Modelling Ethnic Violence
A new paper* in Science purports to predict patterns of ethnic violence from simple assumptions of inter-group dynamics, ignoring all institutional and historical factors peculiar to the situation. Their assumptions are as follows:
1. Individuals preferentially move to areas where more individuals of the same type reside.
2. The size of patches of each type grows as a characteristic power of time.
3. Highly mixed groups do not engage in violence. Neither do highly separated groups. Groups with partial separation and poorly defined boundaries have the highest risk of engaging in violence.
4. There is a critical group size above which a group is powerful enough to impose cultural norms and hence create conflict. Below that size, conflict is unlikely. There is another critical group size beyond which it is difficult for smaller groups to create conflict. Hence, very large or very small groups are unlikely to engage in violence.
The probability of engaging in violence was represented by a mathematical function based on the above criteria (3 and 4). The spatial evolution of groups was modelled by assumptions 1 and 2. The model was tried on the former Yugoslavia and India. The sizes, locations and distributions of groups were assigned pixel by pixel, based on census data. The simulation results were compared with actual reports of violence. A correlation of 0.9 between the simulated locations of violence and real-life locations of violence was found (I wish they said more about how this ‘correlation’ is defined and calculated).
The authors used this correlation to infer to the truth of their assumptions, suggesting that influencing the spatial structure of groups could “address the conditions that promote violence”. As an example of where spatial rearrangements have worked such magic, they point to Singapore’s HDB racial quotas. But as we all know, Singapore’s lack of violent conflict is likely due to the stringent government control allowed by a lack of civil liberties. They also suggest that “regions of width less than 10km or greater than 100km may provide sufficient mixing or isolation to reduce the chance of violence.” They also see their results as support for their assumption that extreme separation can be a way to hold off violent conflict, and hence put forward the probably somewhat unpopular idea that “peaceful coexistence need not require complete integration.” Clearly an unwarranted conclusion, though, given that their model is sufficient to predict some outbreaks of violence, but may not be complete — the historical and institutional factors that they left out may well account for outbreaks of violence in situations of extreme separation between groups. I would be interested to see how their model fares when applied in other (geographical) regions of conflict.
*Lim M, Metzler R, Bar-Yam Y. Global Pattern Formation and Ethnic/Cultural Violence. Science, Vol. 317, No. 5844. (14 September 2007), pp. 1540-1544.
Filed under: mathematics | Leave a Comment
No Responses Yet to “Modelling Ethnic Violence”