Smart city approaches which seek to utilise large-scale flows of urban data have become an exemplar of contemporary geomedia. In this article, I trace the migration of systems analysis and computational modelling from the RAND Institute’s pioneering nuclear war scenarios in the 1950s to their application to a broader set of social and urban problems in the 1960s. Focusing on the influential New York City-RAND Institute established in 1969, I analyse the socio-political context of data-driven approaches to urbanism and urban planning. I argue that the data-driven decision-making enabled by urban computing constitutes a critical threshold in terms of who is able to speak with authority about the city. This provides a productive lens for articulating the early history of urban computation with contemporary debates about smart cities and the ‘platformization’ of urban digital infrastructure.