All Commentary
Tuesday, March 15, 2016

Why Riots Happen

Spontaneous (Dis)order: The Incentives and Entrepreneurs of Mass Crime

After the Los Angeles riot in spring of 1992, almost every pundit in the country took a turn at explaining why riots occur. The conventional wisdom on the subject went something like this: certain dramatic events such as political assassinations or unpopular jury verdicts crystalize riots from social rage.

So to understand riots, one must understand the causes of social rage, usually said to be racism, poverty, lack of economic opportunity, and why people who experience this rage manage it in such a destructive manner. The usual suspects include breakdown of the family, television, and a generalized cultural disorientation.

All of these explanations have some truth in them, but are evidently incomplete. First, they explain too much. The predisposing social conditions are with us all the time, yet riots are episodic.

Second, they explain too little. Many mob actions, like European soccer riots or the increasingly predictable civil meltdowns in the home cities of National Basketball Association champions, are triggered by good news, and not obviously related to social injustice or existential anomie.

Indeed, during the Los Angeles riots, anyone with a TV set could see that jubilation rather than fury best characterized the mood of the people in the streets. It is hard to credit that these exhilarated looters with their new VCR’s and cameras were protesting the jury system, the state of race relations in Southern California, or anything else.

They were, in fact, having a party. Moreover, many of those who risked life and limb opposing the more outrageous excesses of the rioters were themselves poor, unemployed, and victims of racism.

All the conventional explanations for riots explain too much, and too little.Conversely, a crowd is not an incipient riot merely because it assembles a great many people with the predisposing demographic characteristics. For example, every Fourth of July in Chicago’s Grant Park there is a fireworks display that usually attracts about a million spectators. In certain parts of the grounds, people are packed together like sardines, so that individuals substantially lose their ability to decide where to go. One goes where the crowd goes. Going against it is impossible, and even leaving it (unless one is near the edge) may be difficult.

Some people dislike the experience, but whatever its discomforts, the Fourth of July crowd at Grant Park is not a riot in the making. The crowd is big, it is loud, it is unmanageable, it is filled with people who have suffered from racial discrimination and economic deprivation, it has, in aggregate, drunk a lot of beer (which is legally for sale at dozens of kiosks at the event); but it is only a crowd, not an incipient riot.

Day in and day out in any big city, police blotters will reflect the existence of a fairly steady background supply of theft, mugging, arson, and homicide. But this jumble of criminal mischief does not amount to a “riot”; riots are the coordinated acts of many people. If they are coordinated, who coordinates them? Authorities looking for ways to explain why trouble has broken out on their watch sometimes ascribe exaggerated organizational powers to “outside agitators.”

While, as we explain, there is definitely a leadership niche in the ecology of a mob, it seems to become important only after the crowd has assembled. Riots are not, as a rule, plotted and scripted affairs.

It would be very difficult indeed to “stage” a riot. A person who set out to do so would encounter a series of difficult challenges. When should the riot be held? Where? How should the participants be notified? Once marshaled, how should they be instigated to behave in a way that would expose them to arrest? Trying to organize a riot as though it were a company picnic would quickly attract the attention of the police. And with the police watching, who would be brave enough to cast the first stone?

How, then, do riots begin?

Assembling the Crowd

When something happens that causes a large number of riot-prone people to mass spontaneously in one place, while police cannot mass at a correspondingly rapid rate, the cost of starting a riot, as any one participant would figure it before the fact, would begin to decrease dramatically.

It would decrease still more if it seemed to a prospective troublemaker that his own hopes and expectations about the potential behavior of the crowd were also the hopes and expectations of many of its other members, because in that case it would seem a better bet that if he did cast the first stone, many others would follow.

The most obvious way to get a riotous crowd to assemble is the occurrence of what could be called a “Schelling incident,” after Thomas Schelling, the great master of strategic theory. In The Strategy of Conflict (1960: 90) Schelling wrote,

It is usually the essence of mob formation that the potential members have to know not only where and when to meet but just when to act so that they act in concert. Overt leadership solves the problem; but leadership can often be identified and eliminated by the authority trying to prevent mob action.

In this case the mob’s problem is to act in unison without overt leadership, to find some common signal that makes everyone confident that, if he acts on it, he will not be acting alone. The role of “incidents” can thus be seen as a coordinating role; it is a substitute for overt leadership and communication. Without something like an incident, it may be difficult to get action at all, since immunity requires that all know when to act together.

It is not crucial, in the generative stage of a riot, that the participants act literally simultaneously. What is crucial is that offenses occur rapidly enough to overwhelm the police. From the rioter’s viewpoint, there is safety in numbers. There comes a point at which the police pass from inadequacy to impotence.  In the Los Angeles riot, the police actually pulled back from the trouble when it became obvious to everyone, including themselves, that there was nothing constructive they could do.

Certain kinds of high-profile events have become traditional “starting signals” for riots.Certain kinds of high-profile events have become traditional “starting signals” for civil disorders. In fact, incidents can become signals simply because they have been signals before. What ignited the first English soccer riot has been lost in the mists of history; but they had become a troublesome problem sometime during the nineteenth century, as Bill Buford (1991) makes clear in quoting old newspaper accounts in his Among the Thugs.

Today, there is a century’s weight of tradition behind soccer violence. People near a football ground on game day know that a certain amount of mischief, possibly of a quite violent kind, is apt to occur. Those who dislike that sort of thing had best take themselves elsewhere. Certain people, though, thrive on the action — relish getting drunk, fighting, smoking dope; enjoy the whiff of anarchy, harassing and beating respectable people and vandalizing their property.

Such people — hooligans — make a point of being where the trouble is likely to start. The sort of “soccer fans” about whom Buford wrote were mostly interested in barbarian camaraderie, not soccer. Some of them do not even go inside the stadium, and some spectators do not watch the game but pass their time in petty thievery. Hooligans’ game is being a part of the crowd that congregates near a soccer stadium, belonging to and sharing its power, especially its power to flout the law.

A Schelling incident is not a signal that tells a person what to do. It is a signal that tells a person what other people will probably do. In the United Kingdom even an ordinary minor league soccer match might well be a Schelling incident. Buford gives several examples.

In the United States, that sort of game would not be — but having one’s team win a National Basketball Association championship increasingly seems to be. In Detroit in recent years, “Devils Night” (the night before Halloween) has become a springboard for multiple, independent, almost simultaneous acts of arson. These are examples, baleful ones, of how culture, habit, and tradition can overcome major organizational barriers to cooperative social endeavors and lower the cost of transacting business.

As word spreads of a conventional triggering event — whether it is shocking (like an assassination) or rhapsodic (a three-peat) — crowds form spontaneously in various places, without any one person having to recruit them. Each member of the crowd will know more about the intentions of fellow crowd members than people usually know about the intentions of strangers, because once a starting signal has been given, people know that a riot is impending. They gather into crowds because they want to participate and they know why the other people in the crowd, or at least a great many of them, have come.

Not every crowd threatens to evolve into a riot. In fact, the opposite is more often true: people bent on criminal mischief usually do not want lots of witnesses and possibly hostile bystanders around when they commit crimes. And so the psychology of the crowd’s members is crucial. A significant number of the crowd’s members must expect and desire that the crowd will become riotous. That is, there has to be a critical mass of people in the crowd who are making accurate judgments, not about their own desires and intentions, but about the riotous desires and intentions of other members of the crowd.

The Role of the Entrepreneur

For a riot to begin, it is necessary but not sufficient that there be many people who want to riot and who believe that others want to riot too. One more hurdle has to be overcome. Even in an unstable gathering, the first perpetrator of a misdemeanor is at risk if the police are willing and able to zero in on him. Thus, someone has to serve as a catalyst — a sort of entrepreneur to get things going — in Buford’s account, usually by breaking a window (a signal that can be heard by many who do not see it).

In civil rights, anti-war, or anti-abortion marches, it is probably pretty common to find participants eager to expose themselves to arrest in exchange for the chance to optimize the desired impact of their protest.

Riots await events that surmount the risk of sacrifice.This sort of self-sacrifice is certainly rare in ordinary riots, where potential rioters’ behavior is consistent, we suppose, with something like the following calculation: “If somebody else gets the riot started, I can participate without much risk. But if I stick my neck out and nobody follows, I’ll be the only one arrested. So I’ll wait for somebody else to go first.”

If every would-be rioter reasoned thus, nobody would cast the first stone, and the riot would not ignite. This is a typical free-rider problem, as economists have called it. It is usually sufficient to prevent riots from occurring, even where there is a plentiful supply of disposed participants. Riots await events that surmount the free rider problem.

The entrepreneur will throw the first stone when he calculates that the risk that he will be apprehended for doing so has diminished to an acceptable level. The risk of arrest declines as a function of two variables — the size of the crowd relative to the police force available to control it, and the probability that others will follow if somebody leads. This latter point could potentially be tricky, because as we have noted, crowds will generally be inhospitable to the commission of violent acts. But it is possible for a crowd to telegraph its willingness to riot.

Buford’s account (1991: 81—85) of a soccer hooligan rampage in Turin furnishes an example. Members of the crowd marched themselves around in a spontaneous formation with a stilted, unnatural gait, chanting the name of their team. This unmistakable token of cohesion stopped well short of anything that the Italian police could plausibly charge as solicitation or incitement, but served to assure the members of the crowd that a critical mass had formed.

Sometimes a crowd will not clearly commit itself to riot, and in such instances an entrepreneur will take more of a risk getting things started. But if he has done his implicit calculations properly, once the first plate-glass window is broken, the looting will begin and will spread and continue until the civil authorities muster enough force to make the rioters believe that they once again face a realistic prospect of arrest.

The Formation of Action Nodes

As we saw in the case of Los Angeles, riots do not occur everywhere at once. Most of the homes and businesses in south-central L.A. and Koreatown (which cover a number of square miles) were untouched by the riot. Damage was concentrated at certain intersections and along certain strips, what we call “action nodes.” How did the rioters know where these action nodes were?

Schelling (1960: 54—58) again offers a framework for analysis by offering powerful evidence for the existence of focal points in social life. People who may never have met are nonetheless capable of coordinating their behavior under some circumstances. In one experiment, two people were instructed to think of a number between one and ten and told that both would be paid a reward if each arrived at the same answer. Subjects’ ability to psyche one another out far exceeded chance.

Perhaps even more surprising, certain open-ended questions can elicit a high amount of agreement. For example, in one experiment Schelling asked his subjects what they would do if they were simply told to go and meet someone in New York City on a certain day. Out of all the possibilities for when and where to meet, a majority, trying to intuit where and when other people would expect them to be, would have converged at the information booth in Grand Central Station at high noon!

Nothing paranormal is reflected in these experiments. Although it goes beyond what is definitely known to say what makes for a focal point, some features do seem to emerge pretty clearly from Schelling’s experiments. For one thing, uniqueness seems to be important.

When asked to pick a point on a map to await another person with the same map but with whom no meeting place has been arranged, many people will select a house on a map with one house and many crossroads, but will select a crossroads on a map with one intersection but many houses.

And, of course, uniqueness makes sense when selecting focal points. Even if both parties select a house in the latter instance, the chance that they will select the same house is small. If one of many houses is distinct, however, it may be selected by some participants — a single mansion may be selected as a focal point even on a map with many houses.

Another element that seems to figure in establishing a focal point is what could be called contextual prominence — for example, the number “one” in a series of numbers, or the center of a circular area or a mountain rising from a plane.

Rioting would be a lot less likely if information about the location of viable focal points were harder to come by.We cannot say how a resident of South-Central L.A. might go about selecting a focal point. In fact it seems consistent with Schelling’s experiments that there would have been a number of focal points, although substantially fewer than there were residents. For example, any of several major intersections, parks or schoolyards may have seemed the natural place for a large number of riot-disposed people to gather following the acquittals in People v. Powell (the original Rodney King beating case), which amounted to a Schelling incident at least in part because for weeks it had been advertised as such by TV and newspaper accounts of the trial.

One can hardly doubt that many residents of South-Central bent on making trouble arrived at places they expected to be “focal” only to find them largely deserted. But Schelling’s work implies that a substantial number of others would have guessed right — would have gone to a major intersection, Korean strip-mall parking lot, or other public space and found the crowd they had expected to find nearing its critical mass — waiting for some of the outliers from non-viable focal points to find their way to more promising locations.

But here is a problem. Those who selected a non-viable focal point — in other words, those who guessed wrong — would now have to find out where everyone else went in order to join them. How did they get this information? Los Angeles’ television stations’ aggressive news coverage of the disturbance from its very beginning seems to have played a key role.

Within minutes after the verdicts were announced in Powell, minicam crews were doing news “live from the scene,” letting everyone in town know where the trouble was. Innocents thus learned what neighborhoods to avoid; but non-innocents, who wanted to take part in the looting, also found out where to go.

Although inadvertently, the stations lowered the search costs for aspiring rioters. Without TV, other techniques would surely have been used by people trying to find out where to go in order to loot and burn with little fear of arrest. But the broadcast media are by far the best way to get accurate information to many people at once. Especially in spread-out places like Los Angeles, rioting would be less likely to occur if information about the location of viable focal points were harder to come by.

The Role of Reputation

Although the conventional “racism-poverty-lack-of-opportunity” explanation is overly broad and somewhat shopworn, we do think it useful in explaining the makeup of a riotous crowd. Racism and poverty would clearly merit social concern even if they had no connection to people’s disposition to engage in rioting. But these are indeed predisposing conditions. One seldom sees riots break out at a convention of orthodontists. Why?

Reducing hooligans’ disposition to violence involves getting them to identify with the larger community.Respectability — a reputation for behaving in a predictable, socially benign manner — is an extremely valuable asset for most people who live in the middle-class world. It is one of the key ingredients in career and personal success, and the need for it serves as a sort of performance bond to keep middle-class people in line. A person to whom respectability matters much should demand better odds before risking arrest and disgrace than would a football hooligan or a member of the American urban underclass or any other socially marginal character to whom respectability is of relatively little value.

Such a person has something that a middle-class person lacks — a great deal of nihilistic freedom of the “nothing to lose” variety. Such freedom, experience suggests, is a perplexing and often malignant possession. Any social policy that would materially improve the life chances of a potential rioter would concurrently raise the value of respectability to such a person, and thus dampen the incentive to participate in civil disorders.

This is not to suggest that reputation matters less to a hooligan than it does to an orthodontist. The question is, reputation for what. A valuable reputation among the thugs is a reputation for hard partying, physical toughness, “sticking by your mates,” and, above all, an ability to engage in predatory behavior without being arrested.

British football hooligans and members of American street gangs do not direct their aggressive behavior at members of their own group, but only at outsiders. Reducing these individuals’ disposition to violence would seem, therefore, to involve getting them to identify with the larger community — making them middle-class, in other words.

Alas, that is easier said than done. Many years of heavy social spending and a “war on poverty” have established that social and economic privations are very difficult to remove even in the long run, and in the short run can hardly be influenced at all. It follows, therefore, that riots are likely to be with us recurrently for the foreseeable future, and that the focus of public debate about riot management should concentrate on symptomatic remedies. Here, at least, some constructive ideas seem worth exploring.

Stopping a Riot

Once it gets started, rioting is difficult to stop by authorities as constrained as American police forces are. Indeed, two different kinds of constraints are important. The more obvious are the rules of constitutional law, which set stringent limits on how police officers may behave toward those whom they try to arrest. Second are the budgetary facts of life that guarantee that modem urban police forces will always be staffed well below peak load demand levels.

Both these constraints should affect the probability of riots occurring and their duration and severity if they do occur. Traditional deterrence theory teaches that in order to discourage crime at the margin, one or both of two things have to happen: either the probability of catching the offender has to be visibly increased, or the harshness of the consequences to the offender in case he is caught have to be tangibly enhanced.

In the case of riots especially, there is not much that police forces can do about either option.

The constraints that exist against riot control spell practical legal immunity, and every rioter understands that.It is hard to imagine that the public would be willing to staff the police department at levels sufficient to deal with a riot immediately if one should break out: it costs a city’s budget about $60,000 or more to add just one additional officer to the force. Nor will there ever be enough prosecutors to try every rioter that could be arrested, nor enough prisons to house them if convicted. Every rioter understands these practical constraints very well. They spell practical legal immunity so long as a riot continues.

If the police try to cover most of the serious action nodes that develop, they will be spread too thin to do much good anywhere. If they abandon some action nodes to concentrate on a few, the trouble can be stopped at the selected locations, but the procedure is like nailing Jell-o to the wall. The riot simply flows around the impediment and goes to locations the police have not covered. Until the National Guard arrives, quadrupling or quintupling available manpower and increasing the apparent risk of arrest, matters simply run out of control.

Of course, authorities prepared to resort to brutality can terminate riots promptly. Buford gives the example of how the Sardinian police militia smothered a soccer riot during the 1990 World Cup matches. Hundreds of rowdy English soccer fans had flown in on chartered planes, and were determined to find trouble. The police did not try to cover every action node at once; this would have left them outnumbered everywhere.

Instead, following textbook military strategy, they massed forces and surrounded first one, then another group of hooligans inglisi, rendering each in turn hors de combat by beating them senseless with truncheons. Few of the Englishmen actually had to be arrested (which would have been very time-consuming for the police). Nevertheless, because they were not allowed to innocently transpire through police lines to re-appear at some less well-defended action node, the riot soon collapsed.

No one would suggest that American police should emulate this style of riot control. And almost as objectionable for other reasons would be censoring television or radio news in order to impede the formation of action nodes.

The Supreme Court has often stated (although not often acted upon) the principle that censorship for compelling reasons of national security does not offend the First Amendment. Even if this tenet is defensible in wartime, it stretches the point considerably to apply it to riots. Clearly there would be serious danger of political opportunism if authorities were permitted to interdict the flow of news merely because they asserted a fear that riots might otherwise ensue.

Presently, every politically incorrect public manifestation would be subject to seizure or arrest. Such a thing actually happened in Chicago some years ago, when a gang of vigilante aldermen, on the unlikely premise that they were trying to forestall civil unrest, stormed an exhibition hall at the Art Institute and commandeered an oil painting that disparagingly portrayed the recently deceased Harold Washington, Chicago’s first black mayor.

Sheer numbers of anti-riot personnel may be more important than tactics or training.There is a third option, however, that might ultimately prove more palatable. According to our analysis, the proximate trigger of a riot is an entrepreneur’s calculation that he is unlikely to be arrested if he breaks a window. If as swiftly as they developed action nodes actually could be covered by the authorities, riots might not begin at all. Cities should consider how they might accomplish this objective.

Experience has shown that the National Guard is not well adapted to the mission of early containment of a riot. It takes the Guard several days to get into action because when it is called, it is not merely foot soldiers that are summoned, but their entire apparatus of logistics and command that must be mobilized as well.

Moreover, even the hint that authorities are thinking about calling out the National Guard could be seen as a provocative acknowledgement of a riot’s incipiency. Public appeals that the Guard be summoned may therefore amount to a sort of focal incident and do almost as much to choreograph the beginning of a riot as to deter its occurrence.

Of course, once it gets into action, the Guard does seem to pacify full-blown riots fairly swiftly. This fact suggests that sheer numbers of anti-riot personnel may be more important than tactics, training, or other variables in quietening civil unrest.

For this reason, cities might well consider the benefits of using a civilian auxiliary to reinforce and supplement the police force. Such a force could be deployed rapidly and demobilized just as fast once the trouble had died down because its command infrastructure, that of the municipal police, is always up and running. Of course it is out of the question for police departments permanently to maintain as many full-time officers as might be required by peak load demand.

An analogy might be drawn to volunteer fire fighters, who receive training, though far less than their full-time professional counterparts, to enable them to meet contingencies too remote to justify commissioning full-time personnel.

The original idea of the militia, as envisioned by the drafters of the United States Constitution, reflected something of the notion that ordinary citizens bore the final responsibility for the security of the communities in which they lived (Dowlut 1983: 93). When not burdened with a command and control superstructure, but simply used to supplement law enforcement resources already in place, a modem equivalent to the militia might well serve to stop trouble before it started.

According to our analysis, riots are apt to be a more or less recurrent, if unpredictable, feature of social life. It is odd that our law enforcement apparatus seems to be designed for a world in which riots do not occur at all. With some imagination, public administrators could ensure that these destructive episodes become rare indeed.


  • Buford, B. (1991) Among the Thugs: The Experience, and the Seduction, of Crowd Violence. New York: W.W. Norton & Co.
  • Dowlut, R. (1983) “The Right to Arms: Does the Constitution or the Predilection of Judges Reign?” Oklahoma Law RevIew 36(1): 65—105.
  • Schelling, T.C. (1960) The Strategy of Conflict. Cambridge: Harvard University Press.

This article was first published in the Cato Journal in 1994 under the title “Understanding Riots.” © Cato Institute. All rights reserved. Reprinted with permission.