Computational Social Science

Computational social science (CSS) is the interdisciplinary study of social phenomena through computational methods — including agent-based models, network analysis, machine learning, and large-scale data simulation. It sits at the intersection of sociology, computer science, economics, and complexity science, treating social systems as complex adaptive systems rather than as aggregates of rational actors or as structures determined by macro-level forces alone.

CSS emerged in the early 2000s as digital trace data — social media, mobile phone records, financial transactions, geolocation — became available at unprecedented scale. Traditional social science relied on surveys and small-N experiments. CSS leverages millions of data points and simulated populations to test hypotheses about diffusion, polarization, segregation, cooperation, and institutional change that were previously untestable.

The field's central tension mirrors that of complexity science generally: does a simulation that reproduces a social pattern explain it, or merely demonstrate that something could produce it? CSS practitioners increasingly argue that the standard of proof should be predictive accuracy on out-of-sample social data — a standard that most current models do not meet.