Genocidal Mutation and the Challenge of Definition

Henry C Heriault. Metaphilosophy. Volume 41, Issue 4. July 2010.

The definition of the term “genocide” has been contested since soon after Raphael Lemkin first advanced it in Axis Control of Occupied Europe, so much so that two standards in the field, Frank Chalk and Kurt Jonassohn’s The History and Sociology of Genocide (3-53) and the Encyclopedia of Genocide include extensive considerations of the different views of the term’s proper definition and of the associated controversies.

The definition produced through great‐power negotiations, and stipulated in the United Nations’ Convention on the Prevention and Punishment of the Crime of Genocide, is often taken as the starting point. According to the Convention, “genocide” is any of the following acts committed with the intent to destroy, in whole or in part, a national, ethnical, racial or religious group, as such:

  • Killing members of the group;
  • Causing serious bodily or mental harm to members of the group;
  • Deliberately inflicting on the group conditions of life calculated to bring about its physical destruction in whole or in part;
  • Imposing measures intended to prevent births within the group;
  • Forcibly transferring children of the group to another group.

This is sometimes tested against the definition developed by Lemkin, which included broader forms of societal and cultural destruction not necessarily focused on direct physical killing of members of the group. Even within the United Nations this issue has been raised, particularly by the Whitaker Report. This evaluation of the implementation of the Convention raised a number of important issues, including the problems of intent, recognition of violence against a gender group as genocide, and inclusion of political groups as a type of target group.

Among scholars views have run over a wide range. At one extreme are Stephen Katz and others, who directly violate Lemkin’s basic concept of genocide by asserting that genocide is the total physical destruction of a group and that in history only the Holocaust fits this definition. At the other extreme are usages of the term that stretch it to cover many different forms of human suffering and processes, such as terrorist acts or ecological impacts that harm groups of people either by undermining their cohesion in a certain culturally important geographic area or causing death.

The usual critique of narrow definitions of genocide is that they exclude too many cases and thus render unavailable to many groups the protections of the U.N. Convention that they should have. The usual critique of wide definitions is that they expand the term to vagueness and render it of little conceptual or legal value for understanding, preventing, and intervening against genocide: if all sorts of disparate events and acts are genocidal, then genocide does not have a specific meaning that separates it from other acts and allows clear understanding and is a commonplace that does not demand our attention.

The focus on getting the right definition, however, even where it has led to important theoretical insights, misunderstands the nature of “defining” as it should pertain to the concept “genocide.” The problem is not that there is not a good, universal concept of genocide—almost all theorists would agree to something like “genocide is the destruction of a group.” What is at stake, rather, is the way this general understanding is used to designate which events are genocides and which are not—that is, the way in which it is applied or “operationalized” in legal and scholarly practice. Thus, most definitions gain precision by promoting a specific notion of the kind(s) of group destruction to count as genocide. Further, the U.N. definition and most others require intentional destruction, which is generally taken to mean that genocide results from purposeful acts by perpetrators whose goals are identifiable and have a substantial role in producing the resulting destruction.

Not all specification is limiting. The U.N. definition, for instance, allows for partial destruction and even partial targeting of a defined group, which widens application. This range is then narrowed by the specification of only four types of groups who can be victims of genocide. And so on.

Definitional Denial

Regardless of how precise a definition is, there is a complicating factor that operates independently of the scholarly and legal quality of a definition. Most cases of genocide are met with denials that they are, in fact, genocide. A denial claim can be made through an argument (1) that the events being called genocide are not happening or did not happen at all or (2) that they are or were something other than genocide. Both typically entail a falsification of facts, but this second form of denial can be tremendously effective because it does not rely on as extreme a falsification of facts. The usual approach is to argue the events claimed to be a genocide do not fit “the definition of genocide,” which is usually the U.N. definition but might be another. Israel Charny has called instances of the second kind of denial that focus on features of the referent definition of genocide “definitional denial” or “definitionalism” . As the U.N. Convention has become better known and legal cases based on it more frequent and prominent, especially in the 1990s and 2000s, this form of denial has become prominent.

Definitional denials can be categorized temporally. First are denials of past genocides, such as the Armenian Genocide with the related genocides of Assyrians and Pontian Greeks. Those who wish to deny or cover up a past genocide usually present manipulated historical data against a standard definition of genocide. For instance, in the case of the Armenian Genocide, some deniers reject the veracity of the historical documents directly exposing the organizing leadership of the Young Turk government. They then claim that the events of 1915 were local massacres or spontaneous intercommunal violence rather than a centrally intended, planned, and controlled mass extermination. In this case, the emphasis of the U.N. definition on “intent” allows deniers to make their case with a relatively small level of falsification. By rejecting just a few documents, they get the perpetrators off on a technicality.

This has a greater effect than is at first apparent. Even when statements failing to acknowledge the Armenian Genocide as genocide recognize that Armenians suffered significantly, the problem is serious. Regardless of how much lip service is paid to Armenian suffering, by explicitly preventing the use of the term “genocide,” deniers insinuate into the minds of outside parties a subtle fallacy: because what happened to Armenians is not marked by the word “genocide,” it is not as bad as what is claimed and thus is not bad on an objective scale. We can see this kind of diminution of the Armenian Genocide in the denial argument of D. Cameron. Watt’s approach is an example of a second subtype of definitionalism: deniers sometimes manipulate the generally understood definition of genocide itself. Watt argues that what happened to Armenians was not genocide because Turks coveted Armenian women and took many into “Turkish harems.” For Watt, the fact that not all Ottoman Armenians were targeted for death, contrary to his claim about European Jews in the Holocaust, shows that this was not genocide. Watt does not discuss the definition of genocide he is using, so readers must assume he is using the term as generally understood, that is, with the meaning given in the U.N. Genocide Convention. But his argument ignores a prominent feature of the U.N. definition, is generally accepted in other definitions as well, that a victim group might be targeted only in part. Thus, leaving intact a segment of the victim group does not necessarily disqualify the events as genocide—especially when the segment is relatively small, as in the Armenian case. The fact that the women in question were typically sexually enslaved with the same attitude that Armenian sheep were incorporated into the herds of Turks, as dehumanized objects rather than out of respect for their lives, shows how minor this exception was relative to the general genocidal intent of the perpetrators. Indeed, with the growing recognition that sexual violence against women is a method of genocide, this “exception” can in fact be recognized as a central part of the Armenian Genocide itself: women and girls were forcibly converted to Islam and “married” to Turkish and other Muslim men, which not only destroyed their identity as Armenians and thereby reduced the Armenian community but also destroyed Armenian family, community, and identity structures, a recognized element or goal of genocide. Watt also states that the Armenians did not suffer genocide because their persecution was based on “religious” rather than “racial” prejudice. This, of course, ignores an equally obvious feature of the U.N. definition, that the target can be a religious group. By thus misrepresenting the definition of genocide, Watt transforms an obviously genocidal feature of the Armenian Genocide into supposed evidence that the events in question did not constitute genocide.

Genocides are also denied as they occur. For example, during the Rwandan Genocide of April to June 1994, officials of the United States and certain European countries as well as most representatives on the U.N. Security Council refused for weeks and more to characterize the mass violence as genocide, despite the clear evidence they possessed that it fitted the U.N. definition. Even before April some knowledgeable sources had been predicting genocide, based on good evidence. The initial killing should have obviated doubts as to whether the perpetrator group was going to act on its stated intention to destroy the main victim group, the Tutsi minority in Rwanda. Once hundreds of thousands or even tens of thousands had been killed by order of or at the direct instigation of government leaders, there should have been no doubt that the intent to commit genocide was being actualized. Instead, U.S. and European governments and the U.N. Security Council refused to acknowledge publicly the obvious fact of genocide in Rwanda. When pressed, they claimed that the facts were still unclear and so it would be hasty to call the situation genocide. There is good evidence that it was a conscious policy in the U.S. administration and in the Security Council to avoid use of the term “genocide.”

The objective of these definitionalist maneuvers was to forestall intervention. The maneuvers succeeded brilliantly. It is not unreasonable to argue that the entire genocide could have been prevented once the initial intent was clear months before the killing began, if the United Nations or some other outside power had intervened decisively. Even more compelling is the argument that once tens or hundreds of thousands were killed there could be no reasonable doubt that genocide was transpiring; the failure to recognize the events as genocide at that point prevented intervention that still could have saved many of the 850,000 victims.

In retrospect, the events in Rwanda have been universally recognized as an absolutely clear instance of U.N.‐defined genocide. In 1998, President Bill Clinton publicly apologized for his own and others’ failure to intervene. Yet, he still begged out of meaningful responsibility by reiterating his administration’s original definitionalist claim that during much of the genocide they really did not understand that what was happening was genocide.

In Rwanda, the stakes of definition could not have been higher: the consequence of definitional denial was the genocide itself. What is more, the fact that such a broad array of leaders charged with enforcing the U.N. Convention on Genocide misrecognized and even misrepresented the events in Rwanda to prevent application of the definition implies strongly that whatever the particulars of the definition taken as standard, these or similar officials would have falsely perceived and even falsified the data to prevent its application.

The Armenian case fits the U.N. definition very tightly as well. Indeed, Armenians were subjected to all five methods described in the U.N. definition, and there do exist some documents showing the direct intent of the perpetrators. It might seem that the responses to each of the two definitionalist approaches discussed in reference to the Armenian Genocide are fairly straightforward. The facts of the case or the correct definition need simply be reasserted to refute the denial argument in question. But in the first case, the intent requirement of the U.N. definition is stringent, and introduction of the merest question about the sources demonstrating the intent of major perpetrators has been enough to raise doubts for those who do not have expertise on the documentation or an understanding of the denial process. Thus, deniers are able to exploit an element of the U.N. definition. Watt’s approach is also more of a problem than it might seem, because reassertion of the U.N. definition of genocide is not decisive in a context in which the definition of genocide is recognized as contested. If nothing else, it is clear that scholars do not agree on the definition and some would support Watt’s usage. Similarly for Rwanda. The failure of the definition was due to political manipulations by parties who refused to recognize the genocide for what it was. The problem is, the fact that events in Rwanda were tested against a fixed definition that they eventually fitted means that, even if a deliberative process eventually—after genocide was accomplished—determined that genocide had occurred, the stalling tactic would have been used no matter what definition were the standard. Clearly, the process of applying a definition of genocide needs to work when it can actually contribute to prevention of all or some of the killing. The fact that the definition of genocide could be used as an obstacle to prevention and intervention suggests deep questions about how definition of genocide is approached.

We are left with three serious problems: (1) definitional denials of a past genocide can be used to forestall perpetually recognition of that genocide, (2) definitional denials of an ongoing genocide, even when ultimately overcome, will likely do their job of stymieing intervention efforts, and (3) improving the standard fixed definition will not solve these two problems, as denials are advanced despite the fact that the U.N. definition might—but is not allowed to—work in these situations.

As I argue below, it is not that such a definition is flawed in a way that can be addressed that is the problem, it is that the usual approach to definition of genocide opens the door to definitional denial. To begin to see this, consideration of a third type of definitionalism will be helpful. This might be termed “anticipatory denial” of a planned but not yet executed genocide. In such a case, perpetrators intentionally misrepresent and tweak planned actions themselves to maximize the possibility of definitional denial once the actions are taken. This might be the case in Sudan today, where the widespread use of local militia apparently not directly under central governmental control is the main mechanism of violence. To the extent that it is intentional, this measure anticipates application of the U.N. definition, with the focus on “intent” that is typically understood as central governmental decision making and direction, and preemptively defends against it.

Similarly, in the case of Serbian genocidal violence against Bosnian Muslims, it is possible that leaders deliberately organized activities in such a way that they did not seem quite to fit the U.N. definition of genocide. As I discuss more fully below, the perpetrators often used rape instead of direct killing to accomplish genocidal goals of (1) preventing births in the victim community, (2) destroying the social and cultural identity of the victim group, and (3) inflicting destructive bodily and psychological harm on members of the victim group. The use of rape instead of direct killing made what Serbs were perpetrating appear not to be genocide to many outside observers. The Serbs could accomplish their goal of producing the deaths of many Bosnians, changing the demographics of Bosnia, and gaining land in Bosnia with little interference. Their denial tactics helped stall intervention for years, even if there was eventually some kind of intervention that prevented further gains by the Serbs and after the fact some measures were taken to prosecute some perpetrators. The relative success of this approach to genocide is clear in the land gains allowed to Serbs as part of the “resolution” of their “conflict” with Bosnians through the Dayton Accords. Anticipatory definitionalism is yet more effective in contexts where the political and strategic considerations of outside powers run against even eventual intervention or mitigation.

The Serbian case highlights two additional problems. (4) The definition of any crime allows potential perpetrators to modify their anticipated acts, but in the case of genocide, modification does not mean genocide will not occur. Regarding murder, for instance, a modification of the plan means that the intended victim will not be killed, even if a lesser crime will be committed. In the case of mass killing, however, the fact that perpetrators modify their acts not to fit a restricted definition of genocide does not mean that they will not slaughter, torture, and rape masses of victims. Thus, modification is not necessarily successful deterrence. The more potentially restrictive a definition of genocide, the more it allows this kind of modification without successful prevention or mitigation of full or partial destruction of a group. (5) Anticipatory denial can work with any fixed definition of genocide. Each fixed point in a definition occasions different approaches that can accomplish group destruction. The fixed points give perpetrators something explicit to work around. For instance, if possible victim groups are limited to four types, this allows the perpetrators to target or claim to be targeting a group that is not one of the four types.

The Metaphysics of Genocide Definition: Inevitable Gaps and Ambiguities

This last problem suggests a more general feature of all fixed definitions. Whether a definition is general or specific, abstract or detailed, it will inevitably contain points of ambiguity or points requiring interpretation. The U.N. definition, as the standard, is an instructive case in point. We can begin with the provision that a group might be targeted “in whole or in part.” The “in part” element allows recognition that in most if not all cases of genocide there are exceptions to the destruction. But addressing this problem raises others: How big a part? What kind of part? How does one define “a part”—by class, gender, geography, and so on, or arbitrarily (one of ten, and so on)? Setting a percentage threshold is not consistent with other elements of the definition, especially the idea that genocide occurs when specific people are targeted because they are members of a group. An example might be the residents of Nanjing in 1937-38, who were subjected to mass killing by the invading Japanese Army in order to intimidate the remaining Chinese population of more than a hundred times as many to stop resisting Japanese conquest. It is clear that not all seven hundred thousand residents were targeted, and probably between one hundred thousand and two hundred thousand were killed, with tens of thousands of women raped, but even if all had been killed, they would have represented less than 1 percent of the total Chinese population at the time, which could be considered to have been a national and to a large extent an “ethnical” group. Such a small percentage would likely disqualify this event if a percentage threshold were used, despite the scale of the violence. Similarly if a number threshold is used: this would simply privilege larger groups and disqualify small groups from consideration, regardless of the percentages concerned. A five‐hundred‐person indigenous group might be entirely wiped out, but because the casualty number would be relatively low on an objective scale, the destruction would not count as genocide.

A second point of ambiguity has already been mentioned. What is a proper kind of target group of genocide, that is, what kind of group must be targeted for something to be considered genocide? Political groups were excluded in order to appease the Soviet Union, whose mass killings under Stalin typically targeted politically designated groups. What of targeting primarily women and not men within a given racial, national, religious, or ethnic group, as in the early modern European witch hunts? Gender groups would also appear to be a legitimate inclusion, as argued in the Whitaker Report. But no matter how many kinds of groups are included as potential targets of genocide, there is a deeper problem, familiar to anyone who has studied concepts of such entities as “nations” and “races.” These are not natural categories, such as “plant,” and their general meanings are much contested, as is membership in any particular instantiation. To deal with this complex problem, the Nazis went as far as making laws that stipulated who was Jewish and who not, a definition that conflicted with the religious self‐identification of many thus defined as Jews. The conventional nature of group definition in this case is also reflected in such things as differences between the status of children of mixed Jewish/non‐Jewish marriages in Nazi Germany and in other European countries. A similar problem, though not specifically genocidal, comes through in the history of U.S. Jim Crow segregation and beyond, as state governments often developed bizarre rules to determine who was and was not “negro” because the lines between “white” and “black” had become so blurred after years of miscegenation. One might also consider the Hemshin of Turkey, a historically Armenian group that converted to Islam over a long period of time starting in the seventeenth century yet whose members retain many vestiges of their Armenian heritage, including in many cases speaking Armenian. Are they Armenian or not? Having converted to Islam long before the twentieth century, they were not targeted in the Armenian Genocide. And what of Janissaries and their progeny, forced from non‐Muslim families to be raised as elite Ottoman soldiers—were those from, for instance, Armenian origins Turkish or Armenian? What of Protestant and Catholic Armenians, who were not part of the official Armenian millet, and who were sometimes targeted in the Genocide and sometimes exempted from it? In these cases the complexities of biology and especially culture, heritage, ancestry, self‐identification, and perception by others are all operative and challenge uncomplicated, straightforward application of any definition of genocide that incorporates designation of specific kinds of groups as the targets.

The concept of “intent” is likewise the site of debilitating complexities. Does the fact that a large mass of human beings has been killed necessarily imply that the intent for group destruction was there, even if no direct evidence of perpetrator intent exists? Certainly in many cases of genocide, direct intent was there, with careful plans and command structures (the Armenian Genocide, the Holocaust, and the Rwandan Genocide are examples), but just as often the issue of intent is not clear in this way. Native American genocides were accomplished by a range of perpetrator groups (pick your major European or Western Hemisphere European‐derived home government) and by a range of actors within the various North American and South American colonies and then states. While there was central governmental authority, for instance, in the genocidal actions and policies of the U.S. government in the eighteenth, nineteenth, and twentieth centuries, a significant portion of the overall destruction was initiated by local groups (individuals, families, communities, companies, and so forth) without direct central governmental control. What is more, the motives of particular, especially local, participants in the overall destruction of Native Americans might not have entailed the overall destruction of Native Americans or even of a local tribal group. The goal might have been land acquisition, “pacification,” ranching space, and so on. The acts could have been single murders or small‐scale deportations and dispossessions. What if all participants had partial conceptions of a given genocide, such that the concept “we are taking actions to destroy this group in whole or part” was not present in the mind of any single individual perpetrator, yet together the “vector sum” of the various motivations and attitudes necessarily entailed a process to destroy the genocide’s victim group? While many participants’ actions or statements might be necessary for the destruction that occurs, none might be sufficient for it, that is, no specific intention motivates genocide (as a whole), no particular order commands others to commit genocide. “Intent” as measured by usual standards in criminal cases, such as murders, is not present, yet mass destruction has resulted.

The concept here is not the same as “functionalism” as typically used in the case of the Holocaust. There the question is temporal: Did the total destruction of Jews evolve through a series of approaches to the “Jewish Problem” or was the intent to destroy Jews present at the beginning of the series of actions we now recognize as the process of the Holocaust? For instance, were the Nuremberg Laws intended as a step toward total destruction of the Jews or did they merely contribute to that destruction in a way that can be perceived only in hindsight? My concept here is not temporally indexed. At no point does a total destruction scheme emerge that all perpetrators share in; there is no concept of the “Final Solution,” but a final solution nevertheless is accomplished. In the case of the Holocaust, from early on various approaches to solving the entire Jewish Problem were considered; the total destruction of Jews was the one ultimately enacted. Thus, some kind of dramatic action by Nazis as a whole was envisioned from the start. In some other cases of genocide, this kind of overarching concept is not as clearly present, even if some in the perpetrator group(s) might have it in mind. Contrary to reducing the effects of genocide, however, the absence of an overarching concept can actually make genocide much more difficult to stop, as hydralike it has no single ideology to oppose, no single perpetrator motive or set to stop.

The problem raised by this kind of genocide is more than an issue of imperfect definition. In such cases as this, there is a different kind of event causation, a “social causation.” Frequently, genocide perpetrator groups are conceived as identity groups. The conception can be crude or more refined, such as philosopher Karen Kovach’s approach. But no matter how sophisticated, the notion is still that genocide results from a lining up of shared ideas or ideological commitments within a group. This follows a standard view of national groups as pure identity formations, that is, sets of individuals “identical” in some key way(s), such as ideological commitment, language, and/or culture. This notion is central not only in obviously flawed work on nations and nationalisms such as but also more subtly in sophisticated treatments such as. The approaches of Partha and, whatever its shortcomings, Eric put the lie to claims of homogeneity for nations. While it might be tempting to see such extreme cases as Nazi Germany as homogeneous in genocidal intent, this is far from true, as suggested by the jailing of figures such as Armin T. Wegner and thousands of other Germans for opposition to the Nazis. The same is true for most other cases of genocide. Nations (and ethnic groups) are much more internally complex and varied than a model such as Kovach’s allows. Thus, in the typical cases of genocide perpetrated by one national or ethnic group against another or others, the perpetrator groups are not homogeneous.

It is this heterogeneity that grounds a notion of social causation as something more complex than a large individual acting. If there are challenging philosophical questions about the nature of individual intent as it is translated into action, they are multiplied greatly in the case of a social act. Genocide is a social act, requiring a large group of people to carry it out. Intent, however, is applied in human rights law and almost always in genocide studies as an inherently individual concept—the notion of intent is either that specific individuals direct a genocide and have an overarching intent (thus, intent is located in an individual mind) or all participants in a genocide share the same ideology, goal, and so on, and thus intent is located in each individual. As argued above, this notion might not capture the complex reality of what produces a given social act, particularly a social act involving a large number of people. While it might be useful as a tool to determine the relative guilt and appropriate punishment of individual perpetrators of a genocide (a person who is motivated by something less than a full genocidal intent would presumably be guilty of the genocidal equivalent of manslaughter or negligent homicide, rather than the genocidal equivalent of first degree murder, the absence of specific individuals with full genocidal intent should have no determinative bearing on whether or not genocide as a social act has occurred.

What is more, individuals might pursue courses of genocide without recognizing that this is what they are doing. That is, they might be unaware or self‐deceiving about what they are doing or feel it is justified and thus believe themselves to be doing something other than committing genocide. For instance, many Turks, victims of their own government’s propaganda, felt that, by driving out and killing Armenians, they were defending Turkey against an internal enemy that had colluded with the Russians and/or the British and French; similarly, propagandized fears of Jews as a conspiratorial danger drove many in Germany to support their elimination from Europe. Perpetrators might even conceive of themselves as benefiting the victims, for instance by “saving their souls,” as in the case of purification killings of “witches.” Rationalizations of violence are strongly supported when attitudes and acts are part of a large‐scale, ideologically grounded group activity, complete with peer reinforcement and pressure. Similarly, preexisting structural domination and/or widespread prejudice can produce culpable violence without fitting standard notions of self‐aware intent.16 The notion of intent becomes so complex in these kinds of cases that it seems to lose explanatory power—and yet the body counts continue to mount.

It is possible to counter that standard notions of intent that focus not on motives but on purposive action would cover such cases. For instance, as long as perpetrators intended to kill women condemned as witches, their beliefs about what they were doing and motives for it are irrelevant. They intended the killings. This line of argument does not appear to succeed in practice, however. One might consider the case of the Ache, an indigenous group targeted for genocide in Paraguay in the 1970s. The Paraguayan government’s defense against charges of genocide was simple: the purpose was not to kill the Ache but to gain possession of their lands. A similar argument can be made regarding the Herero Genocide of 1904 to 1907 in what was then German South West Africa, though there was an extermination order by the relevant German governmental authority: the Herero stood in the way of German expansion in the region, including blocking a railroad project. They revolted against that German expansion, and in order to allow that expansion, the Germans killed off the vast majority of them. Had the Herero not resisted that expansion (after all, they could have just relocated themselves into the prime real estate of the Kalahari), the Germans would not have killed them off, and so it was not their overarching intention to destroy the Herero. On this view, Ache and Herero deaths were similar to—to use current euphemistic terminology—”collateral damage” in the process of land acquisition.

And what of “free‐fire zones” in the Vietnam War, in which the U.S. military declared a certain zone, which could be hundreds of square kilometers, free of all but enemy combatants and authorized soldiers, pilots, and the like to use any weapons against anyone found within the area, that is, to kill people within the zone freely. The fact that such zones were usually not at all clear of noncombatant civilians indicates the problem. Here we have the designation of a national group (Vietnamese) in part (those who inhabit a given territory) targeted simply because of their status as members of this part of that group. Yet, the question of intent is complicated, because it is not to kill people based on their identity but to kill enemy combatants, even though in practice the former replaced the latter. The examples could go on, but the point is made: instrumental mass killing can be represented as unintentional destruction of a group. On this basis, most if not all cases of apparent genocide can be claimed not to fit the U.N. definition or any alternative requiring intent. It should be noted that even where the U.N. or another authority rejects this denial argument, the problem is not solved: it is simply the case that the authority in question has by fiat decided that the correct form of intent was present in a particular case. While deliberation with this result is welcome and should be the basis of any application of any definition of genocide, the way definitions of genocide are currently developed, decisions against recognizing such acts as genocide are just as justified—and in practice frequent. Deliberative application of a genocide definition is a step toward the definitional process I present below, but it is still limited by the fixed nature of any definition used.

There is also the issue of incremental intent. In the case of Cambodia, though the original targets, Vietnamese and Buddhists, fit the U.N. definition, the target group continued to expand as the process went forward: new categories of Cambodians were added, until about 1.8 million had been killed. It is likely that the perpetrators did not conceive of this scale of destruction initially, but rather thought to implement a vague “purification” of Cambodian society. At least on a functionalist view, even the Holocaust appears to have evolved somewhat this way, where initial plans to deport Jews and mass killings turned only by 1942 into full annihilation. While intent might be clear late in a process like this, there is a tendency to forget that, in the absence of the emergence of later intent, evaluations of earlier motives would be much more complicated—there would be no working backward from the ultimate destruction and conceptualizations based on it. Among other things, this implies that without preexisting full intent the standard approach to definition is not useful for genocide prevention and intervention.

A related issue is the question of how events are parsed. Before an event or process can be determined to be genocide or not, it must be recognized to be a unified event that can be discussed in this way. When is a sequence of killings a unified event rather than distinct acts that do not qualify as genocide? Perceived unity often depends on evidence of central planning or intention or on the unity of the target group, each of which we can now see is frequently complex and unclear.

The whole/part, group, and intent ambiguities, as well as others not discussed, are not just fodder for deniers. Genuine disagreements about the status of events means that judgments about how to interpret them in a given situation will determine whether those events do or do not constitute genocide, with no final answer determined by rational analysis of the facts against the referent definition. These ambiguities are functions of metaphysical indeterminacies regarding such issues as causation and the mind‐world relationship. Sharper deniers and perpetrators exploit these metaphysical indeterminacies. Recognition of the fundamental indeterminacy of any definition of genocide—which is not the same as the claim that no good judgments can be made as to whether a set of events were or are genocide or not—is captured well in Helen Fein’s insight that genocide is a “fuzzy concept” such that many cases might legitimately be considered genocide or some other type of human rights violation. While some cases will generally be clear and unambiguous, there will always be some cases that are not, their inclusion in or exclusion from the category depending on judgment calls that are ultimately speculative (which is not the same as groundless or arational). What Fein perhaps does not recognize is that there are no core cases of genocide, and that manipulations of the definition or adoptions of definitions different from the U.N. standard can render any case of what is usually considered genocide not genocide. Thus, the boundaries of the term are fuzzy in a manner that can affect any case considered.

Stalemate or Opportunity? From Fixed to Flexible Definition

In the face of these fundamental challenges, it might seem that no definition can work and we must recognize every definition of genocide as provisional, flawed, and manipulable. For a less serious matter, this view might be an academic issue, one that could spur interesting but ultimately low‐stakes debates. However, properly defining genocide is at the core of prevention and intervention, as the cases of Rwanda, the former Yugoslavia, and now Sudan make absolutely clear. It might seem we are in a double‐bind. We need some fixed definition of genocide, yet no definition of genocide can determine its own application objectively and directly, and so application will always depend on judgment calls regarding gaps or ambiguities inevitably present in a definition of genocide. To stop here, however, leaves the current system in place, where a given authority—often the United Nations, the decisions of which regarding genocide are typically the result of geopolitical forces and diplomatic calculations rather than concern about stopping genocide itself, or a state actor with its particular agenda—determines more or less by fiat how the term “genocide” will be applied. This is even worse than the problem with genocide prevention and intervention often recognized: that they depend on bystanders with the power to prevent or intervene possessing morally good motives as well as the will to oppose genocide. Even given ethically committed bystanders, the complexities of what is and what is not genocide might continue to undermine prevention and intervention. Many people deeply ethically committed against genocide and to approaching various cases in a rational, academically responsible manner disagree on which cases are and which are not genocide.

Fortunately, there does seem to be a path out. It depends on rethinking the nature of definition as it is usually understood in reference to “genocide.” Chalk characterizes the effort to define genocide through comparative study as a process of identifying shared features that ground explanations of what causes genocide (1994, 50). While this is an important part of the process, it ignores the fact that variations and innovations in genocide forms—the features that are not shared across cases—are just as important to understanding what genocide is and how it happens. This in turn requires recognizing that genocide is not a fixed object of study. The remainder of this article does not propose an improved fixed definition of genocide or method of applying the U.N. definition or another. Rather, it offers a different way of approaching the process of defining genocide that will address the metaphysical limitations of fixed definition.

Fixed definitions, that is, definitions that set as a permanent universal the concept of genocide, have two salient features. First, they function mainly to mark a line between what constitutes genocide and what does not. While this is an important function of a definition, it is not what Lemkin developed his definition to do. He began not with a concept of genocide but rather with a historical series from ancient times to contemporary of cases of group destruction. Unlike the concept of murder for unjust individual killings, no single notion joined the cases, which meant that no specific law could prevent or punish them. Lemkin’s goal was to develop such a law, which required identifying a single kind of act that could be outlawed. His concept gathered various cases of group destruction that shared key features. His definition of genocide was thus inclusive, that is, it functioned to bring into consideration events that before had not been considered in the way he wished to have them considered. Lemkin employed the concept of genocide to help people perceive as a single category of events apparently disparate historical incidents that humanity had not adequately perceived or conceptualized before him.

Since his time, efforts at defining genocide have focused on setting the border between genocide and nongenocide. The Whitaker Report, as noted above, as well as Ward Churchill’s and Israel Charny’s work are important exceptions to this trend. Churchill and Charny represent a contrapuntal strain of definitional analyses that have attempted to preserve the rigor of precise, clear definition while at the same time ensuring that the term not exclude forms and cases of mass violence and oppression that according to some theorists should be included. They both have developed typological definitions that differentiate a set of categories and degrees of genocide, from the premeditated physical annihilation of all members of a group to cultural destruction through negligent imposition of policies and activities that undermine group viability.

[15] has offered a generic, open definition of genocide meant to allow application to as broad a set of acts of mass violence as is reasonable, coupled with a detailed typology of specific forms of genocide that supports precise understanding of a large number of concrete instances (1994, 1999a). Churchill also begins with a general definition of genocide, and then makes it more precise along two key axes. First, he recognizes that next to direct killing and infliction of bodily harm, other methods for destroying a group usually captured under the terms “ethnocide” and “cultural genocide” should be included. Second, in addressing the intent issue, [22] uses degrees of murder as an analogy for degrees or types of perpetrator motivation, from clear cases of direct, explicit premeditation and full planning to negligent acts that are not necessarily intended to cause mass death but do and should have been known to (1997, 399-444).

Both approaches are valuable and offer much for prevention and intervention efforts. At the same time, to a degree both still function within the Aristotelian categorization method, as have virtually all other efforts at definition, including Lemkin’s. While Churchill and Charny attempt to develop fixed definitions that are open to the variation of what they consider genocide throughout history, they yet treat “genocide” as a fixed object to be fitted into a category within an overarching metaphysical scheme. All definitional attempts differentiate between genocide and nongenocide, but Charny and Churchill then differentiate a range of fundamental types of genocide as an essential part of the definition itself. The use of an expansive set of subcategories increases the breadth and flexibility of their definitions in the face of new forms, but such use remains a function of Aristotelian logic and thus expands the range of covered cases by multiplying the fixed categories rather than moving beyond the notion of a fixed category. The importance of this will be clear below.

After Fein’s fuzziness view of the concept of genocide, Chalk and Jonassohn offer the most important additional insight into defining genocide. They respond to the problem of how to determine what types of groups can be targets of genocide by recognizing that perpetrators determine the categorization they target. They define as genocide the annihilative targeting of a group as defined by the perpetrators. In other words, genocide is any attempt to destroy, in whole or part, any group entirely because of the victims’ membership in that group as ascribed by the perpetrators. This approach allows inclusion not only of political but also of social groups, classes, and arbitrary groups, such as “enemies of the people.”

Genocidal Evolution

Chalk and Jonassohn’s approach suggests an important general insight. The key to developing a new kind of definition is recognition that “genocide” however defined is not a fixed natural object but rather an evolving social construct. No fixed definition is adequate to capture it because inevitably new forms will emerge that do not fit that definition. The deep evolution of genocide might be seen as a process of objective differentiation followed by conceptual differentiation. Genocide is thus a real social object but not a natural category in having always existed or in being fixed and absolute. Originally, genocide might have been a part of intergroup violence and later war, simply the excess of existential conflict and struggles for scarce resources. Eventually, however, the act became separable from such contexts. The ancient genocide of Melos is a key example, where genocide was not the inevitable outcome of a broader process of conflict but a specific contingent state policy instrument. But even as genocide became its own focal point beside war and conquest, it remained intermingled with those other activities and thus difficult to capture conceptually. Perhaps the first explicit recognition takes place in On the Social Contract, where in book 1, chapter 4, makes clear that the killing of enemy individuals after defeat is not part of war or conflict and is morally evaluable as a separate act—an act he judged ethically wrong. But Rousseau did not theorize this mass killing itself; rather, he recognized it as an unnecessary addition to war.

The metaphysical gaps in a fixed definition of genocide, such as the U.N. definition, around intent, whole/part, and group are not simply issues of fuzzy category borders. They represent points of shifting from one form to another in the historical evolution of genocide as well as potential shifts in the future. The metaphysical indeterminacies of genocide are opportunities for multiplicity—it is not that the concept of “intent” fails or is hopelessly ambiguous but rather that more than one kind of intent is possible for genocide, resulting in different possible forms of genocide.20 Some have been actualized in historical practice, and others remain future possibilities.

The evolution of genocide is often expressed in typological schemes, such as Chalk and Jonassohn’s, that recognize changes in genocide from historical era to era. The limitation of previous attempts to capture this development is to assume we are looking back at a completed development, with recent genocides as the culmination of the evolution. However, while fixed definitions are useful as historical analyses that often also apply to present and future to the extent forms are reused and elements replicated, they cannot adequately anticipate what has not already been given in experience, and thus will become obsolete precisely as they are called upon to do their work of prevention. As new forms of group destruction that many reasonable people would see linked to the concept of genocide emerge, a given preexisting definition of genocide will fail to accommodate the new forms and might even prejudice some from recognizing the new events as genocide.

Fixed definitions of genocide do not only exclude future events from qualification as genocide; they exclude past events as well. Each fixed definition is, in essence, merely a placeholder for a set of past genocides (as small as one) that are designated as “genocide.” Thus, each is the stipulation of a certain tradition or canon of genocide. What is outside that stipulated canon is excluded, whether past, present, or future event.

There are two avenues of mutation for genocide. First, new forms will emerge with significant differences from earlier forms that nevertheless bear some key similarities. These might be the function of individual creativity or the result of adaptation to novel situations. Second, new forms will emerge that are intentional reversals of or changes in elements of previous forms in order to prevent application of current fixed definitions, as I claimed the Serbs produced in the 1990s. Both avenues extend a given tradition of genocide by adding new forms that transform that tradition somewhat. There cannot be a static tradition, as change is inevitable and repetition requires strong intentional efforts at imitation or fortuitous replication of circumstances.

This second avenue adds complexity to the concept of intent: if, in order to escape prevention, intervention, or prosecution, it is the intent of a group to commit some form of group destruction that specifically does not fit a mainstream definition of genocide such as the U.N. definition, then even if the intent is not to commit genocide so defined it is in fact to commit genocide as understood conceptually rather than by a restrictive definition. This second‐order intent links the new form that is not yet accepted as genocide to earlier forms that are. Viewed within this new framework, the concept of intent becomes an important driving force in the production of genocide, beyond being a mere measure of culpability. That genocides are planned and even conceived by perpetrators suggests a (quite negative) creative process that brings genocides into being. Destruction of groups can be pursued in a variety of ways, with a variety of meanings for “destruction,” many of which Lemkin tried to capture under his term “genocide” (407-8). The variable nature of genocide reflects the fact that different perpetrators have engaged and will engage in different kinds of processes of genocide, with different ways of defining target groups, different methods and forms of destruction, and different ways of connecting intentions with execution.

A deeper implication follows, suggested by the discussion of Serbian genocidal intent above. A fixed definition occasions attempts to work around it while preserving the goal of group destruction. As I stressed above, this can be deliberate. Before the concept was finally fixed by the 1948 Genocide Convention, those who feared being held responsible for genocide in terms of the definitions under U.N. consideration attempted to manipulate the terms of the definition to prevent possible application to their own activities, or they outright refused to accept the Convention. Once the U.N. definition became universally recognized, however, other ways had to be found to work around it—definitional denial and anticipatory modification. That is, ever since the concept of genocide was fixed by the 1948 Genocide Convention and particularly from the 1990s when enforcement of the Convention became a geopolitical reality, innovators have used it as a negative foil to develop means of group destruction that can be argued not to fit that concept as encapsulated in the official U.N. definition. The evolution of genocide has in a sense become consciously driven. While prior to 1948 the evolution of genocide more or less reflected social, political, cultural, and technological changes in the world—for instance, only in the modern era did one of the types of target group, “nation,” come into conceptual being—the post‐1948 evolution has been in part at least a purposeful improvisation on given prior forms.

A flexible definitional approach can adjust to both kinds of mutation. A further advantage of a flexible approach is that it accommodates context as a factor in determining what is and what is not genocide. Fixed definitions function as absolute universals without situational sensitivity. For instance, as discussed above, genocides against indigenous Americans are often denied because the killings, deaths, and social destruction marking them do not appear to be the product of a unified perpetrator group with a clear intent to commit genocide. It is argued that in the United States there were inconsistencies in Indian policies and that the destruction of Native American lives and societies was not accomplished as part of a single, deliberate plan with a single purpose but occurred piecemeal and haltingly through some government actions as well as frequent local actions by individuals and groups not officially representing the central government. When the Nazis, with their extremely systematic and ordered process of extermination in the tradition of German militarism, are taken as the standard against which to measure intent and method, the U.S. genocide(s) appear to fail the definitional test. But the cultural and political context of U.S. genocide(s) was quite different, happening in a rough, decentralized physical, social, and political setting and a political and economic culture that stressed individual initiative and broad popular participation. The method and structure of intent and execution of the Holocaust, including the organized industrial model of the later phase, reflected the authoritarian, hierarchical, and zealous nature of much of Nazi Germany at the time. Genocide(s) of Native Americans reflected the quite different U.S. social and political form and ideology. The perpetrator group was not unified in a simple way; at different times and places participants reflected different agendas and pursued sometimes semi‐independent courses of action. The U.S. genocides of indigenous groups were, in fact, “democratic” and even egalitarian, allowing a far greater level of individual initiative and agency than the Nazi structure ever could. Perpetrators were not united by a strict and uniform organizational structure but integrated in varying positions in a complex socioeconomic and political structure whose (imperfect) social glue was a racially supremacist ideology that made killing Native Americans or allowing them to die acceptable, and that considered their claims on land to be inferior to Euro‐American claims. Even the question of whether one should consider the destruction of Native Americans a series of genocides or a single genocide is metaphysically complicated, with both being true precisely because of the complex nature of intent in this case.

Historically focused typologies themselves can reflect evolutionary developments, as suggested by Chalk and Jonassohn. But we must be very careful to avoid teleological or progressive concepts here. “Evolution” as used in this article is not necessarily progressive; it is variation in time. Similar to viral mutations, new methods and forms of genocide are superior to earlier ones only in the fact that legal, political, ethical, and other defenses and prohibitions against genocide are less applicable to less familiar, novel forms. The variations are often situational and relative to what has come before. The danger of a progressive model is that it will give way to a hierarchy among historical forms, that somehow twentieth‐century genocide is more advanced and therefore worse than genocide in prior centuries. Thus, the shifts from “genocides of military conquest” such as Carthage, to “genocides of encounter” where one group enters into contact with a second group and the challenge or power differential results in genocidal destruction of the second group, to “incubatory” genocide that occurs within a given society as relations among different groups within it (sometimes new groups that have emerged or developed) deteriorate through tension or power differential to the point where one or more group(s) commit(s) genocide against one or more of the other groups, should be taken not as a progression but a varied revelation in time of potential forms. Each form remains viable in sociopolitical contexts appropriate to it.

As an aside, this recasts the debate over Holocaust uniqueness. It is true that the Holocaust was unique, but so were many other genocides. Uniqueness is a function of difference, and difference is the product of innovation. The history of genocide has been a history of the production of unique horrors.

Selected Evolutionary Steps

It will be useful at this point to consider some concrete examples of the innovative evolution of genocide. While not all of the issues discussed here require revision of the U.N. definition of genocide, previous illustrations used in the discussion of the limitations of fixed definitions of genocide have provided numerous examples of genocide elements that challenge the utility of fixed definitions. The goal for this section overlaps with those previous passages: it is to show the ways in which genocide as an actual social practice has changed over time, as new elements have been introduced in a series of acts of horrific violence. Some of the cases discussed here have also been engaged above with regard to definitional issues; here they are presented as evidence of the evolving nature of genocide.

One the most extensive cases or series of linked genocides has been the near elimination of indigenous peoples from North and South America. I have already described the democratic or egalitarian nature of the United States’ genocide(s) of its indigenous population. This mutation not only made the destruction difficult to recognize as a comprehensive whole; it also greatly increased the resilience of the genocidal process. With many different agents operating with some level of autonomy, obstacles or even pangs of conscience or political or legal challenges for some did not necessarily affect others. This multiplicity was a feature of genocide in North and South America more generally. While for example the defeat of the Nazis marked the end of the Holocaust, the American Revolution simply transferred the ongoing genocidal process in a given territory to another perpetrator state. The bulk of the genocide(s) of indigenous Americans was/were sustained for an incredible four centuries, and genocidal activity persists to this day in numerous countries.

The fragmented nature of the U.S. perpetrator group, that is, the fact that government policies and agents (leaders, military forces, and so on) combined with local and regional governments and groups as well as individuals and families to carry out an overarching process of genocide also involved a typically U.S. privatization of genocide. This technique has been important in later genocides against indigenous Americans, for instance in the Amazon basin, and belies definitions of genocide that restrict it to state‐sponsored acts. The history of genocide is simply not that simple; nongovernmental groups, including corporations and families, can have important, even determinative, roles. Fixed definitions emphasizing state actors cannot accommodate this important mutation that has reflected the continuing ascendance of capitalism. We might see shades of this privatization in British colonial genocides, though the tight government ties and monopolistic nature of British colonial corporations did not develop the innovation as fully as U.S. privatization.

The localized nature of much genocidal violence against Native Americans worked to shield the government as well. The government supported a context of impunity for violence against Native Americans and left locals to their own decision making. Thus, the central government did not always need to order or organize genocide for it to be carried out.

True to its self‐image as a land of innovation, the United States has pioneered other forms of genocide as well. Foremost has been the fostering of “proxy genocides” in recent decades. In Guatemala, for instance, the United States substantively supported a local government to pursue genocide against its indigenous population (281-84). The Indonesian government’s genocide against “communists” can be seen as part of a broader U.S.‐led anticommunist campaign throughout Southeast Asia (378-83). In different contexts, the United States has had varying degrees of responsibility for proxy genocides.

These two U.S. innovations raise important further questions regarding intentionality. Definitions usually assume a criminological concept of genocide, in which perpetrators violate moral principles or laws. Such a model particularly focuses on clear intent or premeditation. In many instances of genocide, this is obvious but impossible to prove directly due to a lack of documentary evidence. In others, the intent is a function of deeply embedded structures of domination and oppression that drive a society toward genocidal acts that have the outward form of spontaneous mass violence rather than of a criminal conspiracy. In such cases, intent should be understood as a complex implicit tendency resulting from the acts and attitudes of a broad range of members of the perpetrator group with varying roles and levels of awareness, where no explicit central plan unifies all of the activity. Genocide is more than a crime. The reality of particular genocides necessarily exceeds the limitations of a criminological concept, which tends to obscure or deemphasize the complex social, political, and cultural dimensions, forces, causes, effects, and engagements that constitute a genocide.

Another U.S. innovation—projected onto other countries in recent years—was the first use of so‐called weapons of mass destruction to perpetrate a genocide, in the Hiroshima and Nagasaki atomic bombings. Arguments of military expediency, however factually correct (and this is debated), do not change the fact that the United States perpetrated genocide (killing civilians) to save U.S. military personnel. The use of a context of “military expediency” to justify genocide, while on the one hand very old, was in fact novel in the age of civil and human rights post‐1776 and ‐1789. In earlier ages, no argument needed to be made.

The use of disease as a method of genocide against Native Americans by the Spanish, the British, and others has been well discussed by. This highly effective tool of genocide is very easily explained away as the function of natural medical processes due to the mutual encounter of previously separate groups. It takes a step of crisp logic to see the fallacy underlying all such denial. Clearly Europeans, with their history of pandemic plagues, were quite susceptible to new pathogens, and presumably plenty of unfamiliar pathogens existed among the thousands of distinct groups they encountered in the Americas. Basic logic suggests that if all this were simply a matter of blind natural illness, Europeans would have suffered just as dramatic a reduction in numbers due to diseases after contact as did indigenous Americans. Yet, Europeans were relatively untouched by disease, which appears to have been a one‐way affair. This is consistent with the use of disease as a weapon, not a random distribution of infection.

A further complexity is that disease that is not the direct product of perpetrator actions—for instance, something manufactured in a laboratory or introduced into a population through a clearly intentional mechanism (as in fact some diseases appear to have been introduced into indigenous populations)—does not fit the usual notion of causation. But there are many ways to cause destruction, and the use of available means that one does not create is no less causation. Killing a person with a gun one did not manufacture or buy, but found ready‐made in a drawer, is no less murder than with a self‐made weapon.

This analogy applies as well to the Great Famine in Ireland from 1845 to 1850, another innovation in genocide. Genocide by direct military killing was not a good option at a point where (1) previous Irish policy clearly conflicted with the rise of progressivism and post‐U.S. and ‐French Revolutionary realities, (2) the nature of British politics had shifted, and (3) colonialism by Britain had, in the face of these changes, to be justified by the kind of supposedly benevolent paternalism later advocated by  (see 1978, 9-10) rather than being a function of raw power and the right of “civilization” over “barbarity”—not to mention the problem of Ireland as a European area in the growing nineteenth‐century framework of race. Indeed, within the growing self‐aggrandizement of Britain’s political and moral superiority, the need to present to the world and themselves a viably moral image made direct genocide so close to home problematic. Genocide could only be accomplished in a morally ambiguous way, where the perpetrators could maintain a kind of plausible deniability for their own and others’ benefit. When a natural factor (potato disease) and the economic situation offered a means of destroying part of the Irish people in an expedient manner, the British government seized the opportunity. Callous adherence to problematic notions of wealth possession and control undergirded callous indifference to human suffering and death that functioned to destroy a significant part of the Irish population and disperse another significant part. The prevention of satisfaction of basic needs of the victims through a combination of an overarching (capitalist) socioeconomic structure and philosophy and active prejudice against the target group was an effective destruction method, and a further evolution of the starvation technique used against Native Americans so effectively in earlier, the same, and later periods. While the U.N. definition would apply to starvation (method c above), it would do so only with very clear intent (more difficult to prove for starvation than direct murder), and it is precisely that intent that becomes muddled when a liberal notion of negative versus positive freedom comes into play to deny governmental responsibility to satisfy subjects’ basic survival needs despite an unfair economic system that has forced the victim population into a precarious situation with respect to basic needs. “Intent” is generally only understood as what directs an action toward a result, not as inaction that fails to prevent a result. Even if the British government’s inaction in not releasing grain in Ireland to starving Irish people was intentional in the sense of not saving them, the further argument would be that the government had no responsibility to save Irish people in this way, and so cannot be seen to have had genocidal intent any more than a person refusing to give food to a homeless person who subsequently starves can be seen as guilty of causing the death of that person, even if the first person is the beneficiary of a system of competitive economic exploitation that severely disadvantaged the second person.

Major standards that clearly fit the U.N. definition involved innovations as well, though their innovations were accommodated by that definition. According to, the Holocaust represented the first purely ideological genocide and the only genocide that targeted victims globally rather than just the segment living in a given area. There are also the oft‐discussed roles of industrial technology and manufacturing processes, which compartmentalized participation in genocide. The Armenian Genocide might be considered the first purely nationalistic genocide, where the perpetrator committed the genocide to create a homogenous nation in the place of a multiethnic society of which the victim group was previously a long‐standing part. It also systematized gender violence as a feature of the degradation and torture of the victims and sexual and domestic slavery as a means of destroying the social structures of the victim group. In addition, the central planners set up a number of groups as well as taking advantage of circumstances producing “spontaneous” local violence that were not themselves centrally planned or ordered—that is, the perpetrators organized as well as extensively used, without directly controlling, pockets of cataclysmic but local “chaos.”

Often, the newness of a form becomes the basis for definitional denial of a given genocide. The Cambodian Genocide was the first “autogenocide,” that is, destruction of part of a group by some of its own members based on constructed or perceived differences. The main phase of the genocide does not appear to fit the U.N. definition, because ethnic Cambodians could not have been targeted merely for their membership in that category if those targeting were also Cambodian, unless they targeted themselves as well. Earlier Soviet and Chinese innovations in creating “arbitrary” victim groups within their own societies through political designations provide a link from ideological genocides against objectively preexisting groups, such as the Holocaust and Armenian Genocide, to largely arbitrary destruction of part of one’s own group.

It might be tempting to see the Rwandan Genocide as a fairly standard form of genocide. Often misunderstood as a “primitive” genocide accomplished by machetes wielded by thugs whipped up into a frenzy by government propaganda, the Rwandan Genocide incorporated at least two novel features. First, the killing methods were highly organized, but not through technological advance or preexisting structures, to create a speed and efficiency in killing perhaps unparalleled in history. In less than three months, at least 850,000 people were killed. Second, the perpetrators accomplished their genocide in the direct presence of U.N. peacekeepers and in full sight of the world community. Through careful manipulation of the interests, fears, realpolitik motives, and sensitivities of outside powers such as the United States and Belgium, the perpetrators thwarted intervention by the United Nations and other forces. This was not simply a function of a general failure of outside powers to follow through on protective commitments. The Rwandan perpetrators were able to tap geopolitical forces and to give themselves cover in a context in which intervention should have happened. For instance, the perpetrators apparently killed a number of Belgian peacekeepers in order to play on the fears and shallow commitment of the Belgian government and society, to get the Belgians to withdraw their peacekeeping troops from Rwanda (332). This is not to suggest that external powers bore no responsibility for the genocide, but to assert that the moral failings and self‐interests of outside powers were part of the genocidal process itself, rather than merely keeping those powers outside an independent process when they should have been involved. The perpetrators in essence drafted the outside powers who were pledged to genocide prevention and intervention to provide cover for their genocide.

While rape has perhaps always been an element in genocide, as discussed above Serbian perpetrators used rape in place of killing, rather than as one method among many of torturing and degrading the victim group. In this manner, they attempted to accomplish the destruction of the Bosnian target group in two ways. First, without as much direct killing as would otherwise have been necessary, they effected a fundamental change in the demographics of Bosnia, according to the definition of the perpetrators. The Serbian ethnic self‐concept holds that a Serb father determines the Serbianness of any child he fathers. By forcibly impregnating Bosnian women, Serbian perpetrators required them to give birth to Serbian children. This not only increased the percentage of Serbs in Bosnia directly; it also prevented Bosnian women from having Bosnian children (as defined by the perpetrators) and thus increased the relative percentage of new Serbian births in the area. Second, the rapes inflicted tremendous trauma on the victims and destabilized their communities, both in terms of the presence of traumatized individuals and because of complex and problematic views of rape victims and the impact of rape on the men of their community as a function of both sexism in that community and legitimate concern. This policy appears to have been deliberate, as a means of destroying the victim community without so much direct killing that the world community would have been likely to recognize the process as genocidal while it was happening.

There is a deeper aspect of gender violence related to genocide. While in very recent years there have been more (long overdue) acknowledgments of the role of rape in genocide, the relationship between violence against women and genocide is actually more complex than typically understood or reflected in definitions of genocide. Rape and other violence, degradation, and torture of women is not simply one more method of genocide, it is also a motivational factor, a different kind of intent that draws some perpetrators into an overall process of genocide. In the Armenian Genocide, for instance, information in eyewitness and survivor accounts suggests that many perpetrators carried out genocidal acts in order to have access to women and girls for rape and sexual slavery. Thus, the Armenian Genocide itself depended on the gender of a substantive part of the target group and a preexisting, overarching gender domination structure and devaluing of women and girls within it. But if this is true, then genocide in this case and others, while certainly being more than an act of gender violence simpliciter, also functioned as a part or episode in an overall history and sociological category of gender violence. That is, genocide in part is an element in a broader feature of human society, gender domination. In this sense, rape and enslavement of women has emerged—perhaps this is a matter of recognition as much as objective change—as genocide itself. This shift requires another redefinition of genocide. It also suggests a similar process regarding colonial genocide that addresses the question, Is colonialism inherently genocidal, or is genocide merely a contingent method sometimes employed in the process of colonization? If the latter, then, as for gender violence, genocide is part of a broader concept of colonialism. Of course, given that colonialist genocide is one form of genocide but does not exhaust the latter category, the reverse would then also be true, such that colonialism is in part an element of the broad phenomenon of genocide. Genocide overlaps with colonialism and gender violence, with genocide exceeding both of the other categories even as they exceed it, and with genocide functioning as a foundation of each of the other categories, and vice versa—depending on context and interpretative framework.

The current case of Sudan represents an innovation in the face of the fixed U.N. definition that has caused the United Nations to reject the claim that what is happening in Darfur is genocide. The mass violence in Sudan is occurring in a context with two important features. First, Sudan experienced civil war for decades, and the present violence can be represented as a further episode of internal conflict. Second, the preexisting war conditions and the presence of rebel activity in Darfur are apparent impediments to central governmental intervention in the region. The central government seems to be at once relieved of the responsibility to protect the victims and clearly not in control of the activities of various local groups, in particular the Janjaweed militias. This is not simply subterfuge—as mentioned above, it might be an accurate characterization of the situation relative to at least some of the militia groups, but that situation is intended by the central government, which could take steps but does not to mitigate the unrest and to rein in the militias. The government can claim to be at worst neglectful and so not genocidal: mass violence is being committed by others without central governmental control or even instigation. In this way, the central government is able to accomplish genocide without causing or even strictly speaking intending genocide—even while its actions and inactions strongly suggest that it does desire the erasure of the target population in Darfur as a viable community in Sudan.

Another important feature of the Sudan genocide is its relatively slow pace. If Rwanda represented a maximal velocity of killing, the Sudanese government appears to be playing off this to present its very slow‐paced but inexorable genocide as nongenocide. The government has found a frequency of killing that is below a critical point at which the pace itself would cause outsiders to perceive the killings as genocide. Because the death rate through killing, starvation, and disease is slow enough, compared especially to Rwanda, the other “big” sub‐Saharan genocide of recent years, the Darfur Genocide seems to lose its unity as an event, which suggests that there is no overarching genocidal intent behind it.

The Nanjing Massacre might be seen as similar. In that case, higher military officials were considered responsible for allowing the rampant killing and rape on the ground, but not for ordering a systematic campaign of intentional genocide. Thus, it would appear not to have been genocide. But this assumes that genocide is a top‐down affair planned and driven by those in power. The Nanjing Massacre represents a variation in which mass death and destruction results from decisions and actions by lower‐level military personnel on the ground. It is then genocide without an overarching leadership structure.

From Caveats and Possible Objections to a New Definitional Process

The rejection of a strict Aristotelian categorization scheme should not be taken as a move toward the postmodern dissolution of all approaches to categorization or definition. The “postmodern turn” properly understood and applied requires recognition that all definitional or categorization schemes are inherently unstable, but this is meaningful only in relation to stable, fixed definitions as produced historically. Regarding the topic of this article, at each historical moment and in each cultural context, there is a “true” definition of genocide that captures accurately what has occurred up to that point historically. However, actualization of unnoticed (and, prior to their actualization, often unnoticeable) potentials in past forms as well as new forms inevitably emerge. Perhaps the dynamic described here follows a pattern similar to Jean Hyppolite’s reading of Hegel’s Phenomenology of Spirit dialectic as a sequence of inherently unstable syntheses that give way to new tensions and contradictions. While the quest for a fixed definition privileges the unification trend of genocide definition, this is continually undermined by fragmentation of the category along lines of innovative difference. Thus, a stable resolution cannot be through the attempt to attain a perfect unification through definition. It can only be through recognizing the importance of the inevitable fragmentations, the automatic differences within the category, as means of anticipating new forms of genocide. Rather than a debilitating problem, the instability of any unified concept of genocide, with the ambiguous meanings of such concepts as “intent” and “group,” invites this anticipation and can be used as a positive means of gaining insights helpful for prevention.

Definitionalism as an obstacle to recognition of the changes in genocide might be seen as a kind of devolution, that is, reverse evolution of the concept of genocide. Rather than the innovative nature of genocide pushing a revision of the definition, the definition imposes a falsely regressive concept of genocide that excludes new forms and favors only specially selected previously existing types—with significant implications for prevention and intervention efforts.

A second general point regarding the evolution of genocide is that while the category “genocide” is a construct through human decision, since it was stipulated by Lemkin and has been popularized in a somewhat different form in the U.N. definition and the subsequent scholarly and public discourse about it, it has functioned as a natural category in now being a prior given for any discussions of what genocide is. It is not possible to impose an arbitrary definition of genocide on the term now, because the weight of the preexisting discourse functions as a limit on what can be assimilated to the term. Thus, scholars for instance might disagree over what the definition of genocide should be, but even those in disagreement will tend to agree about the parameters of definition and which proposed definitions count as legitimate possibilities. A restrictive definition of genocide, for example, will not be rejected outright as an impossible definition but rather will be argued against as less useful and appropriate than other definitions. This complex situation might best be understood through progressivist philosophical inquiry, for instance, as a response to Meno’s challenge to Socrates—roughly, How will you know you have found the correct definition of virtue when you test different definitions of virtue if you do not already know the correct definition of virtue ahead so you have a criterion by which to judge the definitions you encounter? A Millian progressivism, on the other hand, does not require a final stipulated truth up front but can in fact generate new truths by building beyond what is given from past intellectual activity, with the recognition that, just as past ideas can be transcended by new ideas that build on them, any new idea will itself be subject to modification in ways sometimes inconceivable from the current state of intellectual development. Thus, it is based on the past but not strictly bound by it. “Genocide” might then be seen not to be defined by timeless features but rather to be a lineage of evolutionary variations from past to future.

Given this approach, it is important to recognize that the fixing of definitions as summations of prior forms of genocide is not a futile or arbitrary process. It is crucial to identifying forms of genocide that should be prevented, intervened against, and punished, both as a means of recognizing these forms of mass destruction in the future should a perpetrator group employ one of them and in deterring or reducing the availability of particular forms of genocide. It is unlikely, for instance, that a future genocide will look just like the Holocaust, because it is so clearly recognized as a central form according to virtually every credible definition of genocide. What is more, there is no recipe for what will count as a reasonable modification of the concept of genocide. While perpetrator intent to modify the form used can be a helpful guide, it is not always clear or present, and modifications might in fact represent successful deterrence, as destructive tendencies are changed sufficiently to reduce what would have been their destructiveness. What counts as a modification of the concept of genocide can only be determined through deliberations among scholars and others.

The process of defining genocide or, more properly, the process of conceptualizing and reconceptualizing it, is part of a dynamic, with perpetration of new genocides at the other pole. As perpetrators produce new forms of genocide, reconceptualizations capture these emergent forms. Then yet other perpetrators produce new forms that transcend the reconceptualizations and thus require further reconceptualization. This dynamic is ongoing and might suggest a perpetual process such that all gains against genocide are subject to supersession by new forms. As suggested above, however, the possible variants of genocide might be limited such that eventually no further significant mutations that adequately sidestep existing concepts of genocide will be possible. But even if this is not the case, through ethical and political volition, it might be possible to cut off the production of new forms. If the broad goal of genocide prevention includes an end to genocides as we know them now, it must also include an end to the mutation of genocide.

While no definition of genocide can anticipate future variations in specific detail, a good definition of genocide must be open enough not only to accommodate the full range of past events of genocide but also to anticipate variations. Thus, certain definitional insights, such as recognizing that the target group can be any defined by the perpetrator, can capture some future variations without requiring modification of the given definition. Even the ambiguity of “intent” in the U.N. definition, if treated in the opposite manner from how it is now—not as a restrictive criterion based on a narrow view of “intent,” but rather allowing the multiple meanings of “intent” to fit the U.N. definition to a range of cases—offers a point of openness that can accommodate future forms of genocide not conceivable today. Definitions can also include speculations about future genocide mutations. These might project out from current trends as well as anticipate how future perpetrators might respond to new legal cases against genocide and new kinds of political, economic, and social contexts.

While more experience with conceptualizing genocide should mean a fuller identification of the range of potential trends implicit in the concept, even the best methods of speculation will leave much unanticipated, and the problem is compounded by the fact that once a speculation is publicized, sophisticated would‐be perpetrators will attempt to work around the speculations in the same ways they might around features of the U.N. definition. Thus, a useful flexible definition of genocide must have built into it recognition of the possibility of future modification—a deliberative process as a means of keeping the concept of genocide up with the evolution of mass destruction. Recognition of past modifications (for instance, the change from instrumental genocide to ideological genocide) can serve as a guide for the kinds of links to new forms that preserve the concept of genocide. The definition can point to specific kinds of modifications and points of modification (the nature of intent, and so forth) that are likely in the future. Where an evolving or flexible definition is official, the specific academic and political process of modification of the definition can be laid out in advance.

One methodological strategy might be to follow Ross’s approach to Kantian categorical imperatives in ethical decision making. Ross accepted that Kant’s general ethical rules were prima facie correct, but also that circumstances and other factors might make suspension or modification of a given rule in a given context morally correct. Similarly, the given definition of genocide—generally, the U.N. definition—can be taken as appropriate prima facie, while the moral correctness of using a modified definition can be recognized for some cases of group destruction, when sufficient reasons support doing so, such as the perpetrators’ intent to modify the form of violence in order to avoid on technicalities the letter of the definition of genocide, the similarity of a new form of group destruction to certain past forms, the destructiveness of a novel kind of act, and so forth.

[70] claims that some international courts have already taken something like this approach to the U.N. definition. He assumes that the problems with the U.N. definition are not fundamental and argues that international law since 1948 has “evolved” to cover all the “gaps” in the U.N. formulation of the concept of genocide. To the extent Schabas is correct, this could be a promising development: some courts have been flexible in their interpretations of the U.N. definition of genocide. The International Criminal Tribunal for the former Yugoslavia (ICTY) Appeals Chamber’s statement he presents shows good reasoning regarding the “in part” aspect of the U.N. definition. Schabas’s case overall, however, is shaky. In claiming that the problem of non‐state‐sponsored genocide has been addressed, for example, he offers an unconvincing argument. He points out that the ICTY has stressed that a plan or policy of genocide is typically an important factor in determining that genocide has occurred. He then says that nothing in the U.N. definition restricts the plan or policy from being that of an individual rather than a state government, so if an individual altogether independently of any government has a plan or policy, he or she could potentially be tried for genocide. On the one hand, this is along the lines of the kind of identification of open points of a fixed definition of genocide I argue for. On the other hand, the claim that international courts will interpret such points openly rather than trying to close them off is purely conjectural and not based on an actual legal trend: that something could possibly be interpreted a certain way does not mean it will be. In addition, in discussing the problem of what kinds of groups can be targeted for genocide, Schabas emphasizes the statement by the International Criminal Tribunal for Rwanda (ICTR) that the spirit of the U.N. definition includes any “stable and permanent group,” not just ethnicities, races, nations, and religious groups (2006, 99). Such groups are populated by members “who belong to it automatically, by birth, in a continuous and often irremediable fashion” (2006, 99). While this would apply, for instance, to gender groups, at least as usually understood, it is not a sufficient extension of the U.N. definition to cover other important targets of genocide that are not permanent groups, such as many politically designated groups. Indeed, the concept of group used here might even be a step backwards, as it would seem to open up new problems for religious groups that allow conversion or have noticeable apostasy, nations that allow assimilation of new members, and so forth. It could possibly apply to some socioeconomic class systems, such as a rigid caste system, but not to a system with class mobility and changing class structures. Regarding sexuality, the concept of group presented would only apply if a strict biological determinism of sexuality were assumed. Beyond these particular issues, Schabas’s general approach emphasizes a few exceptional points of legal interpretation and ignores the overall emphasis by the ICTR on restrictive interpretations of the U.N. definition around intent and other elements. The broad tendency in international politics and law continues to be restrictive and at times definitionalist.

Even if this were not the case, however, Schabas’s approach is problematic. The main difference between Schabas and those proposing new definitions is that he does not believe a new definition is necessary, because legal interpretations of the U.N. definition in essence have already revised that definition. He appears to maintain the assumption that genocide itself is a fixed form of act and that it is just a matter of developing international law to capture it adequately. If this is so, he fails to recognize the evolving nature of genocide itself and holds the view that a good interpretation of the right fixed definition will converge with reality and ultimately settle all issues for the future. There is a potential danger here, in fact. If the view becomes prevalent that legal interpretations of the U.N. definition have adequately addressed all limitations of the definition, then successful definitional denial and anticipatory modification will become easier, because the manipulable boundaries of the U.N. definition as legally interpreted will have a legitimacy they do not have at present.

In understanding the legitimacy of and need for an evolutionary approach to the definition of genocide, it could help to consider what the definition of genocide would have been had it been formulated prior to modifications that are generally recognized as covered by the concept. For instance, if “genocide” had been defined five hundred years ago, the term might have been tied to such things as (1) destruction of an external enemy (by religion, state, and so on) and (2) instrumental motivation (the concept of ideology not being available prior to the eighteenth century at least). In hindsight, the appropriateness of modifications that allow inclusion of such cases as the Holocaust is clear, and the lesson for future modification should be as well. The definition Lemkin would have developed in the 920s or 930s C.E. would have been quite different from the one he actually developed in the 1920s and 1930s. One could argue that it took modern crystalizations of genocide, such as the Armenian Genocide, to present such clear models that Lemkin could look back historically and see other cases that would not have been so clear, and therefore genocide as a concept only became possible in the twentieth century. Yet, cases such as Melos and Carthage appear to be clear in their own right. Thus, were there a Lemkin a millennium ago, he or she would have come up with a different definition of genocide, based on the cases available at the time. Given the cases we are now aware of, it would have clearly been incorrect to present such a definition as timeless and universal. It is just as inaccurate to characterize any definition produced in the twentieth century as timeless and universal.

This is not meant to suggest that future genocides will not be like past forms in any way. Certainly, elements will be repeated in any future genocide, and some future genocides will enact prior forms. Indeed, in some cases the patterning of present acts of genocide on past is through concrete connection as well as intentional imitation. There were points of continuity between the German genocide of the Herero and the Holocaust, as well as between the Armenian Genocide and Nazi mass violence, including the Holocaust. In the latter case, not only was Hitler once quoted as assuaging concerns about the consequences of brutal treatment of civilians in the invasion of Poland by saying, “Who after all speaks today of the extermination of the Armenians?”, Dadrian discusses a number of prominent Nazis who served in the Ottoman Empire during the Armenian Genocide (1996, 199-203). Further, in carrying out the Armenian Genocide, Turkish military and police personnel sometimes studied prior instances of mass violence such as the Inquisition in order to learn torture techniques. These lines of influence bring the intentionality of later instances into stronger relief and reveal just how fully they are inside the history of genocide. The problem for a fixed definition is that even where similarities and lineages are clear, new forms are often different enough no longer to qualify under a given fixed definition. In fact, despite basing actions on past models, clever contemporary would‐be perpetrators can intentionally subtract just enough to problematize them as genocide according to a given fixed definition.

The risk of a flexible concept of genocide stretching the term into meaninglessness is overstated. The concept “genocide” does not exist in a vacuum. While many treatments of its proper definition are concerned with what is inside its boundaries, a holistic approach to genocide recognizes it to be in part determined by other categories of mass and individual violence. Thus, genocide in some sense is not slavery, is not isolated massacre, is not genuinely unintentional mass death, and so forth—though it may include slavery, massacre, and so on. Genocide borders on such other categories, so whether an event is genocide does not depend on a decision as to whether the event is something meaningful or not but depends rather on what kind of event it is, from a set of distinct human rights violations as well as inadvertent mass deaths. The question is not, Is this particular event a genocide?—asked in a vacuum—but rather, Is this event best understood as genocide, an isolated massacre, spontaneous multiple homicide, a terrorist act, slavery, or something else? While one can argue that many events that fit in these other categories could fit in a flexibly defined evolving category of genocide, if an event fits better in another category, then it should not be considered genocide. This also allows that mutations might carry genocide into one of these nongenocidal forms, and thus a new event would not be genocide even if it has developed out of the previous tradition of genocide. Of course, the borders between different pairs of categories of violence and oppression are not rigid, and so evolutions of categories in addition to genocide must be part of the process of understanding and labeling particular events and processes of mass violence and oppression. What is more, an event or process could legitimately fit into more than one category, which would in fact produce a hybrid category, such as “genocidal mass rape.” In such a case, the “best fit” will be in two or more categories, not just one. This will be not an overstretching of the category of genocide but rather recognition that some events of genocide sit on the border of the category with other categories.

Beyond this consideration, it is important to understand that there are different axes along which complexity can be added to the concept of genocide. The usual assumption is that adding new forms of genocide would expand the term beyond its given borders. While in some cases new forms do expand the borders of the prior definitional standard, this is not always the case. New forms are also added through increased complexity within the bounds of the given definition. For instance, it might be recognized that within the apparently simple notion of “ethnic group” there is actually a range of forms of ethnicity that can be specified individually. The old borders remain intact, but inside them the definitional terms have gained greater precision and differentiation. With a very loose analogy to positive and negative infinity in mind, while some points in this article support an expansion of the current usage of the term “genocide,” many can be interpreted as intensifying the concept (making it more internally complex) without expanding its borders relative to other categories of destructive or oppressive acts and processes.

The use of a flexible, evolving definition might be seen as a hindrance to legal prosecutions of perpetrators after the fact. Yet, lack of a fixed definition of the crimes in question did not prevent prosecution of the Holocaust. While it is true that the major Holocaust perpetrators tried at Nuremberg were not convicted of genocide, it is also the case that they were tried for crimes against humanity based on law put into place after World War II had ended, that is, enacted retroactively. An important basis was that prevailing international norms of ethics that should have been clear to the perpetrators showed their acts to have been egregiously immoral. In addition, Alfred makes a strong case that the 1948 U.N. Genocide Convention can be applied to cases prior to it, specifically to the Armenian Genocide of 1915. His main argument is that the Convention does not establish a new crime per se but rather updates previous international law, such as the Hague Conventions. What I am proposing for genocide is less of a step. Genocide will be outlawed already, but it will be recognized that a political or court process might be necessary to determine in a particular case if the existing definition of genocide is adequate for the current case or needs to be amended. For instance, a new concept or form of “intent” might be recognized or the intent element might be suspended.


I have argued in this article that approaching genocides of the past, present, and future through a fixed definition prevents an adequate understanding of some established and some emerging forms. Recognizing genocide as a changing social practice through a flexible definitional process that can capture mutations can advance the goals of a proper understanding of past genocides, intervention against present genocides, and prevention of future possible genocides better than the use of a fixed definition.

My argument has developed three possible approaches to defining genocide flexibly. The first method would be the Rossian, to begin with a standard definition that could then be modified in particular cases as needed. This approach is inadequate because it preserves the same basic definition of genocide and relegates genuine evolutionary steps to situational exceptions. The second is to change the base definition with each accepted situational modification. The need for timely adaptation to new forms requires an ongoing, proactive process of definitional revision. Waiting until after a genocide has occurred to determine whether or not it was genocide—as was done in the case of Rwanda—means the failure of the definitional process. Each event that can possibly be considered genocide should occasion a deliberative process to determine not just whether it fits the given definition but also whether that definition should be modified in the face of the new event. Events and definitions should be held in a dialectical dynamic whereby the definition is applied to new events for decision making while at the same time events potentially drive changes in the definition.

The third approach is to operate with an intentionally vague general definition of genocide, such as “Genocide is the attempted or accomplished destruction of a group.” Possible events of genocide past and present can be tested against this definition. Each determination of what is or is not genocide depends on a specification of that definition, that is, on an interpretation that identifies certain concrete determinations (for instance, targeting what part of a group is necessary for a charge of genocide). The definition will already have some features determined by past cases accepted as genocide, but through deliberation new cases can be added to the canon of genocide cases as appropriate, resulting in an evolving definition.