Predictive urban infrastructures are increasingly central to the ways cities are organised and experienced. By integrating sensing technologies, data platforms, and algorithmic modelling, these systems anticipate events and adjust conditions in advance. While designed to achieve synchronisation, their operations encounter recurring points of resistance where modelled projections do not align with the temporal rhythms, spatial arrangements, or cultural practices of everyday life. This research introduces the concepts of epistemic friction and systemic misalignment to describe these conditions. The study focuses on Istanbul as a primary empirical setting, where predictive infrastructures coexist with informal adaptations and heritage-preservation constraints, drawing on qualitative sources that include planning documents, secondary literature, and observational materials. Songdo, a master-planned smart city in South Korea, and Venice, a UNESCO-protected lagoonal city in Italy, are used as analytical references. Together, these contexts provide contrasting perceptual environments that illuminate how friction and misalignment emerge under different governance and infrastructural conditions. The analysis employs a qualitative framework grounded in phenomenology of perception, infrastructural theory, and socio-technical systems analysis. Findings indicate that epistemic friction is not a temporary malfunction but a recurrent structural property of predictive governance. Systemic misalignment, in turn, reveals how computational synchronisation collides with heterogeneous temporal and spatial conditions. The paper’s primary conceptual contribution is to formalise epistemic friction and systemic misalignment as analytical tools for examining the limits of predictive governance. This research establishes the groundwork for a framework that examines how predictive systems interact with perception and governance, offering a deeper understanding of how urban life is shaped under conditions of anticipatory computation.