Choosing Victims: Information Problems in Genocide

• In Event: Civilian Victimization and Mass Atrocities

Abstract

How does the availability of information influence leaders' behavior/decisions when attempting to commit genocide? In the civil wars literature, information about the identity of non-combatants is understood as being a key component of a belligerent group’s ability to conduct selective, coercive violence. But because genocide is indiscriminate and not coercive in nature, we might expect that information matters little to leaders. I argue that, counterintuitively, information is a key component of genocide, and introduce two ways of understanding how information plays a role in the production of violence: first, similar to guerrilla wars, leaders must understand who is a part of their target group; second, there is a hierarchy of targeting during genocide based on available information. I test my hypotheses using archival data from international criminal court proceedings and previous ethnographic work.
The minimum threshold of information during genocide is identifiability, in particular, the ability for leaders to know directly or by proxy who the members of the targeted group are. I measure available information in three ways: prior knowledge, ethnic visibility, and proxies for group membership, such as geographic location or income level. When leaders don’t have access to this information, they have to find it or guess by proxy information, a dynamic that is similar to the ones that play out during guerilla wars. Governments can use either direct evidence, such as census data, or proxies for this information, like geographic location. At the individual perpetrator level, interpersonal knowledge is substituted with facial features or names.
These proxies produce a higher error term than direct knowledge, but they also allude to a more complex feature of genocide: a hierarchy of targeting. Genocide is a long process, occuring over a series of months and sometimes years, rather than a single isolated massacre. During that time, the levels of violence fluctuate substantially. I argue that the ease of identification drives the targeting of victims during genocide; with uncertainty sometimes protecting victims, and sometimes causing them to be targeted first. Prior relationships can also protect victims from targeting on an individual level. The assumption that identification is a priori and is automatic belies the difficulty of actually conducting a genocide; in fact, identification is a major component in the production of genocidal violence.

Author

• Sophia Lipkin, UNIVERSITY OF CHICAGO