Interdependent Urbanism

‘Simulating Interdependent Complexity: Beyond Prescriptive Zoning’
Author: John Hong
Published in Ecological Urban Architecture, Birkhäuser | 2012

If zoning at its core is about egalitarian relationships, we are at a crossroads where computation can play a major role in successful urban design. By leveraging performance based zoning standards instead of prescriptive rules, this tool demonstrates that there does not need to be an inverse relationship between density and quality of our urban spaces. Daylighting, building cores, proximity to parks, programming, and a myriad of other factors can be simultaneously evaluated to give immediate feedback to designers, planners, and stakeholders. As a ‘version 1.0,’ this tool will continue to be developed as a viable alternative to prescriptive zoning still largely being practiced today.

Toward a Democratic Process
When thinking broadly about the intention of zoning, especially in regard to urban form, one can conjecture that its purpose is to maintain a level of relational environmental quality and equality within the context of urban density. Its endgame is about democratic access to light, air, public space, and those aspects that are loosely defined as ”quality of life.” For the most part, however, our urban fabric is still governed by prescriptive means that predetermine building mass in ways that are not readily apparent, or worse yet detrimental to neighboring parcels and open spaces, while not even taking into account energy consumption. In the U.S., the most prevalent way of adjudicating urban form and program is Euclidean zoning (ironically named not after the ancient Greek mathematician, but for the town of Euclid, Ohio where the model ordinance was adopted [1]). Because this method is based on setting hard and fast rules, it is simple to implement – thus its ubiquity today. Euclidean methods have recently been criticized for their arbitrariness and lack of transparency. The question arises: If the building out-performs in terms of environmental factors, why is its figure limited to “building block” typologies administered by a set of inflexible codes? If zoning is about negotiating egalitarian relationships, is it not contradictory that we have defaulted to mainly prescriptive means to attain something that is at its core relational?

Running in parallel with conceptual shifts brought on by digital culture, ways of conceiving urban form based on optimization are now being considered on many fronts. Performance Zoning is an emerging concept that brings forth ideas of correlation rather than fixed rules.[2] Influenced by performance-based building codes that are gaining momentum in the realm of environmental stewardship, Performance Zoning criteria are goal-oriented rather than specific. Although praised for its potential to add a high level of accountability to environmental principles, it has not gained much momentum because of its reliance on supervising authorities that must arbitrate proposals through a step-by-step review to determine whether complex criteria have been met. Moreover, the definitions of many performance standards, although clearer at the architectural scale, still remain undefined at the level of urban design. Borrowing from the ”new humanist” attitudes of the 1970s with a mixture of current-day science, notions of integrated urban zoning still lack definition both in criteria and process.

It is here where computation can play a major role. Could we be at an important crossroads, a place in history where complex interdependencies can now be transparently and accurately designed for and represented? This is the core of this contribution – not to prove that any one set of rules should be adopted for their urban performance – but to open up the process of how these factors can be analyzed and resolved simultaneously through a set of parametric urban design tools. This is not to say that the inputs for these tools are wholly unbiased: all data must be validated in any situation. However, how these criteria are interrelated becomes a transparent and decisive methodology, driven by an intentional negotiation of different (and sometimes contradictory) parameters. In fact, achieving democratic results in urban form is a constant confrontation between many factors. Chantal Mouffe elucidates this paradoxical condition: if the agonistic mode of differing forces becomes resolved and fixed, the democratic condition comes to an end.[3] In the same way, this contribution hopes to make an important step away from prescriptive urban form to one that foregrounds negotiation and evolution.



A conversation with Samim Winger and Roelof Pieters of Ethical Machines about the role of machine intelligence and creative AI in urban design.

Investigating Simultaneity: Methodological Considerations
A brief word about the tools used to implement the research: For pedagogical accessibility, Rhinoceros and its Grasshopper plug-in was primarily used in conjunction with the programming language Python. As this is the beginning of a new thinking process where complex criteria can be simultaneously considered, the goal was to create an open-ended framework for students to expand the research in the future. Conceived of as a virtual ‘container,’ individual modules, each controlling a specific design criteria, can parametrically integrate with a myriad of other modules, which functions are limited only by the imagination (and current computation power). An important future goal is to include more cross-disciplinary input from outside the design disciplines including concepts from the social sciences. Therefore, software that could have been more effective in terms of processing power but required specialized know-how was rejected early on for a platform more conducive to collaborative modes of working.

Although there are many factors that can be put into this parametric structure, we focused on perhaps classic drivers of prescriptive zoning, such as solar exposure, daylighting, open space, density, and program, as the exact chemistry between these interrelated factors can be quite complex. We then looked at the way in which these parameters can be simultaneously manipulated, so that the most effective solution to be found is a result of the complex interdependency between all of them.

One of the core questions was the definition of ”optimization.” Optimized for whom? As invisible as the hand of the designer may seem in current manifestations of digital culture, the actual methods of data control (as opposed to the direct manipulation of form) is still a mode of inflection that can infer vastly different social ideologies in the final outcome. What is more, discussion of optimization in the design world treads dangerously too close to the idea of the “lowest common denominator.” We would rather use the term ”minimal solution,” as we envision our tool empowering designers to seek conceptually driven solutions where the resulting urban form effectively resolves multiple related issues rather than allowing optimization in and of itself be a surrogate driver for design.

Although the description seems to follow a step-by-step structure, it is merely the limits of the written word that connotes linearity. Our parametric organization allows one to enter into the manipulation of urban form at any point – there is, in effect, no ”correct” beginning or ending place to investigate a design problem. This is a distinguishing paradigm shift from traditional design modes, where decisions many times follow a linear additive process. Instead of the redundancy of trying to repair the consequences of an earlier choice that works for one logic but not for another, the purpose of our platform is to examine many conditions simultaneously, so that there is a constant feedback from one criterion to the next. Paradoxically enough, it is the power of computation that allows non-linear inductive thinking to prevail over deductive methods.

We can therefore say we “started” with an urban grid.  Although the grid in and of itself may not be the ideal structure for urban development (that is a whole separate discussion beyond this contribution), it is one of the ubiquitous armatures of contemporary urban organization and will be for years to come. Acknowledging the grid allowed us to compare one-to-one the differences and advantages of performance-based urbanism over existing prescriptive solutions. As the ability to accurately map open spaces in relation to both daylight and building mass is a powerful form driver within the urban grid, one of our most significant components calculates solar exposure. It dynamically represents direct sunlight to spaces on the ground (or on any horizontal plane, including elevated ones) according to the annual and diurnal solar position.[4] Building from previous research, this Solar Fan can be used to assure a programmatic relationship between daylight and public open space  – for instance, a schoolyard may require sunlight during very specific times of the day and year, while a semi-public space on a building roof might have entirely different criteria. For our exercise we distributed a series of local parks and ensured that they receive sunlight at different times of day throughout the year, with the prerequisite that the precinct would have at least one open space at any given time during the day with direct solar access.

Rather than fix building heights through prescriptive means, heights can become unlimited but sliced by the solar access to parks, streets, and other buildings.

The open space of the street also factors prominently into the geometry of building envelopes. Typically called “sky exposure planes,”[5] they are prescriptive geometries that control building setbacks in regard to building height. However, adherence to a fixed diagram neither assures a democratic access to the sky, nor does it take advantage of potential density. To address this variable condition we built a dynamic sky exposure generator, Sky Slice that deploys the use of Galápagos from within Grasshopper (an algorhythmic component literally named after Darwinian theories of evolutionary refinement)[6]. Beyond providing the guaranteed minimum sky exposure of prescriptive methods, Sky Slice takes into account varying massings at the localized level, resulting in increased density and, at the same time, greater sky exposure for the public space of the street. It is interesting to note that generating potential density in this way can create a Floor Area Ratio (FAR) of 17 where prescriptive methods with the performance criteria currently used in New York result in an FAR of 12.

While prescriptive zoning creates a constant building setback for sky exposure, a responsive sky exposure that negotiates context can create more democratic access to skylight while maximizing the building envelope.
While prescriptive zoning creates a constant building setback for sky exposure, a responsive sky exposure that negotiates context can create more democratic access to skylight while maximizing the building envelope.

As both the above tools pertain to the generation of maximum envelopes we can now discuss the idea of the voxel (short for volumetric pixel) as a way to subdivide this overall mass and imbue it with qualitative data in the form of both inputs and outputs. Designers have the freedom to assign any number of parameters to the voxels, limited only by computation power. For our test case we included such factors as minimum daylight factor, views, circulation, and proximity to open space. From these inputs, qualitative outputs, or ”readings,” of data are produced. The voxels thereby becomes an interconnected mesh, as data output from one voxel can be fed into the input of another, allowing interdependencies to ripple through the entire model.

It is interesting to note that the idea of the three-dimensional pixel has consistently been in the background of aggregated urban form throughout the Modernist era. Perhaps the clearest examples are the Habitat housing projects by Moshe Safdie or the Metabolist Nakagin Capsule Tower in Japan by Kisho Kurokawa. It is important to make a distinction here however: where these projects were important precedents in regard to urban aggregation, the voxel we are referencing here is first and foremost considered an informational module before a spatial one.

The grid is one of the most ubiquitous armatures of contemporary urbanization. Acknowledging it as a starting point allows to compare one-to-one the difference between performance-based zoning over prescriptive solutions.


By distributing local parks throughout the city, the accessibility to open space can be increased while at the same time a performance-based massing can harbor higher densities.


Computing for both density and daylight allows increased building mass and better access to daylight for public spaces over traditional zoning regulations.


The computation process relies on the ‘voxel,’ a 3-dimensional pixel.


A parametric sky exposure plane (left) allows higher building mass but also a better version of the sky exposure plane where the red shows areas of the buildings that block access to the sky (right).

Investigating Simultaneity: Daylighting and Lot-Lines
Daylighting criteria was applied to the voxels by embedding a customized version of the Rhinoceros-based tool DIVA, developed at the Harvard Graduate School of Design under the direction of Professor Christoph Reinhart.[7] Applying LEED baseline standards, our tool begins to automatically adjust floor-to-floor heights along the entire urban mass to optimize daylighting conditions. Through the willful hand of the designer, one can ”clamp” or fix any number of variables, for example maximum / minimum floor heights or certain programmatic height requirements for specific parts of the urban territory. Vice versa, the platform can run the information back into the model to begin to suggest which areas can accommodate what types of program (for example, housing has different lighting and dimensional requirements than retail).

Fine grain analysis of voxels allows custom floor heights that can optimize daylight factor.
Floor levels are adjusted interdependently in response to daylighting criteria, location of cores, and programmatic considerations.

As part of the visualization interface, voxels that do not meet the predetermined daylight criteria are noted in red, awaiting input as to whether the urban mass should be further optimized or other criteria should be simultaneously imposed. In our demonstration case, we solved for the under-daylit areas by deploying a component that generates lot-lines. (Again, this parametric can be imposed on the system at any given time, to subdivide the mass for improved daylighting , to account for economic factors of building size, or for any combination of interrelated motives.) It is interesting to observe that a parametric subdivision that negotiates multiple factors allows us to rethink the lot-line as a more integrated three-dimensional approach to property – namely a dynamically assimilated version of air rights (as seen in New York’s zoning code). Currently air-rights merely exist as a zoning by-product, to be traded opportunistically in very specific locations creating rarified results. As the increased density of our cities puts more pressure on overall environmental quality, should we not begin to rethink whether a two-dimensional understanding of property is an outdated mode of urbanism, inherited from when our cities were only several stories in height? Now, with floor area ratios easily reaching 15 or more in our densest metropolitan regions, the three-dimensional voxel has the potential to liberate the idea of property subdivision into three-dimensional boundaries that can increase the quality of our cities while maintaining density.

As an example, a new conception of lot-lines can guide the way to either subtract from or add to the overall urban mass for optimizing relationships between daylighting, passive energy strategies, proximity to core, building type, and overall relationships to public open space, to name a few. To demonstrate how different design teams might control this process in nuanced ways, we provided several scenarios as if we were wearing the hats of different stakeholders: the developer might maximize density while maintaining baseline environmental conditions defined by LEED standards, while the individual user may have an opposing strategy, prioritizing higher daylighting criteria, access to views, and access to recreational programs  – a community development group may prioritize public open space, community programs, and urban circulation. Two important related ideas are gained from this experiment: that interdependent data can be arranged into hierarchies that create very different results; and that the specific values applied to the parametric logic of the voxels allow a high level of transparency to the management of this data. The result is an empowering democratic check and balance to the way we represent, evaluate, and produce urban form.


Parametric zoning could use a 3-dimensional property lines to enhance lot subdivisions and maximize environmental quality.
The voxels can be analyzed with a limitless number of criteria. For our test case, we included factors such as max daylight factor, views, circulation, property lines, and proximity to open space.


photo: Beyond My Ken, Wikipedia Commons
Currently, air rights are merely a by-product of the zoning ordinance creating rarefied opprotunies to large-scale developers instead of privilege the formation of more democratic and dynamic urban massings.

Convergent Solutions and the Designer’s Role
Even though these test-case parameters are relatively straightforward, their precise interrelationship puts forth a myriad of complex results, each with their own social connotations. We have always imprecisely guessed that privileging urban economy and efficiency at the urban scale may have negative repercussions on public space. However, through obtaining direct visual feedback on specific optimizations, can we more fluidly negotiate a more convergent solution that can negotiate both developer-driven and community-driven paradigms? One could argue that the schematic design stage of envisioning our cities is the most important – it is at this level where ideas either take flight or are squelched. Instead of resorting to design rhetoric that merely references ambitions without specificity, one could incorporate this parametric platform as part of a public negotiation process that gives immediate and specific spatial feedback to the repercussions of our design decisions.

As the building is understood as voxels, the visualization between interdependencies between interior and exterior spaces and programs can give direct feedback to designers.
As the building is understood as voxels, the visualization between interdependencies between interior and exterior spaces and programs can give direct feedback to designers.

Finally, geometric and tectonic constraints can be incorporated as part of the computational platform. Unlike the visionary images of Hugh Ferris that showed a kind of singular tectonic for New York’s skyline a greater diversity of forms can be produced. Particularly since digital fabrication is entering into a closer relationship to full-scale building techniques, tectonic constraints can be built into the urban model. For instance, curvilinear skins that follow specific disciplinary logics versus panelized surfaces with specific subdivision rules can be quickly tested. The future potential is incorporating this information into a larger BIM model, so that costs and feasibility can be fluidly assessed. This would not just be about value engineering as a final afterthought of the design process, but would place in the hands of the architect and urban designer tools to quickly explore the convergence of environmental performance, scale, iconography, and economy.

Our computational framework is akin to a decision-support tool[8] that productively engages the increasing interdependent complexity of contemporary urbanism as integral design matter. Making a decisive split from previous modes of parametric methods that imply an erasure of the designer’s hand, our goal is to provide a platform to radically increase the efficacy of the designer’s role. Where in traditional linear processes, consequences cascade from hierarchical decisions that are made in isolation to other factors, in this new methodology decision and consequence are simultaneous – one in real-time feedback to the other. Therefore, recombinant factors are set in a field condition rather than a singular logic structure. This allows designers and stakeholders to launch exploration from diverse and collaborative starting points instead of from a fixed prescriptive path.

The subsequent refinement of this multiplicity into what we are calling a minimal solution rather than an optimized one is central. It is a shift from focusing on the lowest common denominator that many times drives optimization, to a mode of design that attempts to bring in the most effective set of hierarchies from within the overall complexity of interrelated factors. In this way, we hope to re-open democratic process as part of the ecological discussion of how our cities are envisioned. Our platform will allow multiple stakeholder points of view to be tested and debated more effectively, as the intentional control of each parametric component will be transparently represented.

Finally, with the onset of more intense environmental and sociological expectations toward the way our cities should be developed – not only in rapidly developing areas in Asia, but also throughout the Americas and Africa – prescriptive means of shaping urban form are perhaps no longer adequate to deal with this crises-level contemporary condition. Relational and performance-driven zoning will allow both flexibility and necessary checks and balances to ensure democratic environmental quality. Perhaps a recent statement preceding the New York City’s zoning text conveys the situation most succinctly: “In a certain sense, zoning is never final; it is renewed constantly in response to new ideas—and to new challenges.”[9]

Different stakeholders can privilege different data sets for results that then can be fed back into the system to continuously refine the solution.


Combined with BIM modeling, a wide variety of tectonic approaches can be explored.
[1] Village of Euclid v. Ambler Realty Co., 272 U.S. 365 (1926).
[2] “Flexible Zoning: A Status Report on Performance Zoning Standards”, Zoning News, January 1998.
[3] Mouffe, Chantal. The Democratic Paradox. New York:Verso Books, 2009.
[4] Bosselmann, Peter, Juan Flores, and Terrence O’Hare. “Sun and Light for Downtown San Francisco” in IURD Monograph No. 34. Report by Environmental Simulation Laboratory, Institute of Urban and Regional Development. Berkeley, CA: College of Environmental Design, 1983; Bosselmann, P., E. Arens, K. Dunker, and R. Wright. “Urban Form and Climate, Case Study, Toronto” in Journal of the American Planning Association, vol. 61, n. 2 (Spring 1995). pp. 226-239. Produced in collaboration with the Berkeley Environmental Simulation Laboratory; General Solar Position Calculation, adopted from NOAA website,
[5] New York City zoning ordinance, from the website
[6] Galapagos Evolutionary Solver, from the Grasshopper website
[7] Diva for Rhino is Alstan Jakubiec, Kera Lagios, Jeff Niemasz, Christoph Reinhart and Jon Sargent
J Sargent, J Niemasz, C F Reinhart, “Shaderade: Combining Rhinoceros and EnergyPlus for the design of static exterior shading devices,” (submitted Building Simulation, BS2011, Sydney) .
J Niemasz, J Sargent, C F Reinhart, “Solar Zoning and Energy in Detached Residential Dwellings,” Proceedings of SimAUD 2011, Boston, April 2011, (submitted Environment and Planning B)
K Lagios, J Niemasz, C F Reinhart, “Animated Building Performance Simulation (ABPS) Linking Rhinoceros/Grasshopper with Radiance/Daysim,” Proceedings of SimBuild 2010, New York, August 2010.
[8] Sol, Henk G., et al. Expert systems and artificial intelligence in decision support systems: proceedings of the Second Mini Euroconference, Lunteren, The Netherlands, 17–20 November 1985. Heidelberg: Springer, 1987.
[9] New York City zoning ordinance, from the website
Ecological Urban Architecture: Qualitative Approaches to Design, Thomas Schröpfer, Birkhauser, Basel, 2012
Harvard GSD Project Team:
Marshall Prado, lead researcher (MDes), Sophia Chang (March I), Jeff Niemasz (March I), Eli Allen (March I), Daekwon Park (March II), Marcela Delgado (March I)
Special consulting:
Christoph Reinhart, Associate Professor Havard GSD; Panagiotis Michalatos, Lecturer in Architecture, Harvard GSD
This Research was made possible with a grant from The Real Estate Academic Initiative at Harvard University
Video soundtrack: