We study the networked behavior of individuals and groups using mathematical modeling, data analysis, and lab experiments, to build a theory of human networked interaction, capable of explaining and predicting the global outcomes observable in a population. We consider scenarios in which interconnected individuals interact in a strategic or non-strategic way (social computation, matching markets, influence, crowdsourcing) and model them with synthetic populations of homogeneous agents who make decisions according to simple rules of local interaction. The resulting population is a descriptive model of the original one, can be analyzed with tools from algorithms analysis, game theory and statistics, and allows to make predictions about the original population, under a variety of hypothesis.