I began this book with the “L” train in Chicago, crossing the boundaries between Black and white. The train made a journey that most Chicagoans, because of the social and psychological space between the two groups – reinforced by segregation – would not make. On the “L” train, we see the great divisions in Chicago. This journey could be repeated for other cities, of course. A bird's-eye view of most of the world's cities large enough to be diverse shows social boundaries mapped onto geographic boundaries. Not all such boundaries are as extreme as Chicago's, with its hyper-segregation between Black and white, but the connections between groups and place are still present. And for most of us, these connections between groups and place are how we bringmeaning to the cities. In Boston, Southie and theNorth End are just names until we associate them with Irish and Italian, respectively. Those geographic boundaries, I argue, don't just coincide with the social boundaries – they reinforce them.

In Figures 1.1 and 1.2 on pages 10–11 (see also Figure 5.2 and Figure 5.3 on pages 132–134), I tried to convey the impact of this social geography by treating the city as a giant puzzle which could be rearranged by an experimenter to disrupt the fit between groups and space by altering the spatial continuity of group boundaries. I claimed that this would break the connection between groups and place and reduce the salience of groups, which in turn would reduce group-based bias.

Of course, in real life, we can't add or remove segregation to alter a group's spatial continuity. The only way, so far, we have been able to test the effects of segregation was through thought experiments, as in the maps, and through statistical comparisons across place (cross-sectional comparisons in the language of statistics). This runs us headlong into what statisticians call the Fundamental Problem of Causal Inference, because the same city cannot be both segregated and unsegregated at the same time. If we compare different cities, theremay be something different about those cities or about the people in those cities – something other than social geography – that is causing differences in group bias.