Of all the varieties of violence of which our sorry species is capable, genocide stands apart, not only as the most heinous but as the hardest to comprehend. We can readily understand why from time to time people enter into deadly quarrels over money, honor, or love, why they punish wrongdoers to excess, and why they take up arms to combat other people who have taken up arms. But that someone should want to slaughter millions of innocents, including women, children, and the elderly, seems to insult any claim we may have to comprehend our kind. Whether it is called genocide (killing people because of their race, religion, ethnicity, or other indelible group membership), politicide (killing people because of their political affiliation), or democide (any mass killing of civilians by a government or militia), killing-by-category targets people for what they are rather than what they do and thus seems to flout the usual motives of gain, fear, and vengeance.

Genocide also shocks the imagination by the sheer number of its victims. Rummel, who was among the first historians to try to count them all, famously estimated that during the 20th century 169 million people were killed by their governments. The number is, to be sure, a highball estimate, but most atrocitologists agree that in the 20th century more people were killed by democides than by wars. Matthew White, in a comprehensive overview of the published estimates, reckons that 81 million people were killed by democide and another 40 million by man-made famines (mostly by Stalin and Mao), for a total of 121 million. Wars, in comparison, killed 37 million soldiers and 27 million civilians in battle, and another 18 million in the resulting famines, for a total of 82 million deaths. (White adds, though, that about half of the democide deaths took place during wars and may not have been possible without them.)

Killing so many people in so short a time requires methods of mass production of death that add another layer of horror. The Nazis' gas chambers and crematoria will stand forever as the most shocking visual symbols of genocide. But modern chemistry and railroads are by no means necessary for high-throughput killing. When the French revolutionaries suppressed a revolt in the Vendée region in 1793, they hit upon the idea of packing prisoners into barges, sinking them below the water's surface long enough to drown the human cargo, and then floating them up for the next batch. Even during the Holocaust, the gas chambers were not the most efficient means of killing. The Nazis killed more people with their Einsatzgruppen, or mobile firing squads, which were foreshadowed by other teams of quick-moving soldiers with projectile weapons such as Assyrians in chariots and Mongols on horses. During the genocide of Hutus by Tutsis in Burundi in 1972 (a predecessor of the reverse genocide in Rwanda twenty-two years later), a perpetrator explained:

The bland military term "siege" hides the fact that depriving a city of food and finishing off the weakened survivors is a time-honored and cost-effective form of extermination. As Frank Chalk and Kurt Jonassohn point out in The History and Sociology of Genocide, "The authors of history textbooks hardly ever reported what the razing of an ancient city meant for its inhabitants." One exception is the Book of Deuteronomy, which offers a backdated prophecy that was based on the Assyrian or Babylonian conquest:

Several techniques, several, several. One can gather two thousand persons in a house -- in a prison, let us say. There are some halls which are large. The house is locked. The men are left there for fifteen days without eating, without drinking. Then one opens. One finds cadavers. Not beaten, not anything. Dead.

In the desperate straits to which the enemy siege reduces you, you will eat the fruit of your womb, the flesh of your sons and daughters whom the LORD your God has given you. Even the most refined and gentle of men among you will begrudge food to his own brother, to the wife whom he embraces, and to the last of his remaining children, giving to none of them any of the flesh of his children whom he is eating, because nothing else remains to him, in the desperate straits to which the enemy siege will reduce you in all your towns. She who is the most refined and gentle among you, so gentle and refined that she does not venture to set the sole of her foot on the ground, will begrudge food to the husband whom she embraces, to her own son, and to her own daughter, begrudging even the afterbirth that comes out from between her thighs, and the children that she bears, because she will eat them in secret for lack of anything else, in the desperate straits to which the enemy siege will reduce you in your towns.

Apart from numbers and methods, genocides sear the moral imagination by the gratuitous sadism indulged in by the perpetrators. Eyewitness accounts from every continent and decade recount how victims are taunted, tormented, and mutilated before being put to death. In The Brothers Karamazov, Dostoevsky commented on Turkish atrocities in Bulgaria during the Russo-Turkish War of 1877-78, when unborn children were ripped from their mothers' wombs and prisoners were nailed by their ears to a fence overnight before being hanged: "People speak sometimes about the 'animal' cruelty of man, but that is terribly unjust and offensive to animals. No animal could ever be so cruel as a man, so artfully, so artistically cruel. A tiger simply gnaws and tears, that is all he can do. It would never occur to him to nail people by their ears overnight, even if he were able to do it." My own reading of histories of genocide has left me with images to disturb sleep for a lifetime. I'll recount two that lodge in the mind not because of any gore (though such accounts are common enough) but because of their cold-bloodedness. Both are taken from the philosopher Jonathan Glover's Humanity: A Moral History of the Twentieth Century.

During the Holocaust, Christian Wirth commanded a slave labor compound in Poland, where Jews were worked to death sorting the clothes of their murdered compatriots. Their children had been taken from them and sent to the death camps.

During the Chinese Cultural Revolution of 1966-75, Mao encouraged marauding Red Guards to terrorize "class enemies," including teachers, managers, and the descendants of landlords and "rich peasants," killing perhaps 7 million. In one incident:

Wirth allowed one exception… One Jewish boy around ten was given sweets and dressed up as a little SS man. Wirth and he rode among the prisoners, Wirth on a white horse and the boy on a pony, both using machineguns to kill prisoners (including the boy's mother) at close range.

Young men ransacking an old couple's house found boxes of precious French glass. When the old man begged them not to destroy the glass, one of the group hit him in the mouth with a club, leaving him spitting out blood and teeth. The students smashed the glass and left the couple on their knees crying.

Glover allows himself a comment: "To this ultimate expression of contempt and mockery, no reaction of disgust and anger is remotely adequate."

How could people do these things? Making sense of killing-by-category, insofar as we can do so at all, must begin with the psychology of categories.

People sort other people into mental pigeonholes according to their affiliations, customs, appearances, and beliefs. Though it's tempting to think of this stereotyping as a kind of mental defect, categorization is indispensable to intelligence. Categories allow us to make inferences from a few observed qualities to a larger number of unobserved ones. If I note the color and shape of a fruit and classify it as a raspberry, I can infer that it will taste sweet, satisfy my hunger, and not poison me. Politically correct sensibilities may bridle at the suggestion that a group of people, like a variety of fruit, may have features in common, but if they didn't, there would be no cultural diversity to celebrate and no ethnic qualities to be proud of. Groups of people cohere because they really do share traits, albeit statistically. So a mind that generalizes about people from their category membership is not ipso facto defective. African Americans today really are more likely to be on welfare than whites, Jews really do have higher average incomes than WASPs, and business students really are more politically conservative than students in the arts -- on average.

The cognitive habit of treating people as instances of a category gets truly dangerous when people come into conflict. It turns Hobbes's trio of violent motives -- gain, fear, and deterrence -- from the bones of contention in an individual quarrel to the casus belli in an ethnic war. Historical surveys have shown that genocides are caused by this triad of motives, with, as we shall see, two additional toxins spiked into the brew.

The problem with categorization is that it often goes beyond the statistics. For one thing, when people are pressured, distracted, or in an emotional state, they forget that a category is an approximation and act as if a stereotype applies to every last man, woman, and child. For another, people tend to moralize their categories, assigning praiseworthy traits to their allies and condemnable ones to their enemies. During World War II, for example, Americans thought that Russians had more positive traits than Germans; during the Cold War they thought it was the other way around. Finally, people tend to essentialize groups. As children, they tell experimenters that a baby whose parents have been switched at birth will speak the language of her biological rather than her adoptive parents. As they get older, people tend to think that members of particular ethnic and religious groups share a quasi-biological essence, which makes them homogeneous, unchangeable, predictable, and distinct from other groups.

Some genocides begin as matters of convenience. Natives are occupying a desirable territory or are monopolizing a source of water, food, or minerals, and invaders would rather have it for themselves. Eliminating the people is like clearing brush or exterminating pests, and is enabled by nothing fancier in our psychology than the fact that human sympathy can be turned on or off depending on how another person is categorized. Many genocides of indigenous peoples are little more than expedient grabs of land or slaves, with the victims typed as less than human. Such genocides include the numerous expulsions and massacres of Native Americans by settlers or governments in the Americas, the brutalization of African tribes by King Leopold of Belgium in the Congo Free State, the extermination of the Herero by German colonists in South-West Africa, and the attacks on Darfuris by government-encouraged Janjaweed militias in the 2000s.

When conquerors find it expedient to suffer the natives to live so that they can provide tribute and taxes, genocide can have a second down-to-earth function. A reputation for a willingness to commit genocide comes in handy for a conqueror because it allows him to present a city with an ultimatum to surrender or else. To make the threat credible, the invader has to be prepared to carry it out. This was the rationale behind the annihilation of the cities of western Asia by Genghis Khan and his Mongol hordes.

When a dehumanized people is in a position to defend itself or turn the tables, it can set a Hobbesian trap of group-against-group fear. Either side may see the other as an existential threat that must be preemptively taken out. After the breakup of Yugoslavia in the 1990s, Serbian nationalists' genocide of Bosnians and Kosovars was partly fueled by fears that they would be the victims of massacres themselves.

Once the conquerors have absorbed a city or territory into an empire, they may keep it in line with the threat that they will come down on any revolt like a ton of bricks. In 68 CE the governor of Alexandria called in Roman troops to put down a rebellion by the Jews against Roman rule. According to the historian Flavius Josephus, "Once [the Jews] were forced back, they were unmercifully and completely destroyed. Some were caught in the open field, others forced into their houses, which were plundered and then set on fire. The Romans showed no mercy to the infants, had no regard for the aged, and went on in the slaughter of persons of every age, until all the place was overflowed with blood, and 50,000 Jews lay dead." Similar tactics have been used in 20th-century counterinsurgency campaigns, such as the ones by the Soviets in Afghanistan and right-wing military governments in Indonesia and Central America.

If members of a group have seen their comrades victimized, have narrowly escaped victimization themselves, or paranoically worry they have been targeted for victimization, they may stoke themselves into a moralistic fury and seek vengeance on their perceived assailants. Like all forms of revenge, a retaliatory massacre is pointless once it has to be carried out, but a welladvertised and implacable drive to carry it out, regardless of its costs at the time, may have been programmed into people's brains by evolution, cultural norms, or both as a way to make the deterrent credible.

Widow: Don Francesco. You murdered my husband, because he would not bend. And his oldest son Paolo, because he swore revenge. But Vitone is only nine, and dumb-witted. He never speaks.

These Hobbesian motives don't fully explain why predation, preemption, or revenge should be directed against entire groups of people rather than the individuals who get in the way or make trouble. The cognitive habit of pigeonholing may be one reason, and another is explained in The Godfather: Part II when the young Vito Corleone's mother begs a Sicilian don to spare the boy's life:

Francesco: He will grow strong.

Francesco: I'm not afraid of his words.

Widow: The child cannot harm you.

Widow: He is weak.

Francesco: He will be a man, and then he will come for revenge.

The solidarity among the members of a family, clan, or tribe -- in particular, their resolve to avenge killings -- makes them all fair game for someone with a bone to pick with any one of them. Though equal-sized groups in frequent contact tend to constrain their revenge to an-eye-for-an-eye reciprocity, repeated violations may turn episodic anger into chronic hatred. As Aristotle wrote, "The angry man wishes the object of his anger to suffer in return; hatred wishes its object not to exist." When one side finds itself with an advantage in numbers or tactics, it may seize the opportunity to impose a final solution. Feuding tribes are well aware of genocide's practical advantages. The anthropologist Rafael Karsten worked with the Jivaro of Amazonian Ecuador (a tribe that contributed one of the long bars to the graph of rates of death in warfare in figure 2-2) and recounts their ways of war:

And come for revenge he does. Later in the film the grown Vito returns to Sicily, seeks an audience with the don, whispers his name into the old man's ear, and cuts him open like a sturgeon.

Whereas the small feuds within the sub-tribes have the character of a private blood-revenge, based on the principle of just retaliation, the wars between the different tribes are in principle wars of extermination. In these there is no question of weighing life against life; the aim is to completely annihilate the enemy tribe… The victorious party is all the more anxious to leave no single person of the enemy's people, not even small children, alive, as they fear lest these should later appear as avengers against the victors.

Half a world away, the anthropologist Margaret Durham offered a similar vignette from an Albanian tribe that ordinarily abided by norms for measured revenge:

In February 1912 an amazing case of wholesale justice was reported to me… A certain family of the Fandi bairak [subtribe] had long been notorious for evil-doing -- robbing, shooting, and being a pest to the tribe. A gathering of all the heads condemned all the males of the family to death. Men were appointed to lay in wait for them on a certain day and pick them off; and on that day the whole seventeen of them were shot. One was but five and another but twelve years old. I protested against thus killing children who must be innocent and was told: "It was bad blood and must not be further propagated." Such was the belief in heredity that it was proposed to kill an unfortunate woman who was pregnant, lest she should bear a male and so renew the evil.

The essentialist notion of "bad blood" is one of several biological metaphors inspired by a fear of the revenge of the cradle. People anticipate that if they leave even a few of a defeated enemy alive, the remnants will multiply and cause trouble down the line. Human cognition often works by analogy, and the concept of an irksome collection of procreating beings repeatedly calls to mind the concept of vermin. Perpetrators of genocide the world over keep rediscovering the same metaphors to the point of cliché. Despised people are rats, snakes, maggots, lice, flies, parasites, cockroaches, or (in parts of the world where they are pests) monkeys, baboons, and dogs. "Kill the nits and you will have no lice," wrote an English commander in Ireland in 1641, justifying an order to kill thousands of Irish Catholics. "A nit would make a louse," recalled a Californian settler leader in 1856 before slaying 240 Yuki in revenge for their killing of a horse. "Nits make lice," said Colonel John Chivington before the Sand Creek Massacre, which killed hundreds of Cheyenne and Arapaho in 1864. Cankers, cancers, bacilli, and viruses are other insidious biological agents that lend themselves as figures of speech in the poetics of genocide. When it came to the Jews, Hitler mixed his metaphors, but they were always biological: Jews were viruses; Jews were bloodsucking parasites; Jews were a mongrel race; Jews had poisonous blood.

Metaphorical thinking goes in both directions. Not only do we apply disgust metaphors to morally devalued peoples, but we tend to morally devalue people who are physically disgusting (a phenomenon we encountered in chapter 4 when considering Lynn Hunt's theory that a rise in hygiene in Europe caused a decline in cruel punishments). At one pole of the continuum, whiteclad ascetics who undergo rituals of purification are revered as holy men and women. At the other, people living in degradation and filth are reviled as subhuman. The chemist and writer Primo Levi described this spiral during the transport of Jews to the death camps in Germany:

The human mind has evolved a defense against contamination by biological agents: the emotion of disgust. Ordinarily triggered by bodily secretions, animal parts, parasitic insects and worms, and vectors of disease, disgust impels people to eject the polluting substance and anything that looks like it or has been in contact with it. Disgust is easily moralized, defining a continuum in which one pole is identified with spirituality, purity, chastity, and cleansing and the other with animality, defilement, carnality, and contamination. And so we see disgusting agents as not just physically repellent but also morally contemptible. Many metaphors in the English language for a treacherous person use a disease vector as their vehicle -- a rat, a louse, a worm, a cockroach. The infamous 1990s term for forced displacement and genocide was ethnic cleansing.

The SS escort did not hide their amusement at the sight of men and women squatting wherever they could, on the platforms and in the middle of the tracks, and the German passengers openly expressed their disgust: people like this deserve their fate, just look how they behave. These are not Menschen, human beings, but animals, it's clear as the light of day.

The emotional pathways to genocide -- anger, fear, and disgust -- can occur in various combinations. In Worse than War, a history of 20th-century genocide, the political scientist Daniel Goldhagen points out that not all genocides have the same causes. He classifies them according to whether the victim group is dehumanized (a target of moralized disgust), demonized (a target of moralized anger), both, or neither. A dehumanized group may be exterminated like vermin, such as the Hereros in the eyes of German colonists, Armenians in the eyes of Turks, black Darfuris in the eyes of Sudanese Muslims, and many indigenous peoples in the eyes of European settlers. A demonized group, in contrast, is thought to be equipped with the standard human reasoning faculties, which makes them all the more culpable for embracing a heresy or rejecting the one true faith. Among these modern heretics were the victims of communist autocracies, and the victims of their opposite number, the right-wing dictatorships in Chile, Argentina, Indonesia, and El Salvador. Then there are the out-and-out demons -- groups that manage to be both repulsively subhuman and despicably evil. This is how the Nazis saw the Jews, and how Hutus and Tutsis saw each other. Finally, there may be groups that are not reviled as evil or subhuman but are feared as potential predators and eliminated in preemptive attacks, such as in the Balkan anarchy following the breakup of Yugoslavia.

So far I have tried to explain genocide in the following way. The mind's habit of essentialism can lump people into categories; its moral emotions can be applied to them in their entirety. The combination can transform Hobbesian competition among individuals or armies into Hobbesian competition among peoples. But genocide has another fateful component. As Solzhenitsyn pointed out, to kill by the millions you need an ideology. Utopian creeds that submerge individuals into moralized categories may take root in powerful regimes and engage their full destructive might. For this reason it is ideologies that generate the outliers in the distribution of genocide death tolls. Divisive ideologies include Christianity during the Crusades and the Wars of Religion (and in an offshoot, the Taiping Rebellion in China); revolutionary romanticism during the politicides of the French Revolution; nationalism during the genocides in Ottoman Turkey and the Balkans; Nazism in the Holocaust; and Marxism during the purges, expulsions, and terror-famines in Stalin's Soviet Union, Mao's China, and Pol Pot's Cambodia.

Why should utopian ideologies so often lead to genocide? At first glance it seems to make no sense. Even if an actual utopia is unattainable for all kinds of practical reasons, shouldn't the quest for a perfect world at least leave us with a better one -- a world that is 60 percent of the way to perfection, say, or even 15 percent? After all, a man's reach must exceed his grasp. Shouldn't we aim high, dream the impossible dream, imagine things that never were and ask "why not"?

Utopian ideologies invite genocide for two reasons. One is that they set up a pernicious utilitarian calculus. In a utopia, everyone is happy forever, so its moral value is infinite. Most of us agree that it is ethically permissible to divert a runaway trolley that threatens to kill five people onto a side track where it would kill only one. But suppose it were a hundred million lives one could save by diverting the trolley, or a billion, or -- projecting into the indefinite future -- infinitely many. How many people would it be permissible to sacrifice to attain that infinite good? A few million can seem like a pretty good bargain.

In Blood and Soil: A World History of Genocide and Extermination from Sparta to Darfur, the historian Ben Kiernan notes another curious feature of utopian ideologies. Time and again they hark back to a vanished agrarian paradise, which they seek to restore as a healthful substitute for prevailing urban decadence. In chapter 4 we saw that after the Enlightenment had emerged from the intellectual bazaar of cosmopolitan cities, the German counter-Enlightenment romanticized the attachment of a people to their land -- the blood and soil of Kiernan's title. The ungovernable metropolis, with its fluid population and ethnic and occupational enclaves, is an affront to a mindset that envisions a world of harmony, purity, and organic wholeness. Many of the nationalisms of the 19th and early 20th centuries were guided by utopian images of ethnic groups flourishing in their native homelands, often based on myths of ancestral tribes who settled the territory at the dawn of time. This agrarian utopianism lay behind Hitler's dual obsessions: his loathing of Jewry, which he associated with commerce and cities, and his deranged plan to depopulate Eastern Europe to provide farmland for German city-dwellers to colonize. Mao's massive agrarian communes and Pol Pot's expulsion of Cambodian city-dwellers to rural killing fields are other examples.

The second genocidal hazard of a utopia is that it has to conform to a tidy blueprint. In a utopia, everything is there for a reason. What about the people? Well, groups of people are diverse. Some of them stubbornly, perhaps essentially, cling to values that are out of place in a perfect world. They may be entrepreneurial in a world that works by communal sharing, or bookish in a world that works by labor, or brash in a world that works by piety, or clannish in a world that works by unity, or urban and commercial in a world that has returned to its roots in nature. If you are designing the perfect society on a clean sheet of paper, why not write these eyesores out of the plans from the start?

Not only that, but consider the people who learn about the promise of a perfect world yet nonetheless oppose it. They are the only things standing in the way of a plan that could lead to infinite goodness. How evil are they? You do the math.

Commercial activities, which tend to be concentrated in cities, can themselves be triggers of moralistic hatred. As we shall see in chapter 9, people's intuitive sense of economics is rooted in tit-for-tat exchanges of concrete goods or services of equivalent value -- say, three chickens for one knife. It does not easily grasp the abstract mathematical apparatus of a modern economy, such as money, profit, interest, and rent. In intuitive economics, farmers and craftsmen produce palpable items of value. Merchants and other middlemen, who skim off a profit as they pass goods along without causing new stuff to come into being, are seen as parasites, despite the value they create by enabling transactions between producers and consumers who are unacquainted or separated by distance. Moneylenders, who loan out a sum and then demand additional money in return, are held in even greater contempt, despite the service they render by providing people with money at times in their lives when it can be put to the best use. People tend to be oblivious to the intangible contributions of merchants and moneylenders and view them as bloodsuckers. (Once again the metaphor comes from biology.) Antipathy toward individual middlemen can easily transfer to antipathy to ethnic groups. The capital necessary to prosper in middlemen occupations consists mainly of expertise rather than land or factories, so it is easily shared among kin and friends, and it is highly portable. For these reasons it's common for particular ethnic groups to specialize in the middleman niche and to move to whatever communities currently lack them, where they tend to become prosperous minorities -- and targets of envy and resentment. Many victims of discrimination, expulsion, riots, and genocide have been social or ethnic groups that specialize in middlemen niches. They include various bourgeois minorities in the Soviet Union, China, and Cambodia, the Indians in East Africa and Oceania, the Ibos in Nigeria, the Armenians in Turkey, the Chinese in Indonesia, Malaysia, and Vietnam, and the Jews in Europe.

Democides are often scripted into the climax of an eschatological narrative, a final spasm of violence that will usher in millennial bliss. The parallels between the utopian ideologies of the 19th and 20th centuries and the apocalyptic visions of traditional religions have often been noticed by historians of genocide. Daniel Chirot, writing with the social psychologist Clark McCauley, observes:

Marxist eschatology actually mimicked Christian doctrine. In the beginning, there was a perfect world with no private property, no classes, no exploitation, and no alienation -- the Garden of Eden. Then came sin, the discovery of private property, and the creation of exploiters. Humanity was cast from the Garden to suffer inequality and want. Humans then experimented with a series of modes of production, from the slave, to the feudal, to the capitalist mode, always seeking the solution and not finding it. Finally there came a true prophet with a message of salvation, Karl Marx, who preached the truth of Science. He promised redemption but was not heeded, except by his close disciples who carried the truth forward. Eventually, however, the proletariat, the carriers of the true faith, will be converted by the religious elect, the leaders of the party, and join to create a more perfect world. A final, terrible revolution will wipe out capitalism, alienation, exploitation, and inequality. After that, history will end because there will be perfection on earth, and the true believers will have been saved.

It was not an accident that Hitler promised a Thousand Year Reich, a millennium of perfection, similar to the thousand-year reign of goodness promised in Revelation before the return of evil, the great battle between good and evil, and the final triumph of God over Satan. The entire imagery of his Nazi Party and regime was deeply mystical, suffused with religious, often Christian, liturgical symbolism, and it appealed to a higher law, to a mission decreed by fate and entrusted to the prophet Hitler.

Finally, there are the job requirements. Would you want the stress and responsibility of running a perfect world? Utopian leadership selects for monumental narcissism and ruthlessness. Its leaders are possessed of a certainty about the rectitude of their cause and an impatience for incremental reforms or on-the-fly adjustments guided by feedback from the human consequences of their grand schemes. Mao, who had his image plastered all over China and his little red book of sayings issued to every citizen, was described by his doctor and only confidant Li Zhisui as voracious for flattery, demanding of sexual servicing by concubines, and devoid of warmth and compassion. In 1958 he had a revelation that the country could double its steel production in a year if peasant families contributed to the national output by running backyard smelters. On pain of death for failing to meet the quotas, peasants melted down their woks, knives, shovels, and doorknobs into lumps of useless metal. It was also revealed to him that China could grow large quantities of grain on small plots of land, freeing the rest for grasslands and gardens, if farmers planted the seedlings deep and close together so that class solidarity would make them grow strong and thick. Peasants were herded into communes of 50,000 to implement this vision, and anyone who dragged his feet or pointed out the obvious was executed as a class enemy. Impervious to signals from reality informing him that his Great Leap Forward was a great leap backward, Mao masterminded a famine that killed between 20 million and 30 million people.

Drawing on the work of the historians Joachim Fest and George Mosse, they also comment on Nazi eschatology:

In the 1st century CE, Tacitus wrote, "A shocking crime was committed on the unscrupulous initiative of a few individuals, with the blessing of more, and amid the passive acquiescence of all." According to the political scientist Benjamin Valentino in Final Solutions, that division of labor applies to the genocides of the 20th century as well. A leader or small clique decides that the time for genocide is right. He gives the go-ahead to a relatively small force of armed men, made up a mixture of true believers, conformists, and thugs (often recruited, as in medieval armies, from the ranks of criminals, drifters, and other unemployable young men). They count on the rest of the population not to get in their way, and thanks to features of social psychology that we will explore in chapter 8, they generally don't. The psychological contributors to genocide, such as essentialism, moralization, and utopian ideologies, are engaged to different degrees in each of these constituencies. They consume the minds of the leaders and the true believers but have to tip the others only enough to allow the leaders to make their plans a reality. The indispensability of leaders to 20th-century genocide is made plain by the fact that when the leaders died or were removed by force, the killings stopped.

The motives of leaders are critical in understanding genocide, because the psychological ingredients -- the mindset of essentialism; the Hobbesian dynamic of greed, fear, and vengeance; the moralization of emotions like disgust; and the appeal of utopian ideologies -- do not overcome an entire population at once and incite them to mass killing. Groups that avoid, distrust, or even despise each other can coexist without genocide indefinitely. Think, for example, of African Americans in the segregated American South, Palestinians in Israel and the occupied territories, and Africans in South Africa under apartheid. Even in Nazi Germany, where anti-Semitism had been entrenched for centuries, there is no indication that anyone but Hitler and a few fanatical henchmen thought it was a good idea for the Jews to be exterminated. When a genocide is carried out, only a fraction of the population, usually a police force, military unit, or militia, actually commits the murders.

If this analysis is on the right track, genocides can emerge from toxic reactions among human nature (including essentialism, moralization, and intuitive economics), Hobbesian security dilemmas, millennial ideologies, and the opportunities available to leaders. The question now is: how has this interaction changed over the course of history?

It's not an easy question to answer, because historians have never found genocide particularly interesting. Since antiquity the stacks of libraries have been filled with scholarship on war, but scholarship on genocide is nearly nonexistent, though it killed more people. As Chalk and Jonassohn point out of ancient histories, "We know that empires have disappeared and that cities were destroyed, and we suspect that some wars were genocidal in their results; but we do not know what happened to the bulk of the populations involved in these events. Their fate was simply too unimportant. When they were mentioned at all, they were usually lumped together with the herds of oxen, sheep, and other livestock."

As soon as one realizes that the sackings, razings, and massacres of past centuries are what we would call genocide today, it becomes utterly clear that genocide is not a phenomenon of the 20th century. Those familiar with classical history know that the Athenians destroyed Melos during the 5th-century-BCE Peloponnesian War; according to Thucydides, "the Athenians thereupon put to death all who were of military age and made slaves of the women and children." Another familiar example is the Romans' destruction of Carthage and its population during the Third Punic War in the 3rd century BCE, a war so total that the Romans, it was said, sowed salt into the ground to make it forever unfarmable. Other historical genocides include the real-life bloodbaths that inspired the ones narrated in the Iliad, the Odyssey, and the Hebrew Bible; the massacres and sackings during the Crusades; the suppression of the Albigensian heresy; the Mongol invasions; the European witch hunts; and the carnage of the European Wars of Religion.

2. The Spanish Conquest of the New World 1492-1600

4. Genocidal Massacres in Early Modern Southeast Asia

3. Guns and Genocide in East Asia 1400-1600

1. Classical Genocide and Early Modern Memory

Part One: Early Imperial Expansion

7. Genocidal Violence in Nineteenth-Century Australia

5. The English Conquest of Ireland, 1565-1603

6. Colonial North America, 1600-1776

8. Genocide in the United States

Part Two. Settler Colonialism

The authors of recent histories of mass killing are adamant that the idea of an unprecedented "century of genocide" (the 20th) is a myth. On their first page Chalk and Jonassohn write, "Genocide has been practiced in all regions of the world and during all periods in history," and add that their eleven case studies of pre-20th-century genocides "are not intended to be either exhaustive or representative." Kiernan agrees: "A major conclusion of this book is that genocide indeed occurred commonly before the twentieth century." One can see what he means with a glance at the first page of his table of contents:

Rummel has fitted a number to his own conclusion that "the mass murder by emperors, kings, sultans, khans, presidents, governors, generals, and other rulers of their own citizens or of those under their protection or control is very much part of our history." He counts 133,147,000 victims of sixteen democides before the 20th century (including ones in India, Iran, the Ottoman Empire, Japan, and Russia) and surmises that there may have been 625,716,000 democide victims in all.

9. Settler Genocides in Africa, 1830-1910

These authors did not compile their lists by indiscriminately piling up every historical episode in which a lot of people died. They are careful to note, for example, that the Native American population was decimated by disease rather than by a program of extermination, while particular incidents were blatantly genocidal. In an early example, Puritans in New England exterminated the Pequot nation in 1638, after which the minister Increase Mather asked his congregation to thank God "that on this day we have sent six hundred heathen souls to Hell." This celebration of genocide did not hurt his career. He later became president of Harvard University, and the residential house with which I am currently affiliated is named after him (motto: Increase Mather's Spirit!).

Mather was neither the first nor the last to thank God for genocide. As we saw in chapter 1, Yahweh ordered the Hebrew tribes to carry out dozens of them, and in the 9th century BCE the Moabites returned the favor by massacring the inhabitants of several Hebrew cities in the name of their god, Ashtar-Chemosh. In a passage from the Bhagavad-Gita (written around 400 CE), the Hindu god Krishna upbraids the mortal Arjuna for being reluctant to slay an enemy faction that included his grandfather and tutor: "There is no better engagement for you than fighting on religious principles; and so there is no need for hesitation… The soul can never be cut to pieces by any weapon, nor burned by fire… [Therefore] you are mourning for what is not worthy of grief." Inspired by the conquests of Joshua, Oliver Cromwell massacred every man, woman, and child in an Irish town during the reconquest of Ireland, and explained his actions to Parliament: "It has pleased God to bless our endeavour at Drogheda. The enemy were about 3,000 strong in the town. I believe we put to the sword the whole number." The English Parliament passed a unanimous motion "that the House does approve of the execution done at Drogheda as an act of both justice to them and mercy to others who may be warned of it."

The shocking truth is that until recently most people didn't think there was anything particularly wrong with genocide, as long as it didn't happen to them. One exception was the 16th-century Spanish priest Antonio de Montesinos, who protested the appalling treatment of Native Americans by the Spanish in the Caribbean -- and who was, in his own words, "a voice of one crying in the wilderness." There were, to be sure, military codes of honor, some from the Middle Ages, that ineffectually attempted to outlaw the killing of civilians in war, and occasional protests by thinkers of early modernity such as Erasmus and Hugo Grotius. But only in the late 19th century, when citizens began to protest the brutalization of peoples in the American West and the British Empire, did objections to genocide become common. Even then we find Theodore Roosevelt, the future "progressive" president and Nobel Peace laureate, writing in 1886, "I don't go so far as to think that the only good Indians are the dead Indians, but I believe nine out of ten are, and I shouldn't like to inquire too closely in the case of the tenth." The critic John Carey documents that well into the 20th century the British literary intelligentsia viciously dehumanized the teeming masses, whom they considered to be so vulgar and soulless as not to have lives worth living. Genocidal fantasies were not uncommon. In 1908, for example, D. H. Lawrence wrote:

During World War II, when Americans were asked in opinion polls what should be done with the Japanese after an American victory, 10 to 15 percent volunteered the solution of extermination.

If I had my way, I would build a lethal chamber as big as the Crystal Palace, with a military band playing softly, and a Cinematograph working brightly; then I'd go out in the back streets and main streets and bring them in, all the sick, the halt, and the maimed; I would lead them gently, and they would smile me a weary thanks; and the band would softly bubble out the "Hallelujah Chorus."

The turning point came after the war. The English language did not even have a word for genocide until 1944, when the Polish lawyer Raphael Lemkin coined it in a report on Nazi rule in Europe that would be used a year later to brief the prosecutors at the Nuremberg Trials. In the aftermath of the Nazi destruction of European Jewry, the world was stunned by the enormity of the death toll and by horrific images from the liberated camps: assembly-line gas chambers and crematoria, mountains of shoes and eyeglasses, bodies stacked up like cordwood. In 1948 Lemkin got the UN to approve a Convention on the Prevention and Punishment of the Crime of Genocide, and for the first time in history genocide, regardless of who the victims were, was a crime. James Payne notes a perverse sign of progress. Today's Holocaust deniers at least feel compelled to deny that the Holocaust took place. In earlier centuries the perpetrators of genocide and their sympathizers boasted about it.

No small part in the new awareness of the horrors of genocide was a willingness of Holocaust survivors to tell their stories. Chalk and Jonassohn note that these memoirs are historically unusual. Survivors of earlier genocides had treated them as humiliating defeats and felt that talking about them would only rub in history's harsh verdict. With the new humanitarian sensibilities, genocides became crimes against humanity, and survivors were witnesses for the prosecution. Anne Frank's diary, which recorded her life in hiding in Nazioccupied Amsterdam before she was deported to her death in Bergen-Belsen, was published by her father shortly after the war. Memoirs of deportations and death camps by Elie Wiesel and Primo Levi were published in the 1960s, and today Frank's Diary and Wiesel's Night are among the world's most widely read books. In the years that followed, Aleksandr Solzhenitsyn, Anchee Min, and Dith Pran shared their harrowing memories of the communist nightmares in the Soviet Union, China, and Cambodia. Soon other survivors -- Armenians, Ukrainians, Gypsies -- began to add their stories, joined more recently by Bosnians, Tutsis, and Darfuris. These memoirs are a part of a reorientation of our conception of history. "Throughout most of history," Chalk and Jonassohn note, "only the rulers made news; in the twentieth century, for the first time, it is the ruled who make the news."

Anyone who grew up with Holocaust survivors knows what they had to overcome to tell their stories. For decades after the war they treated their experiences as shameful secrets. On top of the ignominy of victimhood, the desperate straits to which they were reduced could remove the last traces of their humanity in ways they could be forgiven for wanting to forget. At a family occasion in the 1990s, I met a relative by marriage who had spent time in Auschwitz. Within seconds of meeting me he clenched my wrist and recounted this story. A group of men had been eating in silence when one of them slumped over dead. The others fell on his body, still covered in diarrhea, and pried a piece of bread from his fingers. As they divided it, a fierce argument broke out when some of the men felt their share was an imperceptible crumb smaller than the others'. To tell a story of such degradation requires extraordinary courage, backed by a confidence that the hearer will understand it as an accounting of the circumstances and not of the men's characters.

Though the abundance of genocides over the millennia belies the centuryof-genocide claim, one still wonders about the trajectory of genocide before, during, and since the 20th century. Rummel was the first political scientist to try to put some numbers together. In his duology Death by Government (1994) and Statistics of Democide (1997) he analyzed 141 regimes that committed democides in the 20th century through 1987, and a control group of 73 that did not. He collected as many independent estimates of the death tolls as he could find (including ones from pro- and antigovernment sources, whose biases, he assumed, would cancel each other out) and, with the help of sanity checks, chose a defensible value near the middle of the range. His definition of "democide" corresponds roughly to the UCDP's "one-sided violence" and to our everyday concept of "murder" but with a government rather than an individual as the perpetrator: the victims must be unarmed, and the killing deliberate. Democides thus include ethnocides, politicides, purges, terrors, killings of civilians by death squads (including ones committed by private militias to which the government turns a blind eye), deliberate famines from blockades and confiscation of food, deaths in internment camps, and the targeted bombing of civilians such as those in Dresden, Hamburg, Hiroshima, and Nagasaki. Rummel excluded the Great Leap Forward from his 1994 analyses, on the understanding that it was caused by stupidity and callousness rather than malice.

Partly because the phrase "death by government" figured in Rummel's definition of democide and in the title of his book, his conclusion that almost 170 million people were killed by their governments during the 20th century has become a popular meme among anarchists and radical libertarians. But for several reasons, "governments are the main cause of preventable deaths" is not the correct lesson to draw from Rummel's data. For one thing, his definition of "government" is loose, embracing militias, paramilitaries, and warlords, all of which could reasonably be seen as a sign of too little government rather than too much. White examined Rummel's raw data and calculated that the median democide toll by the twenty-four pseudo-governments on his list was around 100,000, whereas the median death toll caused by recognized governments of sovereign states was 33,000. So one could, with more justification, conclude that governments, on average, cause three times fewer deaths than alternatives to government. Also, most governments in recent periods do not commit democides at all, and they prevent a far greater number of deaths than the democidal ones cause, by promoting vaccination, sanitation, traffic safety, and policing.

But the main problem with the anarchist interpretation is that it isn't governments in general that kill large numbers of people but a handful of governments of a specific type. To be exact, three-quarters of all the deaths from all 141 democidal regimes were committed by just four governments, which Rummel calls the dekamegamurderers: the Soviet Union with 62 million, the People's Republic of China with 35 million, Nazi Germany with 21 million, and 1928-49 nationalist China with 10 million. Another 11 percent of the total were killed by eleven megamurderers, including Imperial Japan with 6 million, Cambodia with 2 million, and Ottoman Turkey with 1.9 million. The remaining 13 percent of the deaths were spread out over 126 regimes. Genocides don't exactly fall into a power-law distribution, if for no other reason than that the smaller massacres that would go into the tall spine tend not to be counted as "genocides." But the distribution is enormously lopsided, conforming to an 80:4 rule --80 percent of the deaths were caused by 4 percent of the regimes.

Also, deaths from democide were overwhelmingly caused by totalitarian governments: the communist, Nazi, fascist, militarist, or Islamist regimes that sought to control every aspect of the societies they ruled. Totalitarian regimes were responsible for 138 million deaths, 82 percent of the total, of which 110 million (65 percent of the total) were caused by the communist regimes. Authoritarian regimes, which are autocracies that tolerate independent social institutions such as businesses and churches, came in second with 28 million deaths. Democracies, which Rummel defines as governments that are open, competitive, elected, and limited in their power, killed 2 million (mainly in their colonial empires, together with food blockades and civilian bombings during the world wars). The skew of the distribution does not just reflect the sheer number of potential victims that totalitarian behemoths like the Soviet Union and China had at their disposal. When Rummel looked at percentages rather than numbers, he found that totalitarian governments of the 20th century racked up a death toll adding up to 4 percent of their populations. Authoritarian governments killed 1 percent. Democracies killed four tenths of 1 percent.

Rummel was one of the first advocates of the Democratic Peace theory, which he argues applied to democides even more than to wars. "At the extremes of Power," Rummel writes, "totalitarian communist governments slaughter their people by the tens of millions; in contrast, many democracies can barely bring themselves to execute even serial murderers." Democracies commit fewer democides because their form of governance, by definition, is committed to inclusive and nonviolent means of resolving conflicts. More important, the power of a democratic government is restricted by a tangle of institutional restraints, so a leader can't just mobilize armies and militias on a whim to fan out over the country and start killing massive numbers of citizens. By performing a set of regressions on his dataset of 20th-century regimes, Rummel showed that the occurrence of democide correlates with a lack of democracy, even holding constant the countries' ethnic diversity, wealth, level of development, population density, and culture (African, Asian, Latin American, Muslim, Anglo, and so on). The lessons, he writes, are clear: "The problem is Power. The solution is democracy. The course of action is to foster freedom."

What about the historical trajectory? Rummel tried to break down his 20th-century democides by year, and I've reproduced his data, scaled by world population, in the gray upper line in figure 6-7. Like deaths in wars, deaths in democides were concentrated in a savage burst, the midcentury Hemoclysm. This blood-flood embraced the Nazi Holocaust, Stalin's purges, the Japanese rape of China and Korea, and the wartime firebombings of cities in Europe and Japan. The left slope also includes the Armenian genocide during World War I and the Soviet collectivization campaign, which killed millions of Ukrainians and kulaks, the so-called rich peasants. The right slope embraces the killing of millions of ethnic Germans in newly communized Poland, Czechoslovakia, and Romania, and the victims of forced collectivization in China. It's uncomfortable to say that there's anything good in the trends shown in the graph, but in an important sense there is. The world has seen nothing close to the bloodletting of the 1940s since then; in the four decades that followed, the rate (and number) of deaths from democide went precipitously, if lurchingly, downward. (The smaller bulges represent killings by Pakistani forces during the Bangladesh war of independence in 1971 and by the Khmer Rouge in Cambodia in the late 1970s.) Rummel attributes the falloff in democide since World War II to the decline of totalitarianism and the rise of democracy.

Sources: Data for the gray line, 1900-1987, from Rummel, 1997. Data for the black line, 1955-2008, from the Political Instability Task Force (PITF) State Failure Problem Set, 1955-2008, Marshall, Gurr, & Harff, 2009; Center for Systemic Peace, 2010. The death tolls for the latter were geometric means of the ranges in table 8.1 in Harff, 2005, distributed across years according to the proportions in the Excel database. World population figures from U. S. Census Bureau, 2010c. Population figures for the years 1900-1949 were taken from McEvedy & Jones, 1978, and multiplied by 1.01 to make them commensurable with the rest.

FIGURE 6-7: Rate of deaths in genocides, 1900-2008

Rummel's dataset ends in 1987, just when things start to get interesting again. Soon communism fell and democracies proliferated -- and the world was hit with the unpleasant surprise of genocides in Bosnia and Rwanda. In the impression of many observers, these "new wars" show that we are still living, despite all we should have learned, in an age of genocide.

The historical thread of genocide statistics has recently been extended by the political scientist Barbara Harff. During the Rwanda genocide, some 700,000 Tutsis were killed in just four months by about 10,000 men with machetes, many of them drunkards, addicts, ragpickers, and gang members hastily recruited by the Hutu leadership. Many observers believe that this small pack of génocidaires could easily have been stopped by a military intervention by the world's great powers. Bill Clinton in particular was haunted by his own failure to act, and in 1998 he commissioned Harff to analyze the risk factors and warning signs of genocide. She assembled a dataset of 41 genocides and politicides between 1955 (shortly after Stalin died and the process of decolonization began) and 2004. Her criteria were more restrictive than Rummel's and closer to Lemkin's original definition of genocide: episodes of violence in which a state or armed authority intends to destroy, in whole or in part, an identifiable group. Only five of the episodes turned out to be "genocide" in the sense in which people ordinarily understand the term, namely an ethnocide, in which a group is singled out for destruction because of its ethnicity. Most were politicides, or politicides combined with ethnocides, in which members of an ethnic group were thought to be aligned with a targeted political faction.

The graph shows that the two decades since the Cold War have not seen a recrudescence of genocide. On the contrary, the peak in mass killing (putting aside China in the 1950s) is located in the mid-1960s to late 1970s. Those fifteen years saw a politicide against communists in Indonesia (1965-66, "the year of living dangerously," with 700,000 deaths), the Chinese Cultural Revolution (1966-75, around 600,000), Tutsis against Hutus in Burundi (1965-73, 140,000), Pakistan's massacre in Bangladesh (1971, around 1.7 million), north-againstsouth violence in Sudan (1956-72, around 500,000), Idi Amin's regime in Uganda (1972-79, around 150,000), the Cambodian madness (1975-79, 2.5 million), and a decade of massacres in Vietnam culminating in the expulsion of the boat people (1965-75, around half a million). The two decades since the end of the Cold War have been marked by genocides in Bosnia from 1992 to 1995 (225,000 deaths), Rwanda (700,000 deaths), and Darfur (373,000 deaths from 2003 to 2008). These are atrocious numbers, but as the graph shows, they are spikes in a trend that is unmistakably downward. (Recent studies have shown that even some of these figures may be overestimates, but I will stick with the datasets.) The first decade of the new millennium is the most genocide-free of the past fifty years. The UCDP numbers are restricted to a narrower time window and, like all their estimates, are more conservative, but they show a similar pattern: the Rwanda genocide in 1994 leaps out from all the other episodes of one-sided killing, and the world has seen nothing like it since.

In figure 6-7 I plotted Harff's PITF data on the same axes with Rummel's. Her figures generally come in well below his, especially in the late 1950s, for which she included far fewer victims of the executions during the Great Leap Forward. But thereafter the curves show similar trends, which are downward from their peak in 1971. Because the genocides from the second half of the 20th century were so much less destructive than those of the Hemoclysm, I've zoomed in on her curve in figure 6-8. The graph also shows the death rates in a third collection, the UCDP One-Sided Violence Dataset, which includes any instance of a government or other armed authority killing at least twenty-five civilians in a year; the perpetrators need not intend to destroy the group per se.

Sources: PITF estimates, 1955-2008: same as for figure 6-7. UCDP, 1989-2007: "High Fatality" estimates from http://www. pcr. uu. se/research/ucdp/datasets/ (Kreutz, 2008; Kristine & Hultman, 2007) divided by world population from U. S. Census Bureau, 2010c.

Harff was tasked not just with compiling genocides but with identifying their risk factors. She noted that virtually all of them took place in the aftermath of a state failure such as a civil war, revolution, or coup. So she assembled a control group with 93 cases of state failure that did not result in genocide, matched as closely as possible to the ones that did, and ran a logistic regression analysis to find out which aspects of the situation the year before made the difference.

FIGURE 6-8: Rate of deaths in genocides, 1956-2008

Some factors that one might think were important turned out not to be. Measures of ethnic diversity didn't matter, refuting the conventional wisdom that genocides represent the eruption of ancient hatreds that inevitably explode when ethnic groups live side by side. Nor did measures of economic development matter. Poor countries are more likely to have political crises, which are necessary conditions for genocides to take place, but among the countries that did have crises, the poorer ones were no more likely to sink into actual genocide.

The other three predictors are familiar from the theory of the Liberal Peace. Harff vindicated Rummel's insistence that democracy is a key factor in preventing genocides. From 1955 to 2008 autocracies were three and a half times more likely to commit genocides than were full or partial democracies, holding everything else constant. This represents a hat trick for democracy: democracies are less likely to wage interstate wars, to have large-scale civil wars, and to commit genocides. Partial democracies (anocracies) are more likely than autocracies to have violent political crises, as we saw in Fearon and Laitin's analysis of civil wars, but when a crisis does occur, the partial democracies are less likely than autocracies to become genocidal.

Harff did discover six risk factors that distinguished the genocidal from the nongenocidal crises in three-quarters of the cases. One was a country's previous history of genocide, presumably because whatever risk factors were in place the first time did not vanish overnight. The second predictor was the country's immediate history of political instability -- to be exact, the number of regime crises and ethnic or revolutionary wars it had suffered in the preceding fifteen years. Governments that feel threatened are tempted to eliminate or take revenge on groups they perceive to be subversive or contaminating, and are more likely to exploit the ongoing chaos to accomplish those goals before opposition can mobilize. A third was a ruling elite that came from an ethnic minority, presumably because that multiplies the leaders' worries about the precariousness of their rule.

Another trifecta was scored by openness to trade. Countries that depend more on international trade, Harff found, are less likely to commit genocides, just as they are less likely to fight wars with other countries and to be riven by civil wars. The inoculating effects of trade against genocide cannot depend, as they do in the case of interstate war, on the positive-sum benefits of trade itself, since the trade we are talking about (imports and exports) does not consist in exchanges with the vulnerable ethnic or political groups. Why, then, should trade matter? One possibility is that Country A might take a communal or moral interest in a group living within the borders of Country B. If B wants to trade with A, it must resist the temptation to exterminate that group. Another is that a desire to engage in trade requires certain peaceable attitudes, including a willingness to abide by international norms and the rule of law, and a mission to enhance the material welfare of its citizens rather than implementing a vision of purity, glory, or perfect justice.

The last predictor of genocide is an exclusionary ideology. Ruling elites that are under the spell of a vision that identifies a certain group as the obstacle to an ideal society, putting it "outside the sanctioned universe of obligation," are far more likely to commit genocide than elites with a more pragmatic or eclectic governing philosophy. Exclusionary ideologies, in Harff's classification, include Marxism, Islamism (in particular, a strict application of Sharia law), militaristic anticommunism, and forms of nationalism that demonize ethnic or religious rivals.

Almost all genocides and politicides of the last half-century were either ideological, exemplified by the Cambodian case, or retributive, as in Iraq [Saddam Hussein's 1988-91 campaign against Iraqi Kurds]. The scenario that leads to ideological genocide begins when a new elite comes to power, usually through civil war or revolution, with a transforming vision of a new society purified of unwanted or threatening elements. Retributive geno-politicides occur during a protracted internal war… when one party, usually the government, seeks to destroy its opponent's support base [or] after a rebel challenge has been militarily defeated.

Harff sums up the pathways by which these risk factors erupt into genocide:

The decline of genocide over the last third of a century, then, may be traced to the upswing of some of the same factors that drove down interstate and civil wars: stable government, democracy, openness to trade, and humanistic ruling philosophies that elevate the interests of individuals over struggles among groups.

For all the rigor that a logistic regression offers, it is essentially a meat grinder that takes a set of variables as input and extrudes a probability as output. What it hides is the vastly skewed distribution of the human costs of different genocides -- the way that a small number of men, under the sway of a smaller number of ideologies, took actions at particular moments in history that caused outsize numbers of deaths. Shifts in the levels of the risk factors certainly pushed around the likelihood of the genocides that racked up thousands, tens of thousands, and even hundreds of thousands of deaths. But the truly monstrous genocides, the ones with tens of millions of victims, depended not so much on gradually shifting political forces as on a few contingent ideas and events.

The appearance of Marxist ideology in particular was a historical tsunami that is breathtaking in its total human impact. It led to the dekamegamurders by Marxist regimes in the Soviet Union and China, and more circuitously, it contributed to the one committed by the Nazi regime in Germany. Hitler read Marx in 1913, and although he detested Marxist socialism, his National Socialism substituted races for classes in its ideology of a dialectical struggle toward utopia, which is why some historians consider the two ideologies "fraternal twins." Marxism also set off reactions that led to politicides by militantly anticommunist regimes in Indonesia and Latin America, and to the destructive civil wars of the 1960s, 1970s, and 1980s stoked by the Cold War superpowers. The point is not that Marxism should be morally blamed for these unintended consequences, just that any historical narrative must acknowledge the sweeping repercussions of this single idea. Valentino notes that no small part of the decline of genocide is the decline of communism, and thus "the single most important cause of mass killing in the twentieth century appears to be fading into history." Nor is it likely that it will come back into fashion. During its heyday, violence by Marxist regimes was justified with the saying "You can't make an omelet without breaking eggs." The historian Richard Pipes summarized history's verdict: "Aside from the fact that human beings are not eggs, the trouble is that no omelet has emerged from the slaughter." Valentino concludes that "it may be premature to celebrate 'the end of history,' but if no similarly radical ideas gain the widespread applicability and acceptance of communism, humanity may be able to look forward to considerably less mass killing in the coming century than it experienced in the last."

On top of that singularly destructive ideology were the catastrophic decisions of a few men who took the stage at particular moments in the 20th century. I have already mentioned that many historians have joined the chorus "No Hitler, no Holocaust." But Hitler was not the only tyrant whose obsessions killed tens of millions. The historian Robert Conquest, an authority on Stalin's politicides, concluded that "the nature of the whole Purge depends in the last analysis on the personal and political drives of Stalin." As for China, it is inconceivable that the record-setting famine of the Great Leap Forward would have occurred but for Mao's harebrained schemes, and the historian Harry Harding noted of the country's subsequent politicide that "the principal responsibility for the Cultural Revolution -- a movement that affected tens of millions of Chinese -- rests with one man. Without a Mao, there could not have been a Cultural Revolution." With such a small number of data points causing such a large share of the devastation, we will never really know how to explain the most calamitous events of the 20th century. The ideologies prepared the ground and attracted the men, the absence of democracy gave them the opportunity, but tens of millions of deaths ultimately depended on the decisions of just three individuals.