What is a Riot?

We think we know what causes mob violence. We're probably wrong.

As I read your comments on the first part of this essay, I realized it might be interesting to you if I explored the literature on rioting in a bit more depth. It seems many of you have a keen interest in the subject of riots, and as some of you pointed out, they’re not a phenomenon unique to France, even if they’re more common here. They seem to be rising in frequency around the world. It might be useful to ask why.

So I thought I’d take a few more days with this topic so best to think the problem through carefully. In that spirit, today’s newsletter asks: What exactly is a riot? How does it differ from other crimes or acts of violence? What, if anything, causes a riot?

Tomorrow, we’ll look at the literature on the prevention of riots and ask how to stop them when they happen despite your best efforts to prevent them. Is there a way for the police and government to respond that minimizes damage and loss of life? Does responding in an effective way reduce the odds that your non-rioting citizens will usher fascists to power at their earliest opportunity because they’re sick of the rioters and want them all shot? (Turns out the answers are “yes” and “yes.”)

Finally, I’ll ask how all of this applies to France. Although France didn’t ask me, I’ll conclude with a ten-point strategy for the French should they ever wish to have fewer and less violent riots. I doubt they will take this advice, but I’m pretty sure it’s correct nonetheless.¹

---

Do riots have a cause?

When riots break out, we generally say they were caused—usually by a proximate event. So it was widely reported that riots erupted in France in 2018 because the government planned to raise fuel taxes. When last March they broke out again, it was because the government proposed to raise the retirement age. Riots occurred last week because the police fatally shot a teenager of north African descent.

These descriptions contain suppressed premises. The first is that riots do not just break out for no reason: They require a cause. The second is that the cause is what the rioters, or the majority of them anyway, say it is. The third is that this notion of causality makes sense.

If any of this is true, it is only partly true. Locutions of this kind are so common that we rarely stop to examine them, but if we do, we quickly see that there must be many steps between the ostensible cause and the effect. If a proposal to raise the retirement age caused a riot in France, why didn’t such a proposal have the same effect in Germany? If the recent riots in France were caused by the shooting of a 17-year-old, why didn’t riots erupt when the police shot a 21-year-old? In 2022, the French police fatally shot 39 people. Why did this shooting cause riots, but not those?

Perhaps, one might reply, the body politic is like the carbon fiber hull of the Titan submersible: It can be weakened by repeated stress. It then becomes prone—under pressure—to catastrophic implosion. The metaphor is attractive, but only as far as it goes. A body politic isn’t really made of carbon fiber. It’s made of millions of individual people, each infinitely complex and distinct. Yet when a riot breaks out, thousands or tens of thousands of people who may have little in common suddenly act in concert, behaving in ways they would never ordinarily behave. Have they all been subject to exactly the same amount of stress? Have they all grown weaker at exactly the same pace?

When we say that a submarine has crush depth of 2400 feet, we don’t mean that sometimes it’s fine at the depths of the Marianna Trench but every so often it implodes at sea level. We mean it implodes at 2400 feet, period. We can predict that reliably. Riots are not like that. If they were, they’d be as rare as imploding submarines, because most of the time, we’d have the good sense to prevent them.

In the past week, many variants on this column have been published:

Juliette Fevre, who grew up close to Nanterre, said racial tension has been brewing in neighborhoods following the death of Adama Traoré in police custody in 2016, as well as the 2005 deaths of Zyed Benna and Bouna Traoré, which also led to riots. She said there is a feeling that Black French residents and those of North African heritage can feel police take pleasure in chasing and arresting them “for almost nothing.”

“Nahël had troubles with the police for refusing to co-operate but people are angry because the policeman wasn’t threatened [by him],” she said. “By all means, Nahël didn’t deserve to die.” Ms. Fevre said, “Protesters wanted justice for Nahël at first. Now, it’s turning into something bigger. Nahël was a first spark, a trigger. People are lacking good housing, good public service, good education and opportunities.”

No one disputes the facts of this account.² But did they cause a riot? Or is it that the meaning we assign to these event retrospectively?

Some things about rioting are so obvious that we tend not to notice them, and when we do, we fail to appreciate that they’re significant. Foremost among them is that rioting is almost always counterproductive. If you’re unhappy that the police view you with hostility, for example, what could be less likely to improve this situation than filling water balloons with your own urine and hurling them at the cops? Only an idiot thinks that’s going to cause a cop to like you better. Yet when rioting begins, this kind of logic takes over: He was angry that he didn’t have good housing, so he burned down his own neighborhood. He wanted better public services, so he burned down the bus station. They were mad about the high cost of living, so they crippled the economy for months.

If you point out this contradiction in the rioters’ behavior, someone will usually offer an explanation that goes like this: Rioters just want to send a message to the government. They don’t understand or care about how the economy really works or how political and cultural change really happen. They’re angry, and they want someone to do something. But would you accept an explanation of this kind if you were trying to understand the behavior of a single individual, as opposed to a large group’s?

“So, this morning Pedro stormed into the C-suite wearing nothing but a gimp mask and threatening to slice off the boss’s cojones and airmail them to Cuba.”

“Why did he do that?”

“He wanted a raise.”

“I can’t think of anything less likely to result in getting a raise.”

“He doesn’t really understand office politics. He’s just angry and he wants the boss to do something.”

“Oh, that makes perfect sense.”

That these explanations don’t make sense is also, perhaps, why they’re often inconsistent. We heard hundreds of mutually incompatible explanations for the riots that broke out here in 2018. Even several years later, no one is sure which one was correct. By contrast, there was exactly one explanation for the Titan’s disappearance, and we knew what it was, for sure, within five days.

---

In the aftermath of the First World War, the British mathematician, physicist, and meteorologist Louis Fry Richardson sought to understand the underlying patterns that led to conflict with the aim of preventing future wars. The result, published in 1960, was Statistics of Deadly Quarrels, a mathematical analysis of conflicts ranging from homicides to full-scale wars between 1820 to 1950. The book laid the foundations of modern quantitative political science.

Richardson endeavored to find variables that reliably predicted the outbreak of conflict. Did the belligerents share a common language? A religion? Borders? Does a strong military force on each side reduce the odds of war or raise it?

His conclusions were highly counterintuitive. Outbreaks of violence, he found, could be closely modeled by a Poisson distribution. They occurred in much the same pattern as tsunamis and meteor strikes—more or less at random.³

Richardson’s thesis remains controversial and contested. To my knowledge, no one has applied these methods to the study of riots, specifically. Nonetheless, the idea is suggestive. Is it possible that what we think we know about why riots break out is wrong?

Whenever riots occur, we tend to accept the explanation for them given by the rioters, whether it makes sense or not, and if the rioters offer a range of explanations, we favor the ones that sound serious. We assume that at some level—even if the rioters themselves can’t articulate it—there is a grievance that warrants this amount of uproar.

When a rioter is asked why, for example he decided to throw a firecracker at a cop, he will quite often say, “I don’t know.” But you never see the headline, “Riots broke out for no reason.” So we ask again until we find a rioter who explains what he did in terms of the anger he feels about a recent grievance, particularly one we think worth that much outrage. We wait for an answer like, “I torched the car to express my anger about lacking good housing, good public service, good education and opportunities.” Usually, as soon as the first journalist gets that quote, the narrative is set: That’s why it happened.

Still, it’s rare for a proximate cause to be so obviously outrageous that an explosion of violence makes perfect sense. So usually we imagine that the proximate cause was merely the straw that broke the camel’s back. We quickly converge on the idea that these riots aren’t just a response to a police homicide, or even a pattern of them. The story tends to expand in assigned significance: The riots are, say, a response to years of exclusion and marginalization, to a France that has never truly accepted immigrants as fellow citizens. Because riots are so destructive, we reason, the larger grievance must be that profound. It is hard to accept that an outbreak of violence so destructive and costly might have been caused by nothing in particular, or by something trivial. If rioters have are burning down whole neighborhoods, they have to be suffering deeply.

We tend, too, to accept an idea popularized by Martin Luther King: that a riot is an act of communication. “I think that we’ve got to see,” he said, “that a riot is the language of the unheard.” In that instance, he explained, the rioters in question were trying to say that “the economic plight of the Negro poor has worsened over the last few years.”

His idea has been widely accepted. Riots are widely believed to be desperate actions taken by people who seek to communicate an urgent message. To accept this, we must believe that normal language has failed. By normal language, I mean English or French, as the case may be. If the problem is subject to political redress, we normally use these languages to speak through political parties, representatives, or interest groups. We give speeches, write letters, protest, march, go on strike, boycott, sue, sign petitions, hold sit-ins, post on Facebook, take out ads in the newspaper, or even hire a skywriter. If we accept that a riot is the language of the unheard, we necessarily assume that these avenues are unavailable to the rioters, or that they’ve been tried and failed, or that the rioters don’t know that about them—perhaps because they are, as the Marxist historian E.J. Hobsbawm described insurrectionist 19th-century Andalusian peasants, “pre-political people who have not yet found, or only begun to find, a specific language in which to express their aspirations about the world.”

But how often is this really true—especially in a country like France, a representative democracy with a free press? Do normal avenues of political communication fail so frequently here as to explain the amount of rioting we see? It seems a bit patronizing to describe French citizens, in the year 2023, as “pre-political.” Perhaps a riot isn’t always an act of communication, then? Perhaps it’s something that every so often, people just like doing?

There’s some empirical research to support the idea that riots are more common under conditions of relative deprivation; that is, it’s not poverty or injustice, per se, that causes riots, but the gap between what people expect and what they have. Criminologists have also found some support for the idea that the most likely spark for a riot is a single, highly visible incident of perceived injustice, especially in a context of underlying social tension.

Beyond that, support for the idea that a riot is a language—of the heard or the unheard—is far from overwhelming. In fact, it does seem that often, riots break out for no good reason at all.

---

We must strictly distinguish between protests and riots. Protests are a normal aspect of politics in a liberal democracy. In any nation where the rights of speech and assembly are secure, protesting is not illegal. Riots are large-scale, collective outbreaks of violence against people and property. The things people do when they riot—set fires, assault the police, vandalize property, loot stores—are crimes.

What distinguishes a riot from a set of distinct crimes is collective action. It is a crowd that riots, not an individual. A riot breaks out when a large number of people, spontaneously, decide collectively to act in concert and break the law in specific ways. The central mystery is why, under certain conditions, people collaborate to behave in ways they would not usually behave, and why being part of a group is, under certain conditions, so disinhibiting.

In 1895, attempting to answer this question, Gustave LeBon published The Crowd: A Study of the Popular Mind. He theorized that people in a crowd lose their sense of self and responsibility, which leads them to behave emotionally and irrationally. Contagion Theory, developed in the wake of LeBon, posited that emotions, behaviors, and ideas can spread through a crowd like a virus. Riots, in this view, occur when this contagion spreads collective excitement

Le Bon, The Crowd.

Elaborating upon these ideas, Ralph Turner and Lewis Killian developed Emergent Norm Theory in the 1950s. Mob behavior, they argued, should not be viewed as completely random and chaotic. They hypothesized that rioters develop new social and moral norms. These may be completely contrary to ordinary norms—“Always disrespect the police,” for example, as opposed to “Always respect the police”—but they nonetheless function as norms usually do, in the sense that rioters are guided by them, and those who transgress them are subject to such punishments as ostracism.

In recent years, these ideas have been supplemented with Network Theory, which investigates, among other things, the effect of the internet and social media on the propagation of collective violence. What’s interesting here is that social media seems to spread emergent norms from one riot to the next, even across cultures and borders. So if France’s protests in recent years have been uncommonly violent and destructive, Network Theorists might say this is not necessarily because France is under unusual social pressure, while Emergent Norms Theorists might add that this is not necessarily a sign of general moral breakdown, either. It may instead be a sign that rioters are now learning how to riot from TikTok videos—which are exposing them to newer, more violent norms. This certainly seems to be the case with the very young rioters who recently set fire to France’s suburbs: The police are reporting, repeatedly, that the adolescents they’ve apprehended are learning from, and trying to create, spectacular TikTok videos.

Some criminologists have challenged Le Bon’s perspective on the irrationality of crowds, arguing that individuals actually make perfectly rational calculations about the costs and benefits of participating in a riot. If the perceived benefits (social approval, drawing attention to a cause, changing government policy, acquiring goods by looting, the fun of participating) outweigh the potential costs (arrest, injury, imprisonment), they argue, people are more likely to join a riot. While these ideas are debated—and I have no firm opinion about the debate—what’s not in contest is that a riot is a distinct and unusual human subroutine that falls under the category of “collective behavior.” The dynamics of collective behaviors—mobs, panics, rumors, mass hysteria, fads, crazes—differ from those of individual behaviors. So if we want to understand riots, we must understand them on their own terms.

Sociologists have identified distinct types of riots. One prominent typology, for example, distinguishes between protest and celebration riots. In the former, rioters express unhappiness about a political, social, cultural, or economic issue; in the latter, they express joy, as for example when riots break out in the wake of a football victory. People tend to believe these are very separate phenomena. But it isn’t clear that they are, and it’s important grasp the significance of this: Riots don’t necessarily emerge because people are unhappy. Sometimes, people riot because the social dynamic lends itself to rioting.

Sociologists have likewise drawn distinctions among purposive, symbolic, revelrous, and issueless riots. Purposive riots are intended to achieve a clear goal. Think of prison riots, for example, in which the goal is to change a specific prison rule. Symbolic riots express general disgruntlement but have no clear objective. Revelrous riots, again, are like those that break out after football games. Then there are issueless riots. They just happen. They have no purpose. Power blackouts are known to give rise to them.

The key is that no, rioting isn’t always “the language of the unheard.” It’s part of the normal human behavioral repertoire. Sometimes people riot because it’s fun and sometimes they riot because they think they can get away with it. When asked, rioters have an obvious motivation to describe revelrous or issueless riots as purposive: It sounds a lot better. But that doesn’t mean it’s true.

When a riot breaks out, we often focus on the question “Why are these people rioting?” But if we start with the assumption that quite often, riots happen for no reason or because people like rioting, the question changes. It becomes: “Why did the mechanisms we’ve collectively put in place to prevent rioting fail?” This may be a more useful question.

And if we return to the original question—why are there so many riots in France?—we may find that it’s more useful to ask, “Why hasn’t France collectively put better mechanisms in place to prevent rioting?”

Next: Preventing Riots.

1. Note: This is a revised version of the original newsletter sent to subscribers. When I re-read it, I found many typos and realized a number of sentences weren’t clear. I’ve changed nothing of significance.

2. More or less. It depends what you mean by “good housing, good public service, good education and opportunities.” No one, after all, is drowning in the Mediterranean in the effort to flee from France. What’s not in dispute is that a cop shot a 17-year-old. No one thinks the shooting was justified, including the cop in question, who faces homicide charges and is said to be overcome with remorse.

3. For a much more complete discussion of Lewis Fry Richardson’s ideas, see my father’s essay, The Best of Times.