First Urban Science

Urban environments are redefined as computable systems—artifacts that can be analyzed, modeled, and even designed through formal computational principles. Rejecting classical descriptive approaches, this perspective proposes a model-first methodology, where the city is understood not by analogy (organism, machine, network) but as a complex computational object governed by rules, feedback, and algorithmic structure. This shift aligns urban studies with developments in artificial intelligence, systems theory, and data-intensive sciences. Instead of treating cities as chaotic aggregates, they are conceptualized as structured rule-based systems: they are constructed from operations (e.g., zoning, regulation, design intentions) that can be represented computationally. These systems are not only legible but also generative. The emphasis is placed on the formalization of urban logic, enabling predictive simulation, multi-scale inference, and the potential integration of normative criteria into generative models. The methodological proposal centers on a computable grammar of urban form, allowing for cross-contextual generalizations without flattening urban diversity. This approach is not merely technical but philosophical: it repositions design as inference and the city as an epistemic artifact—a material expression of collective knowledge processes. Ultimately, this model-first paradigm advocates for a rethinking of urban science itself, where cities are not passively studied but actively formalized and constructed as computational realities.

Vanchinathan, A., & Bettencourt, L. M. A. (2024). ‘Cities as Computable Artifacts: Toward a Model-First Urban Science’. arXiv preprint arXiv:2406.14692.