« Ishmael: The Ideas of Daniel Quinn | Main

Cities and Regions as Self-Organizing Systems: Models of Complexity by Paul M. Allen

(Sorry this is late, but I’ve came down with a case of Bell’s Palsy last week, which is a non-contagious, temporary viral infection that affects the facial nerve on one side of the face, paralyzing it. So basically, I can’t really blink my left eye, nor can I more than half-smile, and I’m talking so increasingly out of the side of my mouth I could almost stand in for Vice-President Cheney. Anyway… see you in class!)

For my second chaos and complexity book, I looked at a book called "Cities and Regions as Self-Organizing Systems: Models of Complexity", by Peter M. Allen, a British ex-physicist who was strongly influenced by Ilya Prigogine before taking up spatial economic modeling. The book was published in 1997, about 10 years ago, but is still one of the main works that explicitly develops urban complexity models and theory available to most academics. (In some ways, the work is similar to, and contemporaneous, with now-NYT op-ed columnist Paul Krugman’s short 1996 lecture, "The Self-Organizing Economy", which takes a simpler look at how feedback models create spatial patterns across inter-urban, regional and national scales.)

As a general goal, Allen is interested in creating dialogue with urban planning community, particularly trying to reframe how planners approach their jobs. At one point he venting some slight frustration current planning practices, based in “make the assumption of spatial equilibrium in modeling the 'changing' spatial pattern" (43). Unfortunately, knowing as little as I do about economic modeling, much of the mathematical content of the book took place over my head. There’s a way in which applying complexity theory to the social sciences can only be ‘performed,’ rather than explained, and the best approach to understanding the kinds of process-based approaches Allen utilizes would be to recreate the models myself, tinkering and re-creating the variability. It made me think that the clearest jumping off point for Allen’s work for our class might be the excellent network illustrations that Jeremy talked demonstrated a few weeks backed during his presentation on Linked In. In particular, both modelers are attempting to graphs how network theory can apply to everyday life, creating patterns out of a blank slate, particularly focusing on how feedback loops persist over time.

(cont. below)

The important difference lies, then, in content, where this work focuses on how cities form economic connections and spatial patterns. Allen is most interested in rethinking the concept of equilibrium, specifically trying to reframe the early 20th c. modeling approaches like Walter Cristaller’s Central Place Theory (which is one of the prime places we wee the ‘rank-size’ power law rules historically developed by social scientists). As Allen mentions, an oft-cited problem with many of these earlier approaches is that they don’t allow for any kind of dynamic change (they have a rather crystalline structure), and these theories have particular trouble explaining how non-optimal situations develop, exist, and persist across the landscape.

We haven’t really talked too much about how and why complexity approaches contradict some kinds of economic theory, but this one of the questions that interests Allen: How can complexity science inform economic modeling along lines of processes-based, non-equilibrium systems that don’t necessary develop so-called ‘optimal’ strategies in which everyone ‘maximizes utility’. Instead, history and chance take large roles on the stage of urban and economic development, as notions like ‘path dependence’ and ‘historical accident,’ develop spatial patterns that persist over time through a process of positive feedback that works much like Kevin Bacon’s c.v.

The book is broken up into a three parts: 1) some theoretical assumptions, 2) a study of inter-urban dynamics, 3) some examples of intra-urban dynamics (the development of small-scale spatial patterns, like neighborhoods, and 4) an attempt to tie modeling behaviors into real-world data, predicting things like how U.S. states arrange themselves into a hierarchical pattern over time.

I’m going to briefly describe one of the models that I found interesting, particularly because of the comparison with the kind of network approaches we’ve already encountered. For example, in Chapter 3 Allen offers a model that attempts to include both economic and processes social processes to identify how spatial patterns over time might behave, in particular, seeing how stable, or path dependent they may or may not be at any given point. Like any model he makes a series of assumptions: for example, jobs create populations of people, populations of people create (a smaller amount of) jobs, choices about distance are affected by transportation infrastructure, and that there are both positive and negative effects to crowding (i.e. people are both ‘attracted’ by density and economies of scale, and repelled by crowding and pollution).

As I said, any discussion of Allen’s particular mathematical choices will have to wait until my comprehension of calculus improves, but, beginning with a empty series of points on a ‘simulation lattice,’ Allen starts stochastically developing economic connections (jobs, or ‘innovation’) across the spatial plane. And after running through the model a few times a few tendencies emerge, namely that the model displays behavior that clusters around an increasingly smaller series of ‘cities’ with differing levels of economic hierarchy (across the four ‘economic functions’ that Allen includes in the model). The structure ‘solidifies’ around five or six economic centers, that are linked across space like a topographical map. In particular, the model displays path dependent behavior, irreversibility, and self-reinforcement, so that, in other words, if one were to try to step the urban lattice back in time (by, say, increasing transportation costs in a situation like an oil shortage), it would not revert back to its earlier form but instead develop new tendencies.

It strike me that the point of creating a model like this, for Allen, is to start to engage with the idea of 'stability,' or what Luke and Jeremy called “network resiliency?, using models like this to, rather than predict specific outcomes, attempt to develop an understanding of how and why certain kinds of government or economic intervention (perturbations in the system) may make larger differences at certain points than other interventions. This, finally, is where Allen’s point really come across. He’s speaking to planners about how and why some tactics are better then others: “what is clear, however, is that he whole question of successful planning and intervention within urban systems must depend on a knowledge of the stability of the structure that is evolving. Without this knowledge many well-intentioned projects will prove abortive, as the system 'unexpectedly fights back'" (58)

In my mind, the goal of this second reading (for this class) was to start applying the diverse scientific context that we’ve gained access to throughout the semester to a particular subject. Many people at the beginning of the semester expressed an interest in seeing how complexity and chaos principles may apply not just to physical, but social systems, and cities have always been (in my mind, and in historical literatures) one of the most obvious ways of describing and visualizing emergent, complex human behavior. Patterns like traffic flow, pedestrian activity, housing developments all share this double-life where decisions made by individuals for their own benefit end up creating systemic patterns and activity on larger neighborhood, regional, or global scales.

I was happy, therefore, to find Allen making some of the same conclusions about his research that my favorite writer, Jane Jacobs, made in her 1960 (polemical) work "The Death and Life of Great American Cities" (which I’d mentioned at the beginning of the semester). Allen ends up arguing that issues of causality probably exceed the limits of what social scientists, economists, and urban planners can hope to know. Instead, we need tu study complex social systems to try and gauge their systematic behavior, how flexible, resilient, adaptable, and changing they are. These rules become principles for institutional, social intervention, as Allen writes:

“In general, then 'small and diverse' will allow for adaptation and change better than the 'large and monumental.' Just as central planning failed because of its rigidity in a changing world, so large monolithic organizations and plans will tend to be unsustainable. The lesson seems to be that the plans which encourage variety and diversity in the inhabitants of a region, in their activities, their means of transportation and in the landscape, tend to lead to relate and adaptive systems capable of generating their own development and in responding to the challenges of the economic, natural, and social environment" (252).

This is very similar to what Jacobs had to say about large-scale interventions in urban systems, and why, for example, large public housing projects were probably destined to fail. I’m left, at the end, with a desire to think through how complexity and chaos theories may have implications within the political realm: How, for example, does allowing social systems to have ‘a life of their own’ reinforce or limit libertarian or liberal ideologies? If we cannot specifically predict anything, but can only develop knowledge of tendencies, how should public policy approaches be reframed? Might complexity thinking actually end up being used to reinforce large-scale, and largely static, market-based economic power structures? I’m afraid, however, these are questions for another semester. It’s been nice getting to know all of you.

Bill Lindeke