This thesis explores the how computational methods can be used to enrich the history of architecture. Using Venice as a case study, it establishes a rich and multifaceted information system, made up of temporal, stylistic, photographic, architectural, and administrative information taken from diverse sources. A complete point cloud of the city, obtained from aerial photogrammetry, is segmented using tailor-made algorithms to produce 3D models of each building in the city. Further methods of segmentation and reprojection are developed to convert those 3D point cloud models into a complete set of orthophotos of Venetian façades.

This unique corpus of over 8,567 buildings and 14,167 façades --- alongside the supplementary information system --- are used as the starting point of a formalist reading of Venice. A series of experiments cross-interrogate the fabric of the city: the morphologies of footprints are mapped and analysed in relation to their date and location; the propagation of architectural motifs and typologies of façades are assessed across the city; the networks of lines of sight are reconstructed to question the visual centrality of buildings, identify strategic positions in the city; the long-mentioned homogeneity of the fabric is challenged and instances of rupture are identified. Each of these experiments is then used to contextualise and better understand the architectural solutions chosen and their relationship to topographical, cultural, and socio-political conditions.

By displacing the focal point from well-studied centres of power and high-profile architecture to the more vernacular and so-called "minor" architecture, this thesis argues that a computational outlook can open new avenues of research, shedding light on the understudied urban fabric hitherto left in the shadows, and offer original ways of looking at the city at scale. Finally, this works attempts to prove that its large-scale visual hermeneutics is at the heart of a fruitful dialogue between a long historiographical tradition and digital methods.