Empiricism for descriptive social network models

Social complex system modeling is at the core of a recent interdisciplinary effort, in which statistical physics has been playing a notable role; yet, the unfrequent validation of behavioral hypotheses possibly leads to normative rather than descriptive models, plausibly less appealing to social scientists. An epistemological insight and hindsight on this whole program is proposed, emphasizing a strong empirical methodology which extends to low-level agent-based dynamics. Special attention is given to the recent interest in knowledge diffusion models.