Risk models of atrocity crimes address the question of where outbreaks may take place and focus on slow-moving structural risk factors, such as regime type and level of economic development. In contrast, early warning models of atrocity crimes address the question of when atrocities may break out and focus on proximate causes in terms of fast-moving factors or triggers. Risk models and early warning models are bricks in the international atrocity crime prevention architecture. This chapter proposes an additional brick to that architecture. Sharing basic features with approaches and concepts within epidemiology and early detection mechanisms for disease outbreaks, this chapter outlines an atrocity crime disease surveillance oriented statistical approach to the early detection of systematic atrocity crimes, and illustrates it through an application on an emerging genocide: Darfur in 2003-04.

The early detection approach outlined in this chapter can serve two functions. First, and as the name suggests, it is forward-looking or prospective in that it can identify systematic atrocity crimes in their early phases, and thus constitutes de facto early warning for genocides but also systematic atrocity crimes that are not genocidal in intent. Second, it may also be applied in a backward-looking or retrospective manner in terms of forensic analysis of past civilian fatalities to establish the extent to which the fatalities were not just intentional but also systematic, and thus whether an atrocity crime has taken place.