Humans exhibit a strong tendency to associate with those similar to them. This tendency, termed homophily in the social sciences, impacts both the structure of society and its outcomes. In this talk, Nicole Immorlica discusses the mathematics of homophily. She first quantifies its theoretical implications for geographic segregation. We will see that even tolerant societies exhibit segregation, as weak local preferences can have ripple effects with global consequences. This geographic segregation, in turn, reinforces homophily. Immorlica then explores the role of homophily in inequality and economic productivity. She starts from the observation that most employment is driven by referrals. As social networks tend to exhibit homophily, these referrals are also homophilous, perpetuating existing inequality. This inequality, in turn, is harmful to productivity, causing inefficiencies in the economy. She concludes by suggesting policy interventions that can both reduce inequality and improve productivity.  Finally, should time permit, she explores a rational basis for homophily in social networks, showing that people prefer to listen to those similar to them because they get more accurate information from such interactions.