The Better Angels of Our Nature: Why Violence Has Declined

by Steven Pinker

  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.159 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.160

  FIGURE 6–7. Rate of deaths in genocides, 1900–2008

  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.

  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.161 Many observers believe that this small pack of génocidaires could easily have been stopped by a military intervention by the world’s great powers.162 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.163 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.

  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.164

  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).165 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.)166 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.

  FIGURE 6–8. Rate of deaths in genocides, 1956–2008

  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.

  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.

  Harff did discover six risk factors that distinguished the genocidal from the nongenocidal crises in three-quarters of the cases.167 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.168 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.

  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.

  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.

  Harff sums up the pathways by which these risk factors erupt into genocide: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.169

  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.”170 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.”171 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.”172 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.”173 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.”174

  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.”175 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.”176 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.”177 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.

  THE TRAJECTORY OF TERRORISM

  Terrorism is a peculiar category of violence, because it has a cockeyed ratio of fear to harm. Compared to the number of deaths from homicide, war, and genocide, the worldwide toll from terrorism is in the noise: fewer than 400 deaths a year since 1968 from international terrorism (where perpetrators from one country cause damage in another), and about 2,500 a year since 1998 from domestic terrorism.178 The numbers we have been dealing with in this chapter have been at least two orders of magnitude higher.

  But after the September 11, 2001, attacks, terrorism became an obsession. Pundits and politicians turned up the rhetoric to eleven, and the word existential (generally modifying threat or crisis) had not seen as much use since the heyday of Sartre and Camus. Experts proclaimed that terrorism made the United States “vulnerable” and “fragile,” and that it threatened to do away with the “ascendancy of the modern state,” “our way of life,” or “civilization itself.”179 In a 2005 essay in The Atlantic, for example, a former White House counterterrorism official confidently prophesied that by the tenth anniversary of the 9/11 attacks the American economy would be shut down by chronic bombings of casinos, subways, and shopping malls, the regular downing of commercial airliners by shoulder-launched missiles, and acts of cataclysmic sabotage at chemical plants.180 The massive bureaucracy of the Department of Homeland Security was created overnigh
t to reassure the nation with such security theater as color-coded terrorist alerts, advisories to stock up on plastic sheeting and duct tape, obsessive checking of identification cards (despite fakes being so plentiful that George W. Bush’s own daughter was arrested for using one to order a margarita), the confiscation of nail clippers at airports, the girding of rural post offices with concrete barriers, and the designation of eighty thousand locations as “potential terrorist targets,” including Weeki Wachee Springs, a Florida tourist trap in which comely women dressed as mermaids swim around in large glass tanks.

  All this was in response to a threat that has killed a trifling number of Americans. The nearly 3,000 deaths from the 9/11 attacks were literally off the chart—way down in the tail of the power-law distribution into which terrorist attacks fall.181 According to the Global Terrorism Database of the National Consortium for the Study of Terrorism and Responses to Terrorism (the major publicly available dataset on terrorist attacks), between 1970 and 2007 only one other terrorist attack in the entire world has killed as many as 500 people.182 In the United States, Timothy McVeigh’s bombing of a federal office building in Oklahoma City in 1995 killed 165, a shooting spree by two teenagers at Columbine High School in 1999 killed 17, and no other attack has killed as many as a dozen. Other than 9/11, the number of people killed by terrorists on American soil during these thirty-eight years was 340, and the number killed after 9/11—the date that inaugurated the so-called Age of Terror—was 11. While some additional plots were foiled by the Department of Homeland Security, many of their claims have turned out to be the proverbial elephant repellent, with every elephant-free day serving as proof of its effectiveness.183