Emergent Urbanism

Rediscovering Urban Complexity

Urban complexity in the practice of urbanism

  1. This presentation was originally given to the Complex Systems Laboratory of the University of Montreal. As such, it was unusual in that of instead presenting complexity science to urbanists, I had the task of presenting urbanism to complexity scientists. I will proceed by first giving an overview of urban complexity theory that has developed in recent decades, followed by a condensed history of urbanism to provide context, and in so doing be able to explain how the practice of urbanism must be reconsidered in light of the theories of morphogenetic complexity, finally providing a model of emergent urbanism.
  2. Complexity science is a field that grew from chaos theory and the study of fractals in the 1980’s. Starting in the 90’s it propagated to many different fields of scientific study. Researchers in geography, architecture and urbanism attempted to import the approach in their studies. I will present the most important results.
  3. One of the first to study urban complexity was the French geographer Pierre Frankhauser, who showed, by conducting multiple studies of urban built surfaces, that there was a real geometric structure to them, and that it is possible to extract the fractal dimension of an urban area by analyzing the built and unbuilt zones. He thus proved that the city is a complex structure with a geometric order, and not simply a mathematical anomaly.
  4. Across the channel two laboratories of University College London followed their own research tracks. Space Syntax, shown here, reduces the urban network to a set of spatial relations. By measuring the « integration » of a space relative to others, it can predict around 60% of the traffic volume the space carries. Another interesting fact is that Space Syntax predicts land use and, if integration analysis is restricted only to local areas, locally-specific land uses as well. Researcher Bill Hillier constructs a theory of the « movement economy » from these results.Note that the visualization of the analysis produces a fractal structure. We can conclude that land use is determined by the totality of the urban network, and cannot simply be imposed in a local sub-area (at least in a system that allows this kind of local adaptation).
  5. Michael Batty, with the UCL Centre for Advanced Spatial Analysis (CASA), dove into complexity simulation methods, such as cellular automata, agent-based models, and catastrophe models. His enormous treatise, Cities and Complexity, is more of a study of complex models than a study of urbanism. The author himself admits in the first chapter that the book is only an exploration of the potential for complex systems to explain urban form. He thus shows the output of cellular automata as morphologically similar to cities.
  6. Nikos Salingaros takes a more material and less analytical approach to urban complexity. Relying on some physics concepts, he explains what he names the « urban web » as a set of connections between physical spaces that are differentiated and complementary, and which cannot be subdivided in sub-sets.Salingaros uses every complexity theme available to him, while avoiding the creation of a complete theoretical system. In the image above, he builds an image of urban public space that shows fractal scaling.
  7. These are all interesting theories if someone is a university-sponsored scientist. However, when I would show complexity theory to urbanists and architects, the question they would always ask me was how to apply this to their work. Admittedly, the results of inquiries made by scientists up to 2000 are not very material. We can better understand the nature of the city, without ever touching the nature of urbanism. Can we really claim that the modern city, with its zoning and automobiles, can have the complexity of a city such as London? To reach such a conclusion, we have to study the production processes of the city.
  8. Contrary to most human creations, we cannot describe the city as being an invention. Cities have been for many centuries simply a part of the human ecology, without people being concerned as to how they were produced. Urbanisation became a preoccupation only when the scale of the city reached such a size that severe problems appeared as a result. Movement and circulation flow has been, and remains, the prime imperative of urbanism, to which has been added over time social and technological development as secondary objectives. With every change in scale, the measures used to solve the problems of the urban crisis have been heavier. However, no attempt has ever been made or succeeded in understanding the nature of urbanization itself. Instead people have sought to constrain urbanization to a stable pattern.As a consequence of this evolution, we can distinguish three characteristic periods in urbanism: organic urbanisation, pre-industrial urbanism, and modern urban planning.
  9. The « original » form of the city is nothing other than the absence of planning. In European civilization organic urbanization ruled as soon as the collapse of the Roman Empire made any organized production of the city impossible. For one millennium, the towns and villages of Europe existed as the backdrop of urban life, without any exceptional quality being noticed about them. It is only in the modern age that we have labelled them with a distinctly organic and natural appearance, and in the last decades enormous planning efforts have been deployed to preserve them in that state.
  10. Organic urbanization possesses characteristics which, I will show later, provide it its organic qualities. Firstly, the occupation of land is done randomly, with no artificial limits such as property lines. Land use is established through negotiation, and so the pattern of built and non-built space is the result of a recursive process where the streets and squares are those spaces that could not be built over without breaking some neighborhood-preserving regulation.An interesting fact to note is that those places in the Mediterranean coast which share a genealogy in building codes have a tendency to produce similar morphology, even though their form adopts a unique configuration in each place.
  11. It must be noted that, despite civilization’s progress, organic urbanization did not disappear with the middle ages. On the contrary, all types of urbanization continue to exist concurrently, particularly in third world countries where governments are more or less potent. On the left is the emergent grid of Tultepec, a suburb of Mexico, whose network is made by natural movement. On the right is the Rocinha favela in Rio de Janeiro, an old shantytown now legalized and improved but which maintains its organic features.
  12. Organic urbanization functions pretty well up to a limit, which the Europeans discover in the Baroque age when urban economic development accelerates and many cities reach populations of more than a hundred thousand. As these cities’ growth is founded on trade, traffic circulation problems are critical. The organic « donkey path » works only for local trips. Whenever the opportunity shows itself, the powers that be impose straight lines for new neighborhoods, but urban renewal plans fail where capital or power is missing. This is fatefully the case for the Plan for London of 1666 (shown right), drawn up after the city had been completely destroyed by the Great Fire. The citizens decided to simply build it back the way it was before.
  13. Here we see the transition to pre-industrial urban planning in the fabric of the Amsterdam city center. The organic medieval heart is encircled by canals constructed in very straight lines added to meet the city’s new scale.Venice is an interesting exception. It was built on a cluster of shallow islands separated by water ways that were already hierarchically structured, giving its commercial operations an extra advantage. The Grand Canal thus ensured efficient traffic circulation up to the day when the city became obsolete as a trade port and transformed itself into a tourist attraction.
  14. In the 18th and 19th centuries urbanization is planned on uniform grids that, according to rationalist theory, will ensure the regularity of traffic circulation and perpetual urban order. Grids are imposed by law, such as in the Code of the Americas that regulates the urbanization of the Spanish Empire and which morphology is recognizable in cities as far apart as Mexico, Havana, Santiago and Buenos Aires.With the new cities being split up in a grid of identical blocks, it follows that buildable lots also should be identical in shape, however those lots are subdivided again and again such that, as is shown in the image of the Eixemple in Barcelona, the urban tissue nevertheless achieves a fractal emergent dimension.
  15. The apotheosis of the urban grid is for most people New York, but New York shows an interesting flaw in the system. When multiple villages merge into a single agglomeration, as was the case for New York in the early 19th century, the order provided by the grid turns into total confusion. For this reason the state of New York merged the entire island into a single monolithic corporation and imposed the plan of 1811 with its world-famous numbered streets and avenues. (Manhattan alone has street numbers reaching 200.)
  16. Despite its simplicity, the grid no longer succeeds at ordering the metropolitan city. Circulation over a hundred streets becomes impossible, just as the subway and the new skyscrapers make possible the super-dense metropolis. Tramways and cars mix to exacerbate urban congestion. The urban grid as imagined in the renaissance is in crisis.
  17. 20th century architects, confident in the promises of rational planning and industrial engineering, re-design the city at the scale of several million inhabitants. The most famous of them, Le Corbusier, proposes the Radiant City, which separates pathways for pedestrians, cars and public mass transit in several distinct spaces, and also separates the activities of the city, residential, commercial, industrial, recreational, in segregated zones. The objective is to untangle the rotten and endless industrial metropolis. This is modern urban planning as we know it, which continues to be enforced, even though few remember the reasons for its invention.Notice that in Le Corbusier’s Radiant City, the architecture is planned as strictly as the network and the land uses.
  18. We see here the modern city as produced by modern urban planning. Residential towers and slabs were rejected in the western world, even if they were successfully imposed for Moscow (left). Elsewhere the housing subdivision easily substituted for the residential slab to fulfill the living function, producing the landscape on the right (Las Vegas). Strangely, the space that the modernist architects had provided for pedestrian and mass transit pathways has been removed from the system.
  19. While it took centuries before the scale limits of organic urbanization and pre-industrial cities were reached, modern urban planning was criticized as soon as it was first applied. The critics protested that the city was sprawled out, there was no community, the trips were endless and consumed too much energy, too much land was consumed to build suburbs, street names confusingly all begin in the same letter or feature trees, one can barely locate one’s own house, the whole is totally alienating, and so on and so forth. Worst of all, the citizen doesn’t know whom to turn to in order to change things.Despite all these criticisms, the system continues, as the only alternative is to roll back to systems which we know are not capable of resolving the metropolitan scale. Complaining that the system is too big is futile when the system’s purpose is to resolve bigness.If there is an apotheosis of modern urban planning, I propose the Dubai Marina development, an immense residential development project in the suburbs of Dubai, where all skyscrapers are built all at once, which is linked to the rest of the city by a single super-highway and a metro line.
  20. As the systems of urban planning progressed the production processes of the city went from a more or less spontaneous, individual act, to ultimately become a civil engineering project, with whole neighborhoods of hundreds of thousands of inhabitants designed on a single drawing board (these days, AutoCAD). In other words, we replaced spontaneous decision in time and place by the individual, with collective decision in advance, with the remoteness and abstraction of an imposed plan. It was in a way inevitable, since we did not know any other way to do things. Today, in light of new science, we must ask ourselves what sort of consequences these production processes have on urban complexity.
  21. Since the year 2000, two grand treatises on complexity have been published practically simultaneously and in parallel. A New Kind of Science, by Stephen Wolfram, completes twenty years of personal research on cellular automata, and The Nature of Order, by Christopher Alexander, completes twenty years of personal research on architecture. Although the authors took completely different starting points, they arrive at essentially the same result. Both scientists study geometric computational programs and demonstrate that by the reiteration of these programs, a complex structure in a superior dimension emerges.With these two theories we obtain a « unified » theory of complexity.
  22. Wolfram proposes a « new scientific method » which remains extremely controversial. In essence, he claims that the classical scientific method which proceeds by observing natural phenomena and formulating a system of mathematical equations which explains it is useful only for simple natural phenomena. For complex natural phenomena, this method has never succeeded at explaining anything. Now that we have computers, we can simulate all sorts of programs at negligible cost. It is therefore possible to observe the computational universe in lieu of nature, and to draw from it conclusions about our real universe, while knowing with full precision the laws whose behavior we are observing.
  23. Wolfram spent ten years exploring the computational universe with his Mathematica software before finishing his treatise. He observed that, in most systems, there existed distinct classes of emergence. These classes were the same with complicated programs as they were with one-dimensional cellular automata. We can therefore generally distinguish process by their level of complexity:
    - Dead or stable processes
    - Linear or regular processes that can be studied with classical science
    - Hierarchical processes which have a complex but predictable structure
    - Random processes, which not only have an unpredictable complex structure, but can also spontaneously emerge other kinds of processes, and thus can be deliberately programmed to run any kind of computation. (This is the Church-Turing thesis of computational universability.)Why is this relevant? Cellular autamata are geometric computers, where certain rules produce certain classes of morphology. It turns out that these classes are also visible in urban space.
  24. This is a housing development with regular morphology. What does it share with a cellular automaton? The development of a neighborhood or even the construction of a single building relies on a selection of geometric building rules. If a builder happens to always select the same rule, it is because the process he is using does not allow more complexity. This is the case with mass production, possible since the beginning of capitalism, seen here in a London neighborhood but also much more obvious in all forms of modern architecture or residential development.
  25. Sometimes architects will deliberately impose a hierarchical structure on a development that is otherwise purely linear. This is a technique that often distinguishes classical architecture from modern architecture. The architect selects nesting geometric rules to break up monotony and force a little bit of complexity. Unfortunately, for such an effort to succeed in urbanism, one has to start with an enormous project and a comparatively enormous capital investment, with the speculative risk that accompanies it. New Urbanism-style development has both succeeded and suffered from this approach.
  26. Yet historic cities are filled with examples of random complexity. That means, as in this example, that builders chose harmonious geometric rules to build randomly as economic and social need required it. The whole is a coherent space. Therefore, there must exist some development processes that are both random, able to adapt to any social-economic context, and also fractal, able to produce structure at a larger scale. They must be re-discovered.
  27. For Christopher Alexander, the problem of complexity is to find out the nature of the natural. Until the modern age, human constructions were in harmony with nature, if not wholly part of the natural landscape. There is no such thing as a distinct natural and artificial for Alexander. A farm is natural so long as the production process for it is natural. Modern farms, he finds, no longer fit this description.Alexander searched for the production processes that nature uses to create nature. He arrives at a general theory of morphogenesis.
  28. Relying on multiple examples from biology and physics, Alexander shows that systems which follow emergent morphogenetic processes, that is to say productions instructions founded upon action and adaption instead of plans that must be strictly follow, arrive at complex forms. He shows for example that the shape of a bone is the result of the actions of cells attempting to balance the stress they feel, that there is no plan for a skeleton, but that the skeleton is the result of a biological computation that itself depends on the architecture defined by the cells at an anterior stage of growth. (This fact explains why so many fractals are found in organic systems.)Alexander suggests the same class of codes in all human productions. If we employ morphogenetic processes, we will arrive at natural structures in equilibrium with the environment.
  29. What is needed to explain the form of historic cities? Nothing very complicated, since these appeared naturally in the landscape. It requires that a group of city-dwellers reiterate the same geometric rule, which has enough flexibility to allow the construction of any random structure, for each new step of construction. Such a rule is as natural as the rules that regulate the organic growth of plants and animals.The more stable the geometric rule, the more buildings are symmetrical with each other, without limiting individuality. It is the same principle as DNA.
  30. What would be the characteristics of morphogenetic urbanization? First, the territory of the city would have to be settled in successive steps, establishing a dimension of time. Second, an increase in the number of inhabitants would also increase the number of actors on the environment and form of the city, thus bringing more information into the growth process. Third, the geometric rules must be shared such that they reproduce symmetry from one environmental growth to another.
  31. In comparison, how does modern urban planning work? First, we slice up the land into zones in the plan, from which we sell superblocks to land developers, which cut them up into identical lots made to settle the largest number of new residents possible. The new inhabitants have a choice between a hamburger house or cheeseburger house for which plans have already been drawn and approved by the planning authorities before they can ever have a say in how they want to live in them.The entire process works backwards, and the outcome is a backward urban morphology.
  32. How can we now invent an emergent urbanism? We have to redesign the role of each actor in the production of cities, and the rebuild the process starting from the small-scale and extending it to the large-scale.
  33. Let’s begin with the way houses are built. In the present commercial process, a builder draws up plans by guessing what the market wants, applies for a building permit for that plan, sells a fully-built home or builds as soon as he can find a buyer. In emergent urbanism, the builder should prepare construction without defining a configuration, therefore allowing the home buyer to choose the home configuration that suits him, and thus apply the building system to this configuration.
  34. In the present subdivision process, a neighborhood is divided in lots before it is even known who will live there. Even worse, it is often zoned to only a single use.In emergent urbanism, the subdivision process must be open to any shape of lots, and where new inhabitants can choose themselves where their lots should be, what size, what use they will make of it, much as was the case at the end of the 19th century. (In this example, Morningside Heights in Manhattan.)
  35. Even the neighborhood development process could be emergent. In modern planning, the comprehensive plan divides the land in blocks connected by arteries. This system costs us all smaller-scale developments that could be needed. In emergent urbanism, the modern supergrid is replaced by a flexible grid. In a flexible grid, a random neighborhood addition will connect to the existing network by providing for circulation at its outside edge. (No more closed subdivisions.) In this way, as the city grows randomly, the network adapts through cellular growth. Streets are made by the joined edges of two blocks or neighborhoods.
  36. As the city grows in scale, the traffic problems of the organic city reappear. The transportation department must then double the space at the edge of the network, creating a boulevard that ensures good circulation in future developments.
  37. Even boulevards eventually hit a scale limit, and the transportation department must double them to create an even larger bypass. This process creates a fractal urban network, adapted in real-time to the random problems of localized urban growth, but also adapted to large-scale movement. Boulevards which at an earlier stage were the city limit, become with subsequent growth the major integrators of their neighbourhoods and the central business spaces. The streets that are not direct remain quiet and thus make cul-de-sacs irrelevant.
  38. The complex grid gives us a good configuration for public space, but how should this space be designed? The modern planning method was to segregate all forms of traffic into its own space, until finally automobile space consumed everything. Despite eliminating all competition for urban space, automobile congestion kept increasing, eventually requiring extremely sophisticated traffic control systems such as the command-center-controlled traffic light.Some traffic engineers realized that traffic control was counter-productive and redesigned public space on the principle of negotiation between vehicles of all types and pedestrians. This is called Shared Space, seen here in its prototype of Drachten in the Netherlands. The main idea behind Shared Space is to remove all sign and all controls, and remake the surface to incite vehicle to be alert to their surroundings when negotiating an intersection. The final result is less wait time, more space for pedestrians and bicycles, fewer accidents and a vehicle traffic pattern that is self-optimizing, therefore emergent.
  39. You should now have an idea of what must be done to produce an emergent modern city. It is necessary to make decisions only within their context in time and space. You must also have idea of what obstacles we must face to make such changes. Every system as they exist to create the city require that the result to be achieved be defined in advance in order to obtain permits and loans, before we can determine if this is what the situation truly requires. This is the outcome of several centuries of scientific consensus, which scientists such as Wolfram and Alexander have only begun to unmake.We will thus need spaces where experiments can take place without disturbing the established order of things.
  40. We therefore come to the ultimate subject of urbanism, governance. As we saw with the case of New York in 1811, the growth of a city is not necessarily linear, but is often achieved by agglomerating multiple communities of different sizes. The solution most often invoked to this situation is to adjust the scale of governance to the entire urban area, which is to say a megamerger. The megamerger results in the same loss of community and individuality as does metropolitan-scale urban planning, but is considered necessary to tackle metropolitan-scale problems. This is not a severe problem for cities such as New York and Toronto, but when we look at an area such as Paris, which region is a patchwork of more than a thousand communities, some with a history lasting more than a thousand years, dissolving them into a monster community is extremely controversial.
  41. We can use fractal geometry as a template for our solution, in this case the Sierpinsky carpet. The area with the most severe political issues in the Paris region are the three départements (counties) of the first ring around Paris and not within the historic city of Paris. It would therefore suffice to create a community which excludes the historic center to resolve these issues. The same principle could also apply to other historic cities within the Metropolitan Paris area, such as Saint-Denis, Versailles, Saint-Germain-en-Laye, and special purpose neighborhoods such as La Défense. They could remain independent communities within the perforated metropolis. This new metropolis would only have the problem of dealing with metropolitan life at its current scale for the current time, and would leave the historic and special purpose cities to deal with their own specific issues.Without having to layer multiple levels of incomprehensible and unaccountable government on top of each other, we manage to solve problems at every scale simultaneously.Once there exists a decentralized process of merger and division, communities would naturally achieve an equilibrium with the scale of the urban area, and it would be possible to launch experiments in emergent urbanism in one of the thousands of micro-communities within the metropolitan city.
  42. To conclude, we have seen that urban planning is a response to scale problems created by spontaneous urban growth. Urban planning then creates its own inverse scale problems. Naturally complex structures, which work at every scale simultaneously, are the outcome of emergent geometric rules. Therefore, each aspect of the urban development process must be redesigned with such rules to achieve an emergent urbanism that will resolve all scale problems.
  43. Thank you all for supporting Emergent Urbanism. You may join our community at the Emergent Urbanism Network.


I think Emergent urbanism is not the same as letting slums happen. Great cities in history have emerged in the same way, and I believe that cities should grow in an organic way to be compatible with natural, economic, social and cultural environment, The natural balance that is missed in today's urban areas.

Further comment

Please send your comments by email at mthl@mthl.info, or find me on Twitter @mathieuhelie. The commenting system is closed at the moment as no measures can hold back blog spamming bots.