Disarmament Insight


Wednesday, 11 April 2007

Trying to account for mass participation in the Rwandan genocide

In a period of about 100 days, from 6 April to mid-July 1994, experts have estimated that civilian militias and members of the national army dominated by the Hutu ethnic group killed some 500’000 to 800’000 Tutsi (another group) and moderate Hutu in the African country of Rwanda. While these figures are striking indicators of the most rapidly occurring genocide of the 20th century, also deeply disturbing was the high level of participation in the killing, estimated at between 200’000 and 500’000 people.

While the Rwandan genocide will forever remain morally unfathomable, how can we begin to account for such a large-scale level of participation in the killing, in the hope that unambiguous danger signs might be identified to prevent it happening again?

In a book published in 2006 (see reference below), Ravi Bhavnani posed the question: did killing Tutsi become the norm in the Hutu ethnic group? Bhavnani argued that the level of participation in the Rwandan genocide can be explained by the “emergence of a violence-promoting norm among the Hutu community”.

In his chapter, Bhavnani noted that studies pointing to explanations such as structural factors or to Rwanda’s culture – described as one “of fear” or “of deep conformity” – fail to explain the emergence of such large-scale violent behaviour. Likewise, the significance of the death of Rwandan President Habyarimana in a plane crash near Kigali airport on 6 April as the major catalyst for the genocide is problematic.

Starting from the idea that norms are emergent properties of social systems, Bhavnani used complexity theory and simulation by agent-based modeling (ABM) to try to develop an explanation for the emergence of a violence-promoting norm among Hutu people.

The existence of a norm depends upon complex patterns of interactions between the individuals forming a society. At the time of the genocide, Rwanda, Bhavnani argued, was a “densely populated, heterogeneous society” with frequent connections and intermarriage between the two major ethnic groups – Hutu and Tutsi – leading to a complex pattern of interactions. Bhavnani added:

“When a population mixes randomly, extremists eventually have the opportunity to interact with moderates, and observe and punish their behaviour.”
In other words, in Rwanda it was about killing or being killed.

By focusing on microlevel dynamics, Bhavnani hoped this ABM could show how local interactions could bring about massive Hutu participation in the killing. Of course, however intriguing, Bhavani’s work on its own is unlikely to settle the debate over what really caused the genocide. But it serves to underline the worth of studying collective violence by using bottom-up approaches and of appreciating the complex nature of many human social interactions or, as Bhavnani put it, that:
“The causes of conflict constitute partial explanations at best, and are inextricably linked to the process by which conflict unfolds.”

Aurélia Merçay


See chapter 6: Agent-Based Models in the Study of Ethnic Norms and violence, by R. Bhavnani in Complexity in World Politics: Concepts and Methods of a New Paradigm, N.E. Harrison & J.N. Rosenau, 2006, SUNY series in Global Politic, State University of New York Press.