2012/ Emergence of social structure

I am involved in a research project on the notion of social structure which is based on a collaboration with a team of sociologists at Sciences Po – Medialab lead by Bruno Latour. This project starts with a basic observation: social sciences lack the tools to model the massive and increasing amount of digital data on social systems which is now available. Our ambition is to bridge the gap that exists between purely descriptive approaches of these data and quantitative models. Our hypothesis is that the gap can be explained because existing models are too simplistic and start with inadequate entities.

We propose to build original multi-agent models starting with complex agents who simplify themselves by interacting (which is a request from the sociologists based on their perception) rather than simple “atomistic” agents (which corresponds more to the physicists’ approach) to build emerging complex social structure at the “macro-level”.

⇒ See our article presenting our approach in the British Journal of Sociology (2012) .

In parallel, we are building on physicists’ opinion models (such as the well known Deffuant’s model) to investigate a question raised by sociologists: how can we explain the existence of lasting entities (institutions) from non lasting agents?

Evolution of the opinions of 100 agents with time. Turnover: agents can die or be born with a random opinion. Interaction between two agents leads to convergence to an average opinion. These panels show the emergence of groups for different interaction forces.


⇒ See our paper in Phys Rev E on this subject [arxiv version].


  • Bruno Latour, Dominique Boullier, Tommaso Venturini & Paul Girard (Medialab)
  • Pablo Jensen (IXXI, ENS Lyon)