This paper takes up synthetic populations as a way to discuss the ethics of uncertainty in data-driven urban processes. Urban-tech tells us that better data is a replacement for more robust democracy; that urban issues are solved through more computation, not more deliberation; and that data can increasingly substitute for political representation. In the face of these ever louder claims for calculable urban futures we examine the logics underlying one of these urban models conceptually and methodologically. Synthetic populations describe a fictitious but statistically representative urban populace. Materially, a synthetic population is a dataset. A dataset comprised of individual-level statistics, (think of age, household income, number of children) which were calculated from aggregate data, for example, the number of 35-40 year olds per census tract. Thus, population synthesis is a method that aims to generate granular data where it didn’t exist previously, a way of estimating specificity. These fictional populations are mobilized in the form of computational depictions of human behavior and decisions in ‘agent-based models’. Synthetic urban residents are used to determine the impacts of transportation systems; to map the spread of disease; and to draw segregation over time; to inform urban policy. This paper presents ongoing research using GIS-based methods to investigate whether synthetic populations evenly represent the cities and citizens they claim to describe. It asks whether mathematical sophistication here obscures an underlying uncertainty, and, in turn, speculates on the stakes of this uncertainty for the creation of a just city.