "application of the scientific method to terrorism research."
Drawing on the data, Sageman offers an arresting conclusion, which is that a major explanation of Islamic terrorism lies in patterns of social interaction that transform moral outrage into extremism. In his account, terrorists are not mentally ill, poor, uneducated, sociopathic, or victims of trauma. In the main, they are ordinary individuals who move to radical positions as a result of discussions with like-minded others. Sageman focuses in particular on the rise of what he calls "global Islamist terrorism" -- a large and loosely organized social movement that is subject to no command-and-control structure and has prospered in the wake of the invasion of Iraq. What makes Sageman's account distinctive is his emphasis on the crucial role of social networks -- in the real world and on the Internet -- and his effort to show that an understanding of those networks has significant and sometimes counterintuitive implications for how to safeguard national security. At the same time, Sageman offers general lessons about how enclosed enclaves of like-minded types help produce political beliefs and action of many kinds, including violence.
Like many other observers, Sageman begins by noting that a sharp distinction must be made between Al Qaeda as a formal organization and Al Qaeda as a social movement. Before September 11, Al Qaeda was synonymous with what Sageman calls Al Qaeda Central, the organization founded two decades ago by Osama bin Laden, whose followers formally swore loyalty to him or to his designated subordinates. In the 1980s and 1990s, many terrorist acts were coordinated by bin Laden and his deputies. But after the aggressive American response to the attacks of 2001, Al Qaeda Central has receded in power and importance. What has arisen is a social movement in the form of "informal networks that mobilize people to resort to terrorism," amounting to the "leaderless jihad." Sageman argues that as an organization, Al Qaeda, though still a threat, is largely contained -- but the social movement has grown significantly. That movement is greatly focused on the redress of what it perceives as past and present grievances, on justice, and on the restoration of past virtue and glory.
The heart of Sageman's analysis draws lessons from his database. Many observers believe that poverty is responsible for terrorism, but Sageman finds no evidence to support this belief. On the contrary, most of the terrorists in his sample are middle class. (Here and elsewhere, Sageman's findings overlap with those in two indispensable books, Robert Pape's Dying to Win: The Strategic Logic of Suicide Terrorism and Alan Krueger's What Makes a Terrorist: Economics and the Roots of Terrorism.) If we wish to make finer distinctions, we will find that those in the first wave of Islamic terrorists, in the 1980s, were equally divided between upper class and middle class, while the second wave, in the 1990s, involved mainly the middle class. The third wave, consisting of the generation that followed the invasion of Iraq, has middle-class to lower-class roots. But even in the third wave, there is no correlation between being poor and being a terrorist.
Sageman finds that terrorists are neither sociopaths nor psychopaths. Very few people in his data set display any kind of mental disorder. Indeed, Sageman notes that the low incidence of disorders suggests that "as a group, global Islamist terrorists may be in better mental health than the rest of the population." Some people have speculated that terrorists suffer from sexual frustration, but the vast majority of global Islamic terrorists are married, and two-thirds of them have children. Most Islamic terrorists lack any kind of criminal history. Many people think that terrorists are poorly educated, but here, too, the evidence goes the other way. Sageman finds that nearly two- thirds attended university, a rate that sharply contrasts with that in their home countries, in which fewer than 10 percent go beyond high school. Indeed, the percentage of terrorists who attended university actually exceeds the average rate in the United States (slightly over 50 percent)! And while trauma is often thought to help explain terrorism, Sageman finds little in the background of Islamic terrorists.
It is tempting to believe that terrorists have been indoctrinated or, in a sense, "brainwashed," religiously or otherwise. Indeed, the Saudi government has explicitly attributed terrorist attacks to religious brainwashing. But Sageman's data does not support this belief. He finds that jihadis are typically not well educated in religion. In Sageman's data set, only about one-fourth were deeply religious when young, and about two-thirds were secular. Nor do Islamic terrorists show much immersion in religion. He finds that the religious understanding of the majority of terrorists "is limited; they know about as much as any secular person, which is to say, very little. Often, they have not started reading the Quran seriously until they are in prison, because then it is provided to them and they have lots of time to read it." (Sageman is not focusing here on the leadership or contending that Al Qaeda leaders are not intensely theological.)
What, then, does lead people to become terrorists? Social networks and group dynamics turn out to be crucial. Here is an initial clue: a strong majority of people in Sageman's sample were expatriates, and joined a terrorist organization while residing in a nation in which they did not grow up. Indeed, four-fifths of his sample included people who were either expatriates or sons or grandsons of Muslim immigrants to the West. They joined the global Islamist terrorist social movement as a result of either friendship or kinship. Coming to the West as young adults, many of them searched for and found childhood friends in their new country, and eventually became involved in terrorism as a result. Many others joined relatives who were already part of the terrorist movement. Still others consisted of "a 'bunch of guys' who collectively decided to join a terrorist organization."
Sageman is especially intrigued by that "bunch of guys." He emphasizes that Islamic terrorism has been spurred to a considerable extent by the spontaneous self-organization of informal collections of people. In most cases, the ideological commitment was pre-dated by social bonds. Friendship and family first; political extremism second. Sageman registers his own surprise at "the lack of a top-down recruitment program into al Qaeda. There was no campaign drive for new members, no dedicated committee for recruitment, and no budget allocated specifically for this task."
Sageman describes a process of radicalization that starts with a degree of moral outrage. Many eventual terrorists see, on the Internet or on television, greatly disturbing events and images, such as the killing of Muslim children. For some people, the resulting sense of outrage becomes part of a broader narrative in which "wannabe terrorists" come to understand themselves as engaged in a kind of war against those who are plotting against people such as themselves. That narrative will not make sense to everyone, of course. It must also resonate with personal experiences. For many Muslim expatriates in Europe, those experiences include exclusion and discrimination. Expatriates and second eneration Muslims "compare themselves to their host peers. In Europe, they do not fare as well as the host young men. . . . They interpret their perceived discrimination in the context of moral violations against Muslims elsewhere, and the notion that their local grievances are part of a more general hostility against Islam appears more compelling to them."
Sageman is well aware that belief in the resulting narratives need not produce terrorism. In his account, a willingness to commit violent acts emerges as a result of social networks of two different kinds: face-to-face groups and virtual, online groups. The former include radical mosques and radical Muslim student organizations, consisting largely of people who came to the West to study. These relatively informal groups exercise a major influence. In the extreme cases, they create a sense of a collective identity, whose members "started living in their own world." Some of Sageman's most illuminating passages trace the distinctive nature of the underlying interactions, in which intense group dynamics, spawning a process of "in-group love," ensure that "the group acts as an interactive 'echo chamber,' encouraging escalation of grievances and beliefs in conspiracy to the point of hatred."
Group members are not exactly irrational; instead, their sources of knowledge become sharply limited. They come to rely exclusively on one another to validate new information, and everything that they believe is a product of interactions within their small enclaves. Thus "they discard information refuting their beliefs as propaganda from the West." It is important to note also that some group members reject the process of radicalization. They leave, ensuring a situation of voluntary sorting and self-selection in which "only the true believers remain." Those believers regard themselves as "best friends and a substitute for family." For them, it is especially difficult to depart.
Until 2004, face-to-face interactions played the key role in producing terrorist networks. More recently, the Internet has assumed great importance. Sageman emphasizes that the traditional hierarchy of terrorist groups is undermined by the Internet, which leads to a form of spontaneous self-organization. Chat rooms and dedicated forums inspire many young Muslims to join the Islamic terrorist movement. As Sageman describes it, the role of the Internet has been unplanned, unruly, and dangerous. As police began to monitor physical meeting spaces, especially in Europe, like-minded young people started to exchange views and information online. Sageman's particular emphasis here is on active rather than passive users. Websites are far less relevant than blogs and chat rooms, in which echo chambers help to radicalize people online. "The new forums have the same influence that these radical mosques played in the previous generation of terrorists," he writes. Conspiracy theories, fueling outrage, are spread in rapid fashion, as "individuals seek and select the rooms most compatible with their views and abandon the ones they disagree with. In a sense, the followers vote with their mice and select the views they like."
Offline, those who might believe that they are isolated in their anger and fear, or in their incipient convictions, might be tempted to start thinking about something else. But they can easily find one another through the Internet, and they become more extreme in their beliefs, simply because so many people seem to agree with them. Sageman thinks that Islamic terrorism reflects an even broader problem here. "Let's assume that a very few people in the world share the same strange belief, say that the moon is made of green cheese. Through a process of self-selection, they find each other on the same forum. . . . Soon, they will assume that everyone shares this conviction because only the true believers air their views and the rest stay silent." Through this process, strange and potentially dangerous beliefs, which would otherwise fall of their own weight, can be hardened as a result of the Internet.
With this understanding of the role of social dynamics, Sageman wants to explain why it is that Europe has experienced a set of attacks after the events of September 11, while the United States has not. He believes that the United States is far safer than Europe, and that the United States does not face a significant threat of homegrown extremism; but he does not think that the explanation lies mostly in the efforts of the United States to combat terror. Instead, he urges that, for cultural reasons, radicalization of Muslims is much more likely in Europe than in the United States. In the United States, the idea of the melting pot promotes the acceptance of people of diverse backgrounds, including foreigners, and increases the likelihood of eventual integration into society. In Europe, Muslims are far more likely to complain about exclusion and discrimination than in the United States. It follows that the United States is simply not breeding Al Qaeda networks: "For most American Muslims, the terrorist message does not become a catalyst to action as it does for their European counterparts, since it does not resonate with their beliefs or personal experience." Here Sageman points to the valuable unintended consequence of an open culture: "For the United States, pursuing policies of political, social, and economic inclusion rather than exclusion has paid an enormous dividend."
Sageman acknowledges that in some areas Al Qaeda is regrouping and consolidating, but he argues that there is no general resurgence. "Al Qaeda Central is of course not dead, but it is still contained operationally. It puts out inspirational guidance on the Internet, but does not have the means to exert command and control over the al Qaeda social movement." Sageman wants the United States to come to terms with the proliferation of homegrown terrorist networks lacking any physical link to Al Qaeda Central. The good news is that its members are "far less skilled and adept than the terrorists prior to 9/11." The bad news is that because "the number of wannabe terrorists remains large . . . by the law of averages some are bound to be smart and bold enough to pull off an attack."
What follows from this diagnosis? A new focus should be placed on the social movement and its appeal to young people, which is "sustained by U.S. actions that are perceived to be a war on Islam." To weaken that appeal, Sageman suggests that the United States should attempt to remove the thrill and glory associated with terrorism, among other things by reducing terrorists to common criminals. To accomplish that task, the United States should cease to hold press conferences in which captured terrorists are prominently mentioned. He believes that, as a tactical matter, security is best promoted through quiet arrests and prosecutions, which deprive terrorists of a public stage. Sageman also wants both the United States and Europe to eliminate discrimination against Muslims -- a step that, in his view, will reduce the appeal of terrorist networks.
Sageman is especially concerned about "jihadi cool," a focus of Muslim enclaves in Europe and in Internet chat rooms. Thus far, he says, the "battle is completely one-sided. The true believers, who populate the radical forums, are working themselves into a frenzy with no moderate voice present to calm them down. This leads to ever greater radicalization on the part of the participants, who slowly take on the views of their friends." Sageman wants "to take the glory out of terrorism by demilitarizing the conflict except for sanctuary denial and by reducing terrorists to the status of common criminals." Hence Sageman seeks to diminish Muslim moral outrage "globally by withdrawing from Iraq and locally through restraint in the aftermath of terrorism operations."
Sageman has produced two overlapping books -- the first relatively narrow, the second far more ambitious. The narrow one features his analysis of his database. The more ambitious book offers general lessons about the arc of terrorism over time and the appropriate response to it. Let us explore these in sequence.
Many of Sageman's conclusions -- for example, that terrorists are not poor, badly educated, or insane -- are not new, and they are supported by other studies. Sageman's distinctive contribution lies in his emphasis on social interactions among like-minded people, and in particular the effects of enclaves of young Muslims. While Sageman does not refer to it, some of the most intriguing work in modern social science explores the very phenomenon that he is describing, where it is known as "group polarization." Sageman is actually offering a particular illustration of that much broader phenomenon. The central empirical finding is that after deliberating with one another, group members typically end up in a more extreme position in line with their pre-deliberation tendencies. Group polarization is essentially a form of radicalization. It is found in jury deliberations about appropriate punishment; it can be found in many other areas as well.
Suppose, for example, that a deliberating group consists mostly of people who are concerned about climate change, or who believe that affirmative action is a bad idea. After group members speak with one another, they will be all the more concerned about climate change, and all the more unhappy with affirmative action. Group polarization has been demonstrated in over a dozen nations and in numerous areas; those found to be subject to polarization include jurors, ordinary citizens, burglars, and entrepreneurs. Enclaves of like-minded people move to the extremes. The liberal citizens of Boulder have been found to move sharply to the left after speaking with one another; the conservative citizens of Colorado Springs have been found to move sharply to the right. Even federal judges tend to polarize. On three-judge panels consisting solely of Republican appointees, Republican appointees show far more conservative voting patterns than on three-judge panels having at least one Democratic appointee. Democratic appointees show exactly the same tendency; they are far more liberal on panels consisting of all Democratic appointees than on panels with one or more Republican.
Why does group polarization occur? There are two major answers, and they bear directly on Sageman's findings. First, the exchange of views and information can intensify pre-existing beliefs. In any group with some predisposition (in favor of the Republican candidate, concern about climate change, opposition to same-sex marriage), the pool of arguments will inevitably be skewed toward the original predisposition. If group members are listening to one another's arguments, they are likely to become more extreme in their views. The second reason involves people's concern for their reputations. Most people want to be perceived favorably by other group members, and also to perceive themselves favorably. Once they hear what others believe, many of them will adjust their positions at least slightly in the direction of the dominant position. Without invoking the concept, Sageman has provided a dramatic case study in group polarization, and both of the standard explanations play key roles in the process of radicalization that he describes.
Sageman also offers a nice twist, involving the crucial importance of voluntary sorting. Radicalization, in his account, is promoted by people's decision to sort themselves into offline and online groups, and it is significantly increased by the fact that at a certain stage, those "who believe that the group has gone too far in their growing radicalization peel off. . . . At the end, only the true believers remain." On the Internet, such sorting is especially important, as online jihadis can choose their preferred forums, and often end up in discussion groups that promote violence. Long before the attacks of 2001, the experimental evidence found that extremists are especially prone to polarization; and if true believers are the only ones left, radical movements should not be surprising.
Sageman's claims here are plausible, informative, and even important, but they raise an obvious question. Has he really succeeded in applying "the scientific method to terrorism research"? To his credit, he has compiled a large database, and he offers some noteworthy findings, such as the low percentage of terrorists who are poor, badly educated, or mentally ill -- and the high percentage who are married, expatriates, and close relatives of those already involved with Islamist terrorism. But he does not bring his database into close contact with his central narrative about the role of social networks. Sageman emphasizes various "prongs of radicalization," including moral outrage, perception of a war against Islam, resonance with personal experience, and mobilization by networks. But he does not specify, as a social scientist would, how these "prongs" help to account for the ultimate paths of fifty, one hundred, two hundred, or four hundred people in his five-hundred-person database. There is no statistical analysis here.
There is a deeper problem, which is known in the trade as "sampling on the dependent variable." Suppose that I have a database of two hundred major-league baseball players; and suppose I find that while very few of them are poor, poorly educated, or mentally ill, a strong majority had at least one parent who was a registered Republican, or who attended church regularly. Would I be entitled to conclude that having such parents "made" people major-league baseball players, or was a necessary or sufficient condition for becoming one? Of course not. Many people with such parents do not become major-league baseball players -- and even if all players had such parents, we could not infer causation.
Here is another way to put the point. Suppose that Sageman finds one hundred groups of people who ultimately become terrorists, and could show that those people are more extreme than before they joined the group. But suppose that there are nine hundred other groups, of which eight hundred consist of people who became no more radical when they joined together, and of which one hundred consist of people who became less radical when they joined together and exchanged views. Then we would see randomness, not any causal phenomenon. In short, Sageman's database and method do not permit him to compare those who became terrorists with those who did not; and for all their plausibility, his claims about the causal force of networks have not been shown to count as social science. So the core narrative of his book has not been adequately corroborated by his own evidence.
Sageman's more ambitious goal is to offer suggestions about how to respond to a significant shift in the nature of the terrorist threat. Emphasizing that Al Qaeda Central has been effectively contained, he urges a new focus on the danger posed by the leaderless jihad. For the United States, he believes that homegrown terrorism is not the most serious problem, because of the absence of (much) discrimination, and because the widely held belief in the melting pot and the commitment to freedom of opportunity help to ensure that Muslims do not become a self-enclosed enclave of people prone to radicalization. For Europe, the prescription is essentially to become more like the United States. And for the United States, the prescription is essentially to maintain our best traditions, because they are a recipe for increased security.
Sageman is right to emphasize the importance of introducing moderation into both offline and online discussions among Muslims. He objects to a process that leads to "ever greater radicalization on the part of the participants, who slowly take on the views of their friends." Sageman wants the online battle to include influential opinion leaders who "embrace nonviolence as a way to fulfill Muslim aspirations. This campaign should be conducted by other Muslims. " On this count, Sageman is on strong empirical ground. It is well established that the process of radicalization, or group polarization, can be reduced if groups begin with a degree of diversity. Such diversity also alters the information that is presented within group deliberations; it alters the social pressures that lead people toward extreme positions.
Unfortunately, many of Sageman's broader claims are more impressionistic and intuitive, and they are not exactly supported by the "scientific method." Sageman thinks that the United States and its allies can reduce the appeal of Islamic terrorism essentially by reducing the volume of the war on terror -- by eliminating press conferences that celebrate captures and victories, and by ensuring that arrests and prosecutions are kept quiet. He adds that the United States faces a limited risk of homegrown terrorism simply because of its openness. More ambitiously, he goes so far as to argue that a military strategy is counterproductive and self-defeating because it helps to radicalize "a new wave of young jihadists who have been led to believe that America is establishing a base in the Middle East in order to exploit oil resources and dominate the region." Strikingly, Sageman concludes that "the only military role in the overall strategy to fight global Islamist terrorism should be to deny terrorists sanctuary."
Maybe so. But Sageman does not show that his own evidence adequately justifies these conclusions. England is hardly the land of the melting pot, but in recent decades its leaders have worked hard to be nondiscriminatory and openly multicultural -- and yet it has spawned a great deal of terrorism. And with all its openness, the United States has not been immune. Jose Padilla and the Lackawanna Six were homegrown terrorists, and the New Jersey Islamic terrorists who first bombed the World Trade Center operated within American borders. Sageman may be too optimistic about the effects of openness and non-discrimination, at least for people who, for one reason or another, are intensely committed to their ethnic or religious heritage, and after banding together, seek to use violence to redress historical grievances.
True, we know that many people become terrorists as a result of a process of radicalization. But we do not know exactly how dangerous members of the "third wave" actually are, nor do we know how much the United States has contributed to their numbers by publicizing the capture of our enemies and through military action. Sageman himself emphasizes that the weakening of Al Qaeda Central was achieved in large part by the detention and death of key personnel and by the closing of training camps in Afghanistan; but those achievements were emphatically a result of military action. Sageman is probably right to say that the war in Iraq has produced a large number of new terrorists, but on this count he has no science to offer. Indeed, nothing in Sageman's book shows that military action is unjustified, on balance, unless it denies terrorists a sanctuary. Where is the empirical support for this sweeping national-security claim?
To this extent, Sageman has not achieved his goal of providing "evidence-based terrorism research." Still, his emphasis on the role of social networks, and on the process of radicalization, greatly illuminates the emerging situation -- and shows why aggressive acts, by the United States and others, may actually decrease security. Equally important, his elaboration of what is in effect a case study in group polarization offers a number of provocative lessons for understanding not only Islamic terrorism, but also other domains in which extremism is the eventual result of voluntary self-sorting -- with destructive and even deadly consequences.
Cass R. Sunstein is currently the Felix Frankfurter Professor of Law at Harvard Law School.
Books mentioned in this post