Tag Archives: urban design

Defining a new traditional urbanism

Sometime last year this website attracted the attention of several members of the International Network for Traditional Building, Architecture and Urbanism, an organization sponsored by the Prince of Wales Foundation in order to support and renew traditions of construction. While this organization does great work to preserve the techniques of traditional building cultures, they have yet to define what the traditional urbanism of their name really implies. The importance of such a definition I believe to be primordial. If modern planning measures continue to be adopted from one country to the next, any traditional technique of construction will become irrelevant, as they have in western industrialized (and post-industrialized) countries.

This all became obvious when a miniature controversy erupted and swept through the various internet discussion groups and blogs of the community over a proposed redevelopment of the Chelsea Barracks in London’s Chelsea borough. An old modernist military installation was to be torn down and replaced by its new owner, the Emir of Qatar, with a new modernist megahousing development designed by Lord Rogers. There was nothing particularly interesting about this Rogers design, but Lord Rogers having written the plan of London, a plan that specifically calls for better design, it made sense that a Rogers design would be swiftly approved by the planning authorities. Hiring Rogers was the most risk-free option available for a multi-million pound development project.

Getting wind of this, and noticing that the Rogers design was an unremarkable piece of rehashed modern housing, the Prince of Wales hired his preferred architect Quinlan Terry to sketch up a counter-proposal that was more in harmony with the architecture of the landmark Royal Chelsea Hospital across the street from the barracks, which he then proposed to the Emir of Qatar through his personal relationship with him. The Emir, alien to the local culture and uncertain of what London considers to be “good design”, then decided to dump Rogers and re-think the development.

The Prince Charles and Quinlan Terry counter-proposal

The Prince of Wales and Quinlan Terry counter-proposal

Lord Rogers' Chelsea Barracks redevelopment proposal

Lord Rogers' Chelsea Barracks redevelopment proposal

I am not going to analyze the controversy from all of its fascinating angles, such as the design quality of the architecture, Lord Rogers (of the House of Lords) teaming up with British Republicans to denounce the monarchy’s interference with civilian life, or the absence of affordable housing in Chelsea. I am interested in only one question: is this traditional urbanism?

At first sight, the Terry design is reminiscent of the 18th/19th century style of palatial construction in Europe. (In fact one of the “blocks” features echoes of Buckingham Palace.) In terms of authenticity, the proposal is flawless. The Rogers proposal is also a palace, although one with much fewer attractive qualities. But does Chelsea really need a palace?

Providing a response to that inquiry is precisely what a system of urbanism is supposed to achieve. The system in place for London unfortunately requires that one have enormous financial means in order to participate in any kind of development, and inevitably that implies that only large speculative development will be so much as imaginable. The Chelsea Barracks proposal is entirely a product of modern urbanism, and by intervening into that system, the Prince of Wales and other traditionalists are sanctioning the very thing they claim to be opposed to.

As luck would have it, I wrote about the different processes of urban development using London neighborhoods such as Chelsea last year. Combining this with our models of the processes of urban emergence, we can develop the idea even further and try to conceive of a proposal for a traditional urbanism that develops the Chelsea Barracks site.

When I last covered Chelsea, I used its housing typology as an example of a linear, non-complex model of housing development. While linear housing is characteristic of the neighborhood, it is not the entire tissue of it. If we analyze the morphology of the neighborhood we find many clusters of housing rows, but these clusters do not necessarily repeat from one block to the next, and they are intermingled with other, uniquely programmed buildings of varying scale, the most prominent of which being the Royal Hospital. This means that, despite the neighborhood’s texture being only semi-random and not completely emergent, it performs at a remarkable level of complexity.


This kind of fabric is very common of British-American subdivision development during the 19th century. Here it is in a pure grid form in one of Montreal’s inner core “Plateau” neighborhoods.


We can observe that the middle of blocks is populated very differently from the major streets, despite the fact that they are not very different from a design standpoint. We don’t need to propose anything more complicated than self-optimization to explain this pattern. During development, housing builders would work from the center of blocks outwards, where there was the least perceptible traffic, and shops, churches and other activities located where there was the most traffic. The outcome is a complex tissue with perceptible characteristics, not only random noise.

In comparison, here is the texture of a new neighborhood in Las Vegas (Henderson), Nevada.


It is the same housing model repeated a thousand times, some lots facing backwards from the main roads. This new neighborhood might as well define linear development processes. The only feature of this neighborhood is the house, and so it can only function at any level of complexity by ejecting its residents out into town for any activity.

Of course some might say this is not a fair comparison. Those old neighborhoods are old, and therefore have had a long time to achieve maturity. But a neighborhood maturing implies that the neighborhood is planned to have a life cycle taking place in time, of which the early stage of growth is critical to its final morphology. What did a young, new neighborhood look like in 19th century British-American urbanism? It consisted mostly of very large lots of gardens and other large events (such as, for example, a Royal Hospital). These new neighborhoods were advertised as a pastoral refuge from the city. Look at this engraving of Milwaukee’s outskirts in 1858.


Its overall density is much lesser than that of Las Vegas new neighborhoods, and it has a distinctively pastoral quality. Yet what happened to those traditional neighborhoods was often that, very rapidly (the span of 2-3 decades) they became very dense urban neighborhoods, at which point the rate of new growth plunged and the fabric remained stable.

See for example this comparison of the urban fabric (1897-1915) of the Morningside Heights neighborhood of Manhattan, from the book Retrofitting Suburbia: Urban Design Solutions for Redesigning Suburbs.


In less than two decades the neighborhood was populated from a pastoral grid dominated by the campus of Columbia University to the dense, New York-style neighborhood it remains today. In these examples, growing a mature neighborhood was intrinsic to the development process adopted, and once this process wound down there was very little left to change to it. The neighborhood was mature because it had reached its equilibrium with the socio-economic context. Correcting deficiencies is what is meant by a neighborhood maturing, and developing a mature city implies that one is avoiding mistakes during its development. If we are employing a development model that is set in advance, no mistake can be either corrected or avoided during production.

Notice then that in traditional neighborhoods, the construction of mass-produced housing came last, after the neighborhood had established itself as a socio-economic system at the fringe of an existing city. Because of this, the mass-produced housing is a complement to that tissue, and contributes to the established complexity of the neighborhood, even super-charging it with population density. This not only ensured that there was no environmental alienation for the new residents, but also that there was a limit to how much repetition there could be from housing builders.

In modern urbanism we require all new developments to be programmed for a certain type of use, whether we are building a housing subdivision, an office or industrial zone, or a “mixed-used” development. If this is not know and debated in public, no development project can be approved. Only when a proposal has gone through this ordeal can anything be built, and making changes involves going through the process again, so the developers just subsidize the mistakes, or leave certain parts of the plan unrealized and a gaping hole.

In traditional urbanism this is never necessary. In fact it is possible for entire blocks to be left as pasture or gardens, creating an ultra-low density urban tissue. Only as further development becomes truly needed are these blocks transformed into housing and other programs. A critical difference is that no planning permit or approval is necessary to further develop a neighborhood. Instead the residents have an envelope of building rights set in building codes, and everything within that space is considered to be automatically approved. Because of this the development of a neighborhood can be undertaken in a large number of successive decisions, where the next building to be added is not only determined by the citywide market but also by the current state of the neighborhood. This in turn allows a local community and economy to grow, which is absent from modern developments.

This is all very interesting for new neighborhoods, but how would that apply to a small urban redevelopment site in the middle of a centuries-old neighborhood? Clearly we aren’t going to be building up from pastures. This is where a “new” traditional urbanism becomes relevant, as we need to invent a new process that restores the features of traditional urbanism, but can also function in the context of mature cities and modern structural requirements.

Although the redevelopment of a large urban block is usually undertaken as a large real-estate project, it can also be considered as a nested process of urbanism (urbanism within urbanism). Much like the city-wide process of urbanism is characterized by regulations intended to achieve equilibrium, the redevelopment of a block of the city should also be designed as to achieve its equilibrium with the city as a whole but (and here is the defining characteristic of a traditional urbanism) also within itself. This is what does not happen in linear development processes such as housing subdivisions, or 19th century housing terraces. They provide equilibrium with the larger scale, but amongst themselves they provide no complement. For this reason, although you’re likely to see a lot of some housing model repeated in one place, you’ll rarely ever see it used again elsewhere. Mass-production does not work for buildings the way it does in automobiles.

Time and interaction are the critical factors. The reason large-scale development like the Rogers and Terry proposals get approved and built is that everything must be conceived and approved in one step. The architectural design is rushed in order to make proposals as soon as possible. The form can’t evolve over the course of development. This process is justified by the need to control the architectural character of the city, but it is not necessarily so. It only follows from controlling architectural character because we rely on static information systems and processes to conduct building. In fact, many of the traditional building techniques that preservationists are attempting to preserve do not translate into modern information systems (building plans). If instead of drawing the full plans, the proposals simply supplied the component patterns and a parameter space for them, then there could be an infinite variety of different instances of these patterns populating the new space, all fitting a particular need and applying a specific method of returning to equilibrium. If we wanted to release control even more, we could define some buildings from the neighborhood as models and whatever patterns they featured as automatically approved. And seeing as this is the 21st century, we could define these patterns inside software that could randomly generate any possible permutation, such as the City Engine.

With the architecture out of the way, there would only need to be a fixed design for the frame of spaces around which the urbanization will take place. Grids are flat and unspecific, and so a good project will have a place structure that creates inequalities of movement. (Even New York’s grid has subtle inequalities in the short-blocked avenues and long-blocked streets, creating vastly different spaces in character.) Crescents, squares and alleys on a completely open surface should be the extent of planning a new neighborhood, and it will be important that this design have value all of its own. It is quite possible, for whatever economic reasons, that only part of the surface will be built, or even that nothing will be built. A good urban design must work in all states, including with nothing on it. Remove the buildings from the Terry proposal and there is still a rather interesting landscape. The Rogers proposal, without its buildings, has nothing. Terry is therefore much closer to the goal.

Negative space in modern plots

Negative space in modern plots

In the final step, how does the developer make money? Sustainable development, after all, has to be profitable in order to be sustainable. In a traditional city, plots were subdivided over time as the need arose. In a modern city lots are defined as a standard shape, and then later sold off for some standard price. This approach has the unfortunate side effect of creating a lot of negative space. The developer of the Chelsea Barracks could instead sell or auction off space as an elastic product. The first buyer would choose the first spot on the open surface, in relation to the hierarchy of the urban grid. The second buyer would place himself in relation to the urban grid and the first buyer. These buyers at first would come from long-time residents of the neigborhood aware of some particular way of extending the neighborhood, but unable to find a lot a space at the right size before this project became open to the public. This process would continue until all the space had been consumed, and the end result would be that all buildings would be related to one another through the sales process. If the space was priced high enough, the later projects would only be initiated after the initial ones had been completed, and the impact of time would generate the demand for building programs complementing the initial projects.

In such a way the urbanism within urbanism would create its own socio-economic subsystem, would feature a randomly adapted but uniform architectural signature, and would complement and extend the external urban tissue.

While I’ve detailed a process for developing a small block within a city, this process is just as applicable for doing development of new cities, or new suburbs of cities. There are fewer constraints and difficulties involved in these other cases, which is why I wanted to use the example of the Chelsea Barracks site. Urbanisation is a universal phenomenon, and although the patterns change, the underlying principles are everywhere the same.

Principles published

The full article conceptualizing the principles of emergent urbanism has been published by the International Journal of Architectural Research volume 3 issue 2. You can download the complete article or read the whole issue.

The Journey to Emergence

This is part I of a series of excerpts of an article to be published in the International Journal of Architectural Research entitled The Principles of Emergent Urbanism. Additional parts will be posted on this blog with the editor’s permission until the complete article appears exclusively in the journal’s upcoming issue.

Of the different domains of design urban design is an oddity. While the design of a machine can be traced to a definite, deliberate act of invention, and even the design of buildings (architecture) is rooted in known production processes, the design of cities was never seriously attempted until well after cities had become a normal, ordinary aspect of civilized living, and while the design of machines and buildings was a conscious effort to solve a particular problem or set of problems, cities appeared in the landscape spontaneously and without conscious effort. This places the efficacy of urban design in doubt. The designers of machines and buildings know fully how the processes that realize their design operate, and this knowledge allows them to predictably conceive the form they are designing. Urban designers do not enjoy such a certainty.

How is it possible for what is obviously a human artifact to arise as if by an act of nature? The theory of a spontaneous order provides an explanation. According to Friedrich A. von Hayek (Hayek, 1973) a spontaneous order arises when multiple actors spontaneously adopt a set of actions that provides them with a competitive advantage, and this behavior creates a pattern that is self-sustaining, attracting more actors and growing the pattern. This takes place without any of the actors being conscious of the creation of this pattern at an individual level. The spontaneous order is a by-product of individuals acting in pursuit of some other end.

In this way cities appear as agglomerations of individually initiated buildings along natural paths of movement, which originally do not require any act of production as dirt paths suffice. As the construction of individual buildings continues the most intensely used natural paths of movement acquire an importance that makes them unbuildable and these paths eventually form the familiar “organic” pattern of streets seen in medieval cities. This process still takes place today in areas where government is weak or dysfunctional, notably in Africa where urban planning often consists of catching up to spontaneous settlement, and in the infamous squatter slums that have proliferated in the 20th century.


A transect of the city of Tultepec in Mexico provides a snapshot of the different phases of spontaneous urban growth. (Google Earth image)

As urbanization becomes denser, the increasing proximity of concurrent, competing individual interests causes conflicts between the inhabitants of the emerging town. Individuals build out their properties in such a way that it interferes with others, for example by blocking paths or views. These acts threaten the sustainability of the spontaneous order, and to resolve this situation the parties involved appeal to the same judges that rule on matters of justice. These judges, again according to Hayek, are required to restore and preserve the spontaneous order with their rulings. These rulings provide the first building regulations and, when government authority becomes powerful enough to do so, are compiled into comprehensive building codes to be applied wherever the force of that government extends. (Hakim, 2001)

The compiled building codes are later brought by colonists to create new settlements, reproducing the morphology across multiple towns but each time in a pattern that is adapted to the local context. Early town planning efforts are attempts at regularizing the building codes in order to plan for long-term organization of cities, but maintain the spontaneous production process. Most notably the rapid urbanization of New York City was accomplished by very simple rules on the size of blocks laid out in the 1811 Commissioners Plan for New York. Unlike the experience of urbanization in previous centuries, where urban growth was slow and often stagnant, the urbanization of New York took place in a time of rapid social and economic changes, and the city government had to invent building codes involving issues that never could arise in a pre-capitalist society: first the tenement, then the skyscraper, and ultimately, the automobile.

Modernism: the replacement for the spontaneous order

Architects and urban planners of the early 20th century, confident in the techniques of engineering and industrial production, believed that the spontaneous city had become irrational and had to be replaced with a new design fully integrating new industrial technology. The Swiss architect Le Corbusier is famous for designing a complete city around the automobile and building models of his design. In so doing he adopted a process of urbanization that was completely planned hierarchically, applying the processes familiar to architects at the scale of an entire city. He also ridiculed the morphology of spontaneous cities as being the product of donkey-paths.


This scale model of Le Corbusier’s Plan Voisin marks the turning point where city plans as constraints on individual initiative are replaced with architectural design at the scale of millions of inhabitants. (Le Corbusier, 1964)

Although the architectural program of high-rise living of Le Corbusier was discovered to be a colossal failure, the modernist process of development replaced spontaneous urbanization in the industrialized world. The housing subdivision substituted adequately for the high-rise tower block, providing affordable housing in large numbers to a war-impoverished society. This production process is still in force today, separating cities into three distinct zones: residential subdivisions, industrial and office parks, and commercial strips.

Modern city planning has been successful at its stated objective, producing a city designed specifically around automobile use, yet it was immediately and has been perpetually the target of criticisms. Most significantly the vocabulary of these criticisms had to be invented in order to spell out the critics’ thoughts because the type of deficiency they were observing had never been seen. Words like placeless or cookie-cutter were invoked but fell on the deaf ears of urban planners who were trained in Cartesian processes and industrial production techniques.

The most devastating criticism of modernist urban planning came in the form of a sociological study and personal defense of the spontaneous city, the book Death and Life of Great American Cities by Jane Jacobs. (Jacobs, 1961) In it she described in great details how the functions of a spontaneous city related and supported each other. Her concluding chapter, the kind of problem a city is, is still the most relevant. In it she attacks the scientific foundations of urban planning at a paradigmatic level, and claims that the methodology of the life sciences, at the time undergoing the revolution created by the discovery of DNA, is the correct approach to studying cities.

Death and Life of Great American Cities has been adopted by contemporary urban planners as a textbook for urbanity. Its descriptions of the characteristics of a city are now the models upon which new developments are planned. The old urban development of housing subdivisions and office parks is being substituted for the new urban development that has streets, blocks, and mixed uses, just as Jacobs had described to be characteristic of life in the city. A major difference between Jane Jacobs’ preferred city and the new urban plans remains. The layout of mixed uses is organized and planned in the same process as Le Corbusier planned his city designs. The scientific suggestions of Jacobs have been ignored.

The discovery of emergence and complexity science

In the time since Jacobs published her attack on planning science molecular biology has made great technological achievements and provided countless insights into the morphology of life. In parallel the computer revolution has transformed the technology of every human activity, including that of design. But the computer revolution brought along some paradigm-altering discoveries along with its powerful technology. In geometry, the sudden abundance of computing power made it possible for Benoit Mandelbrot to investigate recursive functions and his discovery, fractal geometry, generated a universe of patterns that occurred in many aspects of the physical universe as well as living organisms. (Mandelbrot, 1986)

Some thinkers saw that the life sciences were part of a much more general scientific domain. They formed the Santa Fe Institute and under the label complexity studied not only organisms but also groups of organisms, weather systems, abstract computational systems and social systems. This research formed a body of theory called complexity science that has resulted in the creation of similar research institutes in many other places, including some centers dedicated specifically to urban complexity.

Their scientific revolution culminated in two major treatises within the last decade, both from physicists practicing in a field of complexity. The first was A New Kind of Science by computer scientist and mathematician Stephen Wolfram (Wolfram, 2002), where he presents an alternative scientific method necessary to explore the type of processes that traditional science has failed to explain, presenting a theory of the universe as a computational rule system instead of a mathematical system. The second was The Nature of Order (Alexander, 2004) by architect Christopher Alexander, where he presents a theory of morphogenesis for both natural physical phenomena and human productions.

A definition of emergence

To define what is meant by emergence we will use the abstract computational system upon which Wolfram bases his theories, the cellular automaton. Each cell in a row is an actor, making a decision on its next action based on its state and the states of its direct neighbors (its context). All cells share the same rule set to determine how to do this, that is to say all cells will act the same way with the same context. In this way each row is the product of the actions of the cells in a previous row, forming a feedback loop. The patterns of these rows are not in themselves interesting, but when collected in a sequence and displayed as a two-dimensional matrix, they develop complex structures in this dimension.


The 30th rule of all possible rules of one-dimensional cellular automata produces a chaotic fractal when displayed as a two-dimensional matrix, but most other rules do not create complex two-dimensional structures. The first line of the matrix is a single cell that multiplies into three cells in the second line in accordance with the transformation rules pictured below the matrix. This process is reiterated for the change from the second to the third line, and so on. All the information necessary to create structures of this complexity is contained within the rules and the matrix-generating process. (Wolfram, 2002)

The same general principle underlies all other emergent processes. In a biological organism a single cell multiplies into exponentially greater number of cells that share the same DNA rules. These cells create structures in a higher dimension, tissues and organs, which form the entire organism. In the insect world complex nests such as termite colonies emerge from the instinctual behavior of individual termites. And in urbanization, buildings form into shopping streets, industrial quarters and residential neighborhoods, themselves overlapping into a single whole system, the city.


Alexander, Christopher (2004). ‘The Process of Creating Life’, The Nature of Order Vol. 2, Center for Environmental Structure
Corbusier, Le (1964). La Ville Radieuse. Éléments d’une doctrine d’urbanisme pour l’équipement de la civilisation machiniste, Édition Vincent Fréal et Cie, Paris, France
Hakim, Besim (2001). ‘Julian of Ascalon’s Treatise of Construction and Design Rules from Sixth-Century Palestine,’ Journal of the Society of Architectural Historian, vol. 60 no. 1
Hayek, Friedrich A. (1973). ‘Rules and Order’, Law, Legislation and Liberty Vol. 1, Routledge and Kegan Paul, London and Henley, UK
Jacobs, J. (1961). The Death and Life of Great American Cities, Random House and Vintage Books, New York, USA
Mandelbrot, Benoit (1986). The Fractal Geometry of Nature, W.H. Freeman, New York, USA
Wolfram, Stephen (2002). A New Kind of Science, Wolfram Media, USA

Design, configuration and natural form

When did human creations stop being natural? We look at a tower block, a subdivision or a shopping mall parking lot and see the worst of industrial civilization translated into form. We tolerate them as necessary to achieve the material wealth of our civilization. Those human settlements that are still natural we grant special protections through UNESCO and historical preservation laws. We do not have a law that promotes the creation of new historic settlements because we are not quite sure how they are made.

I believe that our mistake is not in the things we make, that there is nothing unnatural about a shopping mall parking lot from a design point of view. What makes the shopping mall parking lots we build so unnatural are errors in configuration of the design elements. To understand this, one must understand the difference between design and configuration.

The form of a tree is an ideal example to illustrate the difference between the two concepts. Any particular species of tree will have a design that is essentially the same from one tree to the next. The design elements in the tree are all the named parts: trunk, branch, leaf, root, bark, and so on. These parts are organized into hierarchical relationships with the whole tree and with each other. We will always find the roots related with the trunk in the same way. This relationship is a design solution that achieves a specific result. However, the position of any of the parts is not fixed. In the DNA of the tree are rules that instruct cells to adapt themselves to the larger context the tree finds itself in. The different design solutions that result from this cellular action will therefore adopt a position that reflects the particulars of time and place, ensuring that the tree’s form is perfectly adapted to its environment. This is why it makes no sense to create a description of the forms of a leaf in order to make another leaf – that form is relevant only to this particular leaf, and another leaf, although it would have the same overall design of parts, will take a completely different configuration.

Adapted to chaos

A chaotic configuration of a standard design

If you’re having trouble seeing this, imagine the following scenario: we take the DNA of a tree and clone it 100 times. Then we lay out a grid 10 trees by 10 trees and watch them grow. What would happen would be that every tree would come out a different way, since the earth around them would be structured differently, the wind patterns would be different, the shade and the moisture would be different. The trees would each have the chaotic, random shape that we know trees to have, yet would all be perfectly symmetrical with one another without being identical. Each clone would adopt a unique configuration of the same design.

When we look at a traditional village, we find that the same house design is repeated time and time again, but configured in such a way that it is differently adapted than the other houses. The reiteration of an often very simple design is all that it takes to create a natural landscape, so long as each house is configured to adapt to its place, and the design elements of the house are themselves configured to adapt to these adaptations.

One design, many configurations

Even today this kind of natural adaptation takes place in modern settlements where planning regulation allows it, or fails at forbidding it.


This is the skyline of Monaco, which by necessity of the small size of the city had to be built piecemeal but yet is still made with an entirely modern building stock. The piecemeal process allowed each building to be configured to its site and thus, despite the fact that the buildings’ design is very basic modern architecture, the whole landscape looks natural. It would be even more natural were the architectural elements also adapted.


This the Rocinha favela of Rio de Janeiro. Here the building design is as bare as could be made, the houses being built by poor residents with little capital to invest. But the resulting configurations adapt perfectly to the shape of the hill and the other buildings, and the overall look of the place is that of a human jungle. (If you have the chance to see this summer’s The Incredible Hulk, the movie makes this point by fading from an overhead shot of Rocinha to that of a tropical jungle, subtlety be damned.) The buildings in Rocinha are just as natural as the trees.

How does that translate back into our shopping mall parking lot? It means that although the relationship between the parts, for example the lanes, the spaces and the paint that demarcates them, must be defined, the length of the spaces or the thickness of the demarcations do not have to be identical from one element to another. The chaos of nature requires that they be slightly different from one to the next, and that means that the people who make them must be able to make decisions while they are building. Simply copying an AutoCAD drawing is unnatural. The design must be translated into a language that instructs the builders to make configuration choices while constructing the defined forms. This kind of language is how builders have made traditional towns and how DNA makes organisms.

Separating design from configuration also allows us to make a second attempt at city planning. The plans of modernists all had fixed configurations, and their failure to adapt to their context meant the failure of urban planning. The conflict between design and configuration planning dates back even further, to the 19th century plans for Barcelona and Paris. In Barcelona, Cerda planned a grid of square blocks through which he ran grand diagonal avenues. Those were only two design elements in a very strict configuration that was made possible only by the enormous economic pressure to expand Barcelona. In Paris, Haussmann did not have the luxury of expanding the city with blocks, he had to upgrade a city of blocks that already existed with a new design element, the grand avenue. He deliberately left the configuration of his avenues open until they were completed, and placed them where he met the least resistance. Their effect on Paris is even today essential to life, and they could not have been realized unless their configuration was left adaptive.

What would a natural urban design look like? It must first be a definition of parts that must be related to each other in order to create urbanity. Describe the relationship between the avenue and the streets, the streets and the alleys, describe the relationship between the avenue and the pavement, the pedestrian crosswalks and the shade trees. Describe the relationships with the buildings without delimiting their size and shape. The city builders will then decide in what configuration these elements need to be to fit their context, and the resulting built form of the city plan will be perfectly natural as well as fully planned.

Classicism describes itself as the imitation of nature. Complexity, on the other hand, does not imitate. It is nature, applied to different problems. To create the urban design of our time requires not adopting a certain style or program, but ensuring that any style or program can be adapted to a particular context. It only requires us to use different tools than what we have become accustomed to.

Further reading:

Complex geometry and structured chaos, part I and part II.

More evidence that New Urbanism is really dense sprawl

From The New Geography magazine.

In Celebration, many of the early residents were Disney executives; only 4 or 5 years after opening did Disney develop office space in Celebration for some of their offices. Baldwin Park, approximately 2 miles from Downtown Orlando, never pretended to capture the employment aspect, instead selling itself (to many Celebration residents who rushed to this newer, hipper version of their town) as a downtown commute. And neither Avalon Park nor Horizon West have employment opportunities within their town centers. What they do have is easy access to the area’s ring road – ensuring vehicular congestion outside of their New Urbanist communities.

What is in their Town Centers? Ironically, you find only a small shopping district and the ubiquitous Publix, Florida’s home-grown grocery store chain. The formula of “live-work-play” must stick in the craw of those who are employed in these stores, because the Publix employees, Starbucks baristas, dry cleaner cashiers, and others who do work in these Town Centers can not possibly afford the New Urbanist real estate. Rather than a social continuum (as was more common in the idealized version of America), there is a new social schism, with the New Urbanist underclass forced to commute to the New Urbanist communities from more affordable but less trendy housing nearby.

In contrast, the region’s native communities have been thriving throughout the same growth period. Communities like College Park, adjacent to Orlando’s downtown, offer something that New Urbanist communities do not: diverse housing, from garage apartments and rental communities up to stately mansions, all within walking distance of each other. They offer an idiosyncratic mix of sacred places, playgrounds, schools, and shops in what the Philadelphia architect and theorist Robert Venturi calls “messy vitality.” No overarching body dictated the form, developed transects, or rigidly controlled the distance between the front porch to the street to achieve these vibrant, socially cohesive, and proud neighborhoods.

New Urbanists claim to reduce the need for cars, but Orlando’s New Urbanist communities make the car more necessary than ever. Built on the periphery of the metropolitan area, they require a vehicle to complete the circle of functions necessary for a healthy society. Orange County planners have been submissive to the New Urbanists – especially after Celebration – but increasingly recognize that they do not solve the problems they claim to solve and instead invent more: higher traffic, less affordable housing near city centers, and lumpy development sprawl.

If you are building in a city at the metropolitan scale, you have to expect your potential residents to live metropolitan lifestyles. A single TND is nothing more than a prettier subdivision, and brings along all the economic risk and maladaptations that other subdivisions do, with none of the flexibility, agility and adaptivity of regular cities. But the blame here doesn’t fall on the developers of New Urbanism, it falls on the county planners who are supposed to enable the flexibility, agility and adaptivity of their metropolis, and who instead create the ideal conditions for unsustainability and subdivision development. There wouldn’t be these TNDS without the ring roads, which immediately become unplanned urbanism. That’s the only reason this kind of development is profitable in the first place.

The challenge of dense sprawl

When looking at such a picture we are at first inclined to make a parallel with the landscape of Los Angeles. It is a foggy urban plain with a cluster of towers popping out over the horizon. This is Dubai from a perspective that is rarely shown, that of the city in the foreground, and its suburban expansions in the background. The towers are the Dubai suburbs, lined up on what used to be the main highway out to Abu Dhabi, now the centre of New Dubai and the urban fringe of Dubai. A picture such as this is significant because the central core of Dubai is almost never seen in the pictures of the huge developments going up in its suburbs. That is unfortunate, since the reason that these suburban developments are economically possible at their size is because they are growths of the old Dubai. The parallel with Los Angeles is therefore incorrect. Los Angeles followed the standard American model of urban growth, developing a central business district on a simple grid where the original center of the city was founded, then later adopted suburban sprawl to continue its growth and grew a cluster of skyscrapers adapted to this sprawl where the center once was. The policy of sprawl is blamed for the impossible traffic congestion that cripples Los Angeles, but generally what is meant by sprawl are the low-density housing subdivisions, office zones and other standard typologies of suburbia. The solution that was called for was more density, even though some geographers pointed out that Los Angeles was already one of the densest cities in America. The result of this choice has been dense sprawl: worse traffic, worse crowding and seemingly no improvement in quality of life.

Clearly the challenge of sprawl has been improperly identified. I will show that sprawl is not about density but about distance between complements, and the extremely rapid urban growth of Dubai, from a small fishing town to an urban metropolis in two generations, makes this visibly explicit.

Perhaps the most fascinating fact about Dubai is how natural and complex the old city seems to be despite having an entirely modern building stock. Most cities associated with a natural or organic morphology are usually pre-modernist cities of old classical or vernacular buildings, or at least preserve some of them. There are no such buildings in Dubai, the city having started its growth period in the 1950’s. The only explanation for the natural form of the old Dubai is a natural process of growth.

The urban fabric of a city is a solution set, each building being a solution to a problem of a particular time and place. The city as a whole is a solution set for its population as a whole, and as times and people change new buildings are added to provide new solutions to these new problems. The reason that naturally grown cities have a chaotic morphology is the same reason why stock market movement is fractal; they are both adaptations to fluctuating circumstances, and they are both limited in size by the previous size of the system. That means that you cannot grow a city by a development that is bigger than the city’s current state of maladaptation to circumstances. Attempting to do so will result in economic failure. Many very small development operations done in succession will be much more adaptive than one large development operation planned at one moment, since each operation adapts to the circumstances created by the previous one.

Because cities grow by correcting maladaptations, the new buildings turn out to be complements of the existing buildings. It makes no sense to make a building that is identical to one that already exists, and in natural cities that will never appear. Each building will be fitted to the particular knowledge of time and place, as Hayek would say. This particular knowledge is itself produced by the presence of existing buildings and the way people use them.

This process is not simply an economic abstraction, it also has morphological consequences.

Jumeirah is a beachfront suburb directly to the south of old Dubai that has become the home of the Anglosphere expatriate community as well as synonymous with upper-class lifestyles. Because it is more recent growth of Dubai’s urban centre it has seen growth in operations of much larger sizes. You can tell that there are now “clusters” of identical buildings, nothing of the scale of a metropolitan subdivision yet, but the building individuality of central Dubai is no longer present. (Cluster housing development is also visible in, for example, the older parts of Las Vegas.) This is pre-subdivision scale, in that an operation creates multiple buildings without having a scale large enough to be “planned” and have status as a named development. For the purpose of the adaptation they bring, they are still only one event, only one adaptation subdivided into multiple buildings. And because they are much bigger than the previous development operations, they create more distance between themselves and the urban fabric they are adapting to. The scale of development is not big enough to be called sprawl yet, but the complexity of this neighborhood is not as advanced as that of the older city. While the buildings in Jumeirah are certainly complements of the whole city of Dubai, they are not complements of each other, which is true in the older city.

Left: Skyscrapers on Sheik Zayed Road, Burj Dubai site on lower left

Right: Palm Jumeirah, Dubai Marina megaprojects along with random subdivisions.

Here are Dubai’s world-famous megadevelopments, which have been made possible as adaptations to Dubai’s metropolitan scale. The development operations are enormous complements of the existing urban fabric, but the fact that they are all being built concurrently means that they cannot adapt and become complements to each other. With little surprise the traffic congestion on the Sheik Zayed Road that integrates them together has skyrocketed. The distance between complements has increased to the scale of the projects, and the only way to move from one to another is by driving down Sheik Zayed Road.

The skyscrapers going up are actually making sprawl worse the same way that a development of 100 houses extends sprawl. The idea that a skyscraper is a vertical city is a myth. Skyscrapers are identical floors of open space and rely on very large networks of complementary urban fabric to work properly. Skyscrapers can contribute to the complexity of a place like New York City because the urban fabric of Manhattan is very rich and can digest density and congestion gracefully, but in a city where urban fabric is undeveloped, as modernist plans for a Radiant City were, or has been de-developed, like the CBD of most American cities, skyscrapers throw congestion and traffic out to the whole city as badly as a housing subdivision.

The solution to sprawl is not increasing density, but increasing complementarity. That means breaking up existing housing subdivisions, office parks and shopping centers into smaller autonomous parts that can grow into buildings that are complements to houses, offices and shops as well as complements to the city as a whole. (This obviously implies abolishing zoning.) It means not trying to design everything that goes into one project, to let growth come to you and to accomodate it, to publicize the chaotic symmetry of old Dubai as a model of natural beauty instead of the iconic forms of the skyscraper cluster. The Dubai Palms and the Burj Dubai district may still grow into something natural and complex.

Creating the emergent dimension, or learning from Wikipedia

In Architecture: Choice or Fate, his manifesto for New Urbanism, classicist Leon Krier produced many inspirational images of urban complexity, going as far as a fractal comparison of modern and traditional buildings. The cover of the book, a fictional resort town for Tenerife, presents a fascinating case study of complex symmetry; no building is the same as another, but all share the same geometric properties. That would not be unusual had it not been an architectural manifesto. Competent artists have always been able to imagine dream cities, and they continue to do so with every blockbuster fantasy movie that hits the screens.

The dream city of Coruscant in Star Wars

The dream city of Not London in The Golden Compass

Like the dream cities of Le Corbusier and his modernist colleagues, the dream cities of artists have in common the fact that none of them have ever been realized. Leon Krier’s dream city, Poundbury, has been realized by the capital backing of a supernaturally rich patron, and even then it has been built very slowly and carefully. In the time it has taken to develop Poundbury, millions of urbanisations have occurred elsewhere. Like the New Urbanist TNDs of America, no matter how much we enjoy their architectural quality, we cannot consider them to be real cities. The real city has been built in the emergent dimension, not by the mind of a single artist but by the material necessities of all people. While it is fairly straightforward for an artist like Leon Krier to invent and apply his own form language to imagine a complex cityscape, in order for this design to be adapted to the material necessities of millions it must also involve millions. We obviously cannot burden a single artist with this task.

And so urbanism has a wholly different starting point from architecture. The artist does not have control. No one can possibly ever have control. Everything is happening all at once everywhere at once. To attempt to stabilize this process has caused chaos everywhere.

The starting point of urbanism is the same starting point that the founder of Wikipedia, Jimmy Wales, set for himself when he established the Internet’s now most indispensable website. Instead of asking how to publish the expert opinion of specialists as an encyclopedia that would compete with the print powerhouses (a venture he had already attempted and failed at), Wales based his system on the theories of economist and complexity scientist Friedrich von Hayek. The idea that Hayek proposes is that there exists specific knowledge that only individuals possess, and that can only be utilized with their cooperation. Wales saw his task as the aggregation of this knowledge into one coherent system.

The world-wide-web had, since the early 90’s, become a massively hyperlinked knowledge network that everyone could publish in. The reality at the time of Wikipedia’s creation was that this power had not produced any kind of coherent system for basic knowledge. Aggregating knowledge had up to then been too complicated. Wales wanted, in his own words, to “make the Internet not suck.” Overcoming this deficiency meant simplifying the production of web pages and hyperlinks, removing some unnecessary choices in the process. This is what the interface of Wikipedia did. Within a year, Wikipedia had grown explosively and exponentially. Making information easier to create and access had made it possible for the sum total of encyclopedic knowledge to be rapidly constituted.

This has come at a cost, however. While there is in theory unlimited freedom to add content that one considers relevant to Wikipedia, the form that this content will take on screen is very rigidly defined. This is necessary in order to achieve the complexity of the system. The design of Wikipedia is constant across every page in order to make it possible to rapidly navigate through all the information without having to relearn the rules for every article. Nevertheless, when we click on a link to a Wikipedia page, we never know what we are going to get. The design acts not as a constraint on the content, but as an enabler of the content. Without the rules enforced by Wikipedia, none of the content would have been added.

Wales and his foundation have been extremely controversial. On the one hand, in order for Wikipedia to work as it does, the foundation must provide all the support structures necessary to enable the users to create knowledge, and on the other hand it must also blindly trust the users to create information that will be accurate. It simply is not possible to control the content of so many millions of articles. To attempt this would necessarily shrink the size and reduce the complexity of the system, destroying what makes Wikipedia useful in the first place. The emergent dimension must either be embraced or rejected.

The lessons learned from Wikipedia can also be learned from the very rich past of urban planning. Historically the most successful cities have not been those who have had the least planning but the most enabling plans. The Manhattan plan of 1811, for example, provided for the flexible extension of a street grid without interfering in what could be built within the blocks, and so enabled a surge in urbanisation that was unmatched in history. Eventually this model reached its complexity limits and a new design for Manhattan was applied (with varying success), such as the building codes that gave us wedding-cake skyscrapers, and the metropolitan transit system of subways and later on expressways.

The darker side of this phenomenon has been the creation of city designs that inadvertently enabled the creation of a type of city that no one wanted. Every city, no matter how loudly the local authorities claim to be planning-free, have a design. Take the classical example of a “no planning” city, Houston. Although it has no zoning codes, Houston has a system for laying down a grid of roads that implies necessarily a large-scale, long-ride, automobile-dependent city. By their very form, these roads make some types of urbanisation easier and others more difficult. Who is going to build for a walkable neighborhood when there are no sidewalks? How can a TND make a walkable city when at every mile a thoroughfare cuts off the pedestrian links? How can a sustainable city emerge when only one form of link, long-range auto trips, can be made between destinations?

The threat that we face today is not suburban sprawl. New Urbanism and Smart Growth have been victorious in that aspect. The danger we face is dense sprawl, (see Eric Eidlin, The Worst of All Worlds Los Angeles and the Emerging Reality of Dense Sprawl) where our disconnected cities become denser and denser without becoming more complex, resulting in even poorer urban conditions. The suburb is not the design. Sprawl is the design.

Our cities may not be what we wanted, but they have not been accidents. They are the result of designs applied by the local (and sometimes not-so-local) authorities. They will only change if we invent and apply new designs for them. They cannot be architectural designs founded upon control of the artist. They also cannot be the endless grid of highway strips. Yet they must have both art and highways. It is a whole new method of design, emergent design, that we must master.


Wikipedia and Beyond: Jimmy Wales’ sprawling vision