News

Complexity science as an ‘insight engine’

Development Policy25 Feb 2011Frauke de Weijer

The Santa Fe Institute in New Mexico is like a Mecca for complexity science. Physicists, biologists, cosmologists, anthropologists, playwrights and even novelists spend their time here, breaking down disciplinary boundaries. This time the focus is on foreign policy and Afghanistan in particular.

The first discussion of the workshop was on the night before the workshop proper with a discussion initiated by David Kilcullen, which exposed the real weaknesses in data analysis and solid theory for working in conflict environments like Afghanistan. As the first day of the workshop started there was therefore a real sense of anticipation, both on the part of the complexity scientists and the foreign policy specialists, on how evidence-based policy making could be strengthened using the models of complexity science.

Interestingly, one of the ideas that some of us from the development community have long had is that complexity science is best seen as a way of generating new insights and ideas to help us re-think our assumptions on what works in development. Exactly this point was made by a senior SFI faculty member, who suggested that the ideas of complexity were an ‘insight engine’ – helping us to falsify old intuitions and build new ones.

Ben Ramalingam, the author of the ODI paper ‘Exploring the science of complexity: Ideas and implications for development and humanitarian efforts’, showed this clearly in his presentation, in which he provided a number of practical examples of how network analysis had broken through prevailing thought patterns; for instance how a reduction in the use of pesticides would in fact increase the efficacy of a malaria prevention program.

These new insights are not always easy to arrive at however. The presentations by the SFI-faculty on specific applications of complexity, machine learning and game theory seemed at first glance a bit removed from aid realities, as they included primate behavior, avoiding plane collisions, and genetic variations. Nonetheless, these examples illustrated very interesting concepts such as causality networks, robustness, vulnerability and human decision making, which do have a direct relevance for foreign policy and counterinsurgency in particular. One of the most interesting examples presented was a game theory model showing a scenario with two possible stable states whereby the trajectory between these was indirect. In other words, a direct imposition of the ideal state would not lead to a stable situation, but the model showed how that state could still be reached through an roundabout and perhaps counter-intuitive way.

Fascinating to think about how such a model could be applied in the context of a more strategic introduction of institutional change into a social system.

At the same time it became quite clear that the translation from these models to the real world still requires a lot of creative thinking. Tomorrow we will continue this journey of a joint exploration of how these models can best be applied in the messy world of development and foreign policy, where data itself is quite a scarce resource.