We investigate the impact of preferences for co-ethnic contact on residential segregation and find support for the Schelling hypothesis that modest preferences can have significant consequences for segregation under certain conditions. Our findings temper and in some instances contradict Laurie and Jaggi's claim that expanding 'vision', the size of the immediate neighbourhood households consider when evaluating residential ethnic mix, makes stable integration a likely outcome in Schelling-like models with weak-to-moderate preferences. We note several reasons why our results differ from Laurie and Jaggi's. The most important of these is that Laurie and Jaggi underestimate the segregation-producing potential of weak-to-moderate preferences because they overlook a powerful interaction between preferences and ethnic demography and perform their simulations using the optimal ethnic mix for achieving integration. We show that the preferences Laurie and Jaggi describe as compatible with stable integration generate high levels of segregation in their model under settings for ethnic demography common in real cities.