A systems model of how ethnic tensions flare into violence has passed a test in Switzerland, where harmony prevails except for one region flagged by the analysis.

The model runs census data through an assembly line of high-powered mathematical processes, but at its root is one basic assumption: that community-level violence is primarily a function of geography, modulated by the overlap of political, topographical and ethnic borders.

Though tests of the model, developed by network theorists at the New England Complex Systems Institute, are still in early stages — Switzerland is the third country to be analyzed — they raise the alluring possibility that propensities to social violence can be formally quantified, predicted and even prevented.

‘Switzerland could have been Northern Ireland, except they made cantons’

“When people think about peace and conflict, they ask, ‘What is the history of each group? The social structures, the economics, the governance?’” said NECSI president Yaneer Bar-Yam. “This is a model that takes the geographical size of groups, and comes up with violence.”

According to Bar-Yam’s team, between-group violence is unlikely when one of two conditions are met: Either diverse communities are so well-integrated as to prevent any one group from dominating, or — lacking such integration — when political or geographic boundaries match demographic borders.

When there’s a mismatch, tendencies to persecution have opportunity to flourish. Majority groups have the power to define local rules, but minority groups are large enough for conflicts to arise. With the right borders, this wouldn’t happen. “Violence arises due to the structure of boundaries between groups rather than as a result of inherent conflicts between the groups themselves,” write Bar-Yam’s team in their Switzerland study, released Oct. 6 in arXiv.

From top to bottom, Switzerland's demography as demarcated by language; with topography added; the propensity to violence, as calculated with the inclusion of political boundaries but without topographical boundaries; and the propensity to violence, calculated with the inclusion of topographical boundaries. Image: Rutherford et al./arXiv

The claim sounds suspiciously reductionist, with complicated social realities subjugated to a few neat parameters. But as demonstrated in the 2007 Science paper where NECSI debuted the model, its calculated local probabilities of violence closely matched the locations of real-world violence in Yugoslavia and India.

In the new study, they test the model with data from Switzerland, a country known for social stability and prosperity, but containing a potentially volatile cultural mix of different languages and religions — French and Italian and German, Catholic and Protestant.

The researchers mapped census information onto a spatial grid, then categorized the patchwork of ethnicities with refined versions of equations typically used to define edges and borders in artificial vision applications. That produced a baseline map of violence risks, which was then cross-referenced with maps of Switzerland’s cantons, or quasi-autonomous member states, and maps of mountains and lakes.

Heightened risks of violence occasioned by religious differences were smoothed out by the inclusion of borders. “Switzerland could have been Northern Ireland, except they made canton,” said Bar-Yam. For linguistic differences, heightened risks of violence persisted only in the northwest, where the Jura mountains form a porous boundary between historically French and German-speaking communities.

As predicted by the model, this is indeed a region where violence erupted in the 1970s.

Bar-Yam hopes the model can be tested in other regions, and perhaps even used to inform diplomacy and political reforms. “Violence doesn’t just come out of people being bad,” he said. “The propensity to violence is not that different between Switzerland and Yugoslavia.”

Clockwise from top left: Protestant (blue) and Catholic (yellow) demographics in Switzerland; a map of the country's cantons; risk of violence as calculated without accounting for canton borders; and risk of violence with administrative borders included. Image: Rutherford et al/arXiv

Image: Aldrin Muya/Flickr

See Also: