Complexity, to employ the definition proposed by Jane Jacobs in the final chapter of Death and Life of Great American Cities, is a juxtaposition of problems. This implies that a complex solution is a juxtaposition of solutions: fractal geometry.
How does the way we build arrive at complex solutions to complex problems without driving the builders to madness? How can we solve problems which exist at every scale in space, but also exist at every scale in time? Let's take a look at St. Paul's Cathedral in the City of London.
Let us focus on two different parts of it, the dome and the belltower. At first sight, there is nothing that a dome and belltower have in common. They are two different forms that solve two very different scales of problems. And if they had been built very far apart in two different neighborhoods of the city, one would never even associate them together. Yet in this case they are not only "dome" and "bell tower", but they are also part of a greater form we call "St. Paul's Cathedral". That is to say, their form not only solves the problem of providing a dome and a bell tower, but it also contributes to solving the problem of providing a cathedral. Several scales of solutions are juxtaposed in the same space in order to form a complex solution. How was this result achieved?
Perhaps the architect Sir Robert Wren was a genius, but intuitively we doubt that, since the geometry in St. Paul's cathedral is very similar to the baroque geometry employed throughout Europe at the time. And when we think back to how the Gothic cathedrals were built, very slowly, sometimes over more than a century, they were necessarily built by more than one architect. If they were all geniuses, then they must have been lucky to find so many geniuses idling about in medieval Europe. That sounds impossible given that medieval cathedrals appear to be even more complex than St. Paul's cathedral, even though more people worked on their construction over a greater timespan. The sublime Antwerp cathedral, for example, was built from 1351 to 1521, and never completely finished.
There has to be a key to this riddle. How did we lose the skill to make this kind of complexity?
Since Leone Battista Alberti heralded modernity (not to be confused with modernism) in architecture, and until the mid-20th century, architects spent their first days in training learning to draw the classical orders. These classical orders supposedly held the finest refinement of western civilization's building culture, having been in use since Greek antiquity and maybe earlier. It was an architect's duty to reproduce this culture by learning the orders. Any deviation would certainly cause the doom of civilization. What the orders actually consisted of were fractal nesting rules, settled on more or less accidentally through the ages. Since the abstract concept of fractal nesting would not be discovered until Benoit Mandelbrot's work in the 1970's, the orders were simply understood to be unquestionable tradition. Since they were very simple local-form rules, any architect could use them to make his building, and they could be taught to any laborer working on any specific sub-section of a building without his having to know his role in the form of the whole. They could even be used to make simulations of the building, drawings and scale models that would later be used to convince patrons to fund construction. The rules were always the same. Only the problems to be solved changed.
Let's take a look back at Wren's cathedral. What does the dome consist of? Nested structures, including columns. What does the bell tower consist of? Nested structures, including the same kind of columns. The two different problems to be solved, dome and bell tower, also happen to share the same nested problems, and when they share a solution to this problem, they become connected into a whole.
Once we are aware of this rule we no longer need a necromancer to reanimate Wren in order to build an addition to the cathedral. We can simply decompile the geometric rules and apply them to solve the new problems we face. Whatever we produce that way will belong to the cathedral as much as the original parts. But we can also extend this to the scale of an entire city. If we apply these geometric rules to build a house or an office tower, it will appear to belong as much to St. Paul's as the bell tower and the dome do. This enables us to achieve the complexity limit of urbanism. And when we look at all the great cities of the past, Paris, Rome, Venice, Amsterdam, Mediterranean hill towns, what we find is that they look whole because the builders who made them were all using the same rules in order to solve their individual problems. They didn't realize they were doing it, they were just doing it because that's how things were done.
If the classical orders were so great, why are they no longer being taught? Up to the 19th century, building technology changed very little, and so simply repeating the tradition was enough to create complexity. When metals and glass became massively affordable in the industrial revolution, architects faced a puzzle. Although the traditions succeeded at creating complex solutions, they were no longer solutions to problems that were relevant to anyone. Some architects experimented with new rules for nested structures using the new materials, more or less compatible with the old rules, and that gave us Art Nouveau and the Eiffel tower, for example. And some more radical architects, such as Louis Sullivan, said that modernity required the invention of a whole new architecture, and this became known as modernism. The modernists were right to declare the classical orders irrelevant, but in their rejection of the very foundations of architecture, the application of simple nesting rules, they also made it impossible for themselves to create complex buildings, and the result is the architectural wreck that unfolded starting in the 1930's. The worse culprits, no doubt, were those modernists like Le Corbusier and even Albert Speer (bet you wouldn't think he was a modernist) who favored abstraction and repetition in architecture. Abstraction is only the denial of complexity, the physical nature of our universe. It is the architectural equivalent of playing ostrich.
Post-modernism tried to bring back traditional forms without really giving up modernism, and that was a disaster perhaps worse than modernism was. Since post-modernists did not create nesting rules for their architecture, and on top of that were bringing up forms that were solutions even less relevant now than when they were abandoned, the result was a worldwide goofy architecture that everyone mocks as pastiche.
Some architects have been stumbling upon the right path these last few decades. The most remarkable effort has been the remodeling of the Reichstag in Berlin by Norman Foster.
The old building represented the federalist traditions of Germany, but also had to be adapted to the new philosophy of popular democracy. Foster built a glass dome from which the people can look at their politicians at work while enjoying a wonderful panoramic view of Berlin. Foster nested a new solution to a new a problem within the traditional geometry of the Reichstag, and thus created complexity that is relevant to the problems of today.
Architecture is, ultimately, just the repetitive computation of simple geometric rules to solve complex problems. Necessarily that creates complex solutions, and truly fractal buildings. With the right ruleset, anyone can do architecture, and by extension, great cities. The rules guide your hand.