Human interaction networks reveal that large cities facilitate segregation

A long-standing expectation is that large, dense, and cosmopolitan areas will support diverse interactions and socioeconomic mixing. It has been difficult to assess this hypothesis because past approaches to measuring socioeconomic mixing have relied on static residential housing data rather than real-life interactions among people meeting at work, in places of leisure, and in home neighborhoods. Here we develop a new measure of interaction segegation (IS) that captures the economic diversity of the set of people that a given person meets in their everyday life. Leveraging cell phone mobility data to represent 1.6 billion interactions among 9.6 million people in the United States, we measure interaction segregation across 382 Metropolitan Statistical Areas (MSAs) and 2829 counties. When averaged across all MSAs, interaction segregation is 38% lower than a conventional static estimate, which means that people meet diverse others mostly when outside their home neighborhoods. But, we also find that interaction segregation is 67% higher in the 10 largest Metropolitan Statistical Areas (MSAs) than in small MSAs with fewer than 100,000 residents. We find evidence that because large cities can offer a greater choice of differentiated spaces targeted to specific socioeconomic groups, they end up promoting -- rather than reducing -- everyday economic segregation. We also discover that this segregation-increasing effect is countered when hubs of interaction (e.g. shopping malls) are positioned to bridge diverse neighborhoods and thus attract people of all socioeconomic statuses. Overall, our findings challenge a long-standing conjecture in human geography and urban design, and highlight how built environment can both prevent and facilitate diverse human interactions.