Advancing social simulation: lessons from demography

Previous work has proposed that computational modelling of social systems is composed of two primary streams of research: systems sociology, which is focused on the generation of social theory; and social simulation, which focuses on the study of real-world social systems. Here we argue that the social simulation stream stands to benefit from recent methodological and theoretical advances in demography. Demography has long been an empirically focused discipline focused primarily on mathematical modelling; however, agentbased simulation have proven influential of late as demographers seek to link individual-level behaviours to macro-level patterns. Here we characterise this shift as a move toward system-based modelling, a paradigm in which the scientific object of interest is neither the individual nor the population, but rather the interactions between them. We first describe the four successive paradigms of demography: the period, cohort, event-history and multilevel perspectives. Then we examine how system-based modelling can assist demographers with several major challenges: overcoming complexity in social research; reducing uncertainty; and enhancing theoretical foundations. We propose that this new paradigm can enhance the broader study of populations via social simulation.