From Drawing Cities to Growing Them

To understand what computation adds, hold the two classic ways a city takes shape. The drawn city is the product of deliberate design by an authority: a plan conceived whole and imposed on the ground, from the Roman grid to the Renaissance ideal city to Baron Haussmann's boulevards cut through Paris to Le Corbusier's and the modernists' planned capitals like Chandigarh and Brasilia. Its strength is coherence, capacity and intent - it can provide infrastructure, order and grandeur at scale, and realise a vision. Its failure, repeatedly, is that a city drawn whole by one mind can be rigid, inhuman and blind to how people actually live: the grand plan that looks magnificent from the air and is desolate at street level, the superblock that killed the street life it replaced. The grown city is the opposite: emergent order with no single author, accreting over centuries through countless small decisions - the organic medieval town, the dense adaptive fabric of an old Indian city, the informal settlement that houses millions. Its genius is fine-grained adaptation, human scale, mixture and resilience - a complexity tuned to real life that no planner could have drawn. Its failures are chaos, inequity, and an inability to deliver large infrastructure or cope with rapid change.

Urbanism has always oscillated between these poles, and the deepest urban thinkers (Jane Jacobs against the grand plan; Christopher Alexander on pattern and organic order) have wrestled with how to get the coherence of the drawn and the humanity of the grown at once. This tension is the essential backdrop for computational urbanism, because its promise is precisely to bridge them: to let a designer set rules and intent (like the drawn city) while generating fine-grained, adaptive, responsive form (like the grown city) - to *grow* a city that is nonetheless *designed*. That is a genuinely exciting proposition. But the same history is a warning: the drawn city's characteristic failure was imposing a designer's abstract order on the messy reality of human life, and computational urbanism, if used naively, is a spectacularly powerful new way to make exactly that mistake - to impose an *algorithm's* abstract order, optimized for whatever happened to be measurable, on a living city. Holding both the promise and the warning is the whole disposition this course tries to build.

Within computational urbanism, keep two ideas distinct, because they are often blurred and they do different things (Module 0.2 goes deeper). Parametric urbanism is design through an explicit model whose behaviour is controlled by *parameters* - adjustable inputs. You build the model once - encoding relationships like 'blocks are this size, streets this wide, buildings set back this far and rise to this height, density is this' - and then you *tune* it: change the block-size parameter, or the density, or the street-grid angle, and the entire plan re-forms instantly and consistently to match. Parametric urbanism does not invent form; it lets you *explore a family* of designs you have defined, rapidly testing 'what if the blocks were bigger?' or 'what if density rose toward the transit stop?' and seeing the whole consequence at once. It is a knob-turning, what-if machine for a design space you have specified.

Generative urbanism goes a decisive step further: instead of you defining the form and tuning it, *algorithms propose the form*. Procedural rules can grow a street network the way an organism grows; optimization algorithms can search for arrangements that best meet goals you set; and increasingly, AI can generate plausible urban fabric. Generative methods *produce and explore* many candidate forms - hundreds or thousands - often surfacing configurations no human would have drawn, which the designer then evaluates and selects among. Where parametric urbanism explores a space you defined, generative urbanism helps *search a space too large to draw*. The two combine: a parametric model defines the moves, generative search explores them, and evaluation (analysis, optimization, human judgement) picks among the results. Both share the core inversion from the drawn city: you no longer place every element by hand; you specify rules, goals and parameters, and the form emerges from computation. That is powerful precisely because a city has far too many interacting parts for a human to optimise by hand - and dangerous for exactly the same reason, because the rules and goals you encode become the city, including everything you left out.

No field in urbanism is more seductive, or more prone to a specific catastrophic error, than the computational one, and an honest course names the danger on page one. The genuine value is real: cities are staggeringly complex, and computation lets us handle that complexity - modelling how a street network performs, testing thousands of massing options for sunlight and density, grounding decisions in real data, exploring far more of the possible than a hand ever could. For analysis, exploration and handling scale and complexity, generative and parametric methods are genuinely powerful and here to stay. But there is a trap at the heart of the field, and it must be stated plainly: a city is a living human, social and political system, not an optimization problem, and the things that make it worth living in are largely the things a computer cannot measure. Computation optimizes what is *measurable* - density, daylight hours, travel time, cost, a walkability score. But a great city is made of things that resist measurement: community and belonging, memory and meaning, justice and dignity, the unplanned encounter, the corner where life happens, the fine-grained mixture that no metric captures. Optimize hard for the measurable and you will quietly sacrifice the unmeasurable - producing a city that scores beautifully and is dead to live in, exactly the failure of the worst drawn cities, now automated and given a false gloss of objectivity.

Worse, the choice of what to measure and optimize *for whom* is never neutral - it is a question of power and equity. Whose comfort, whose access, whose land value does the model optimize? A computational masterplan can encode and entrench the priorities of whoever commissioned it, erase the informal city that does not fit its categories, and present deeply political choices as objective technical outputs - 'the algorithm says'. And cities are not machines you can fully model: they are complex adaptive systems full of emergence, feedback and human unpredictability that defeat any model's assumptions. So the competent stance is neither the techno-optimist's ('optimize the city') nor the romantic's ('computation has no place in the humane city'), but the disciplined urbanist's: use computation powerfully for what it is genuinely good at - analysis, exploration, handling complexity - while holding fast to the truth that the city's real life is not in the metrics, that the binding choices are human, political and democratic, and that equity and the unmeasurable must be defended precisely because the model cannot see them.

This course builds computational-urbanism literacy as a practical, honest, and critically self-aware design skill. You will start with designing cities by rule - from drawing to growing, generative vs parametric, the landscape, the hype (Module 0); then why compute urban form - the complexity of cities, data-driven urbanism, exploring the possible, the caveats (Module 1); parametric urbanism - what parametric means, urban parameters and rules, the parametric model, when it helps (Module 2); generative urban design - what generative means, procedural generation, generative design and AI, evaluating options (Module 3); the rules of urban form - street networks, blocks and density, massing and envelope, mixing uses (Module 4); optimizing the city - multi-objective optimization, sunlight/wind/comfort, walkability and access, trade-offs and Pareto (Module 5); data and analysis - urban data, space syntax and networks, simulation, from analysis to form (Module 6); in the real process - masterplanning with computation, stakeholders and participation, from model to plan, the role of the planner (Module 7); making it real - tools and workflow, computational skills, integration and scale, adoption (Module 8); reality, limits and honesty - parametric-washing, the optimization trap, cities are not machines, equity and power (Module 9); and practice and the future - the urbanist's role, getting started, India, becoming computationally literate (Module 10).

One firm boundary runs through all of it. Generative and parametric urbanism is a set of tools to explore, analyse and test urban form - it supports human and democratic decisions, it does not make them. This course teaches the principles and design judgement, and defers every binding result - actual planning and land-use decisions, statutory approvals, and the social, equity and political judgements about a city's future - to the planning authority, the democratic and participatory process, the affected communities, and the governing planning law and development-control regulations (in India, the master-plan and development-plan process, the applicable development-control regulations and the National Building Code of India). Any tool or generated form named here is illustrative and fast-moving. Studio Matrx is free and not-for-profit, and this course is written to be rigorous and honest - not a computational-design sales pitch but a clear, critical grounding in growing cities by rule, mindful of the Indian context where cities are urbanising at extraordinary speed and scale, where the informal and organic city houses hundreds of millions and rarely fits a computational model's categories, and where top-down, technocratic masterplanning has a long and cautionary history. Understand the drawn-versus-grown inheritance, the parametric-versus-generative distinction, the genuine power for analysis and exploration, and above all the optimization trap and the questions of equity and power - and you will be able to use computation to serve a humane, just city rather than to automate its erasure.