Self-Organization and the City

Order this book

Reviewed by
Alasdair Turner
Bartlett School of Graduate Studies, University College London, UK.

In Self-Organization and the City, Portugali presents within a unified framework the development of many agent-based models over the years. The major part of the book details his experiments using 'FACS' or free agents in a cellular space (agents which are free to move and settle over a cellular-automata base). However, as a whole, the book is about much more than experiments on spatially located agents. Do not mistake 'self-organization' in the title for a general interest about complexity and emergence in agent-based systems. Portugali is endeavouring to present a theory of cities as described by the 'formal mathematical' theory of self-organisation. Whether or not he succeeds is open to question. A familiar challenge to Prigogine can easily be levelled at Portugali's book: what exactly does casting cities as self-organising systems explain? Indeed, his persistence in trying to force the idea of 'city' into a paradigm of self-organisation results in paradox. Portugali denies causality, yet the FACS games he plays are based on causal events; he denies the ability to plan cities, yet looks for a way to plan them. His persistence also tends to grate with the reader, as the eulogy to self-organisation becomes repetitive and turgescent. Nevertheless, despite its flaws, the book is an intriguing testament to Portugali's inventiveness. Even where the experiments described are now well known, Portugali still manages to find a new interpretation or an ingenious explanation.

The book is broken into five parts.

Part I On Cities and Urbanism begins with a tour of various views of 'city' and 'urbanism' from the literature. Within this context, Portugali uses a philosophical discussion of semantic networks to propose his own notion of 'city': the inter-representation network (IRN) city. The IRN city is a complex of internal (semantic) networks and external (environmental) networks, influenced by Giddens' (1984) theory of structuration. The IRN city, claims Portugali, is a self-organising city, and the brand of self-organisation he favours is Haken's (1978) 'synergetics'. This enfant gâté is introduced via Haken's classic example: the laser. Haken calls the resonant light wave of a laser the order parameter, to which the individual atoms releasing energy to add to the light wave are enslaved. The attendant mathematics is not introduced at this stage, but the foundations for examining IRN cities in terms of the vocabulary of self-organisation have been laid.

Against this background, Portugali presents free agents on a cellular space (FACS) as a tool to implement IRN cities. The FACS method presents a structural basis in the form of a cellular automata, to which a layer of agents exhibiting cognition is added. The structural basis acts as the environment (the network of external representations) and the agents as the individuals within it (the network of internal representations). Portugali discusses how FACS cities may produce various phenomena related to self-organisation, how they may resonate with different order parameters, how they may bifurcate from the old to the new, how a multiplicity of states may coexist within a single city. It is a heady mixture, and within it the distinction between theory and speculation is quickly lost - no matter, however, FACS cities offer an enticing investigative tool.

Part II City Games describes a set of 'games' played using FACS cities, labelled City-n, where different n investigate different IRN properties. The simulations are distinguished as 'games' as they do not bear a direct relationship to reality. Of course, there is a problem that, isolated from the phenomenon of city, these city games become mere games, played in isolation of science. Portugali's defence is that self-organisation cannot be predictive, that the models are heuristic (seeking to discover) and hermeneutic (seeking to explain). But the question remains, seeking to discover what, seeking to explain what? Since the games are unpredictable, anything the games may tell us can change from game to game. The most we can learn from such systems is that there is a range of outcomes. Now in fact, the results do often show macroscopic predictability (or we must assume they do) as Portugali gives representative outputs from his experiments in his diagrams, showing the value of some variable at locations on the cellular automata grid. These representative samples must of course be giving human-recognisable patterns for them to be representative. Thus, when Portugali tells us that such and such an order parameter is emerging, it is emerging in a predictable manner. However, this is not a criticism of the games themselves, but of the philosophical import - the games themselves are intriguing.

City[-0] is a thorough investigation of 'blue' and 'green' segregative agents on a cellular automata. In the course of the investigation, Portugali explains notions of stability and instability at local and global levels, and draws out a new principle to insert into Haken's synergetics. City-1 shows how combinations of local parameters can affect global outcomes, while City-2 and City-3 look at the effect of changing either intention or behaviour in order to avoid the 'cognitive dissonance' of a gap between the two; the results are examined in the context of actual changes in attitudes of Jewish inhabitants to having Arab neighbours. City-4 goes further in that it encodes memes for individuals (where in this case a meme is a genotype that might represent 'cultural identity' for example). The results are interesting in that coexisting stable meme populations arise (or demes - see Wright 1969), while the individual memes remain unstable. City-5 shows that the FACS framework does not have to be based upon traditional cellular automata, but may use GIS as the locational input, thus creating a very similar idea to O'Sullivan's (2001) graph cellular automata. City-6 returns to the first City implementation, but imposes hierarchical familial relationships between the agents, so they act according to groups and subgroups. Different 'cultures' are prescribed different hierarchical structures, and the results analysed using Q-analysis.

Part III Self-Organizing Planning takes on a major theme from the philosophical standpoint of the book: how can you plan the unplannable? Of course, as explained before, we do have to take it that cities really are unplannable. If so, though, Portugali suggests that planning should be 'just-in-time' rather than 'just-in-case'. The just-in-time planning should however consist of actual plans, albeit short-range and micro in scale, rather than letting the postmodern milieu take over. Thus, the overall plan is allowed to emerge fromthe system, growing the infrastructure as it goes - along the lines of building the paths where the people actually walk (Helbing et al. 1997, suggest a technique to model such a process). Part III also considers planning games, which allow the planner to 'experience' different planning outcomes. Again, as previously pointed out, it does appear fruitless to do this. If the games really are chaotic, and self-organisation results in different order parameters based on initial conditions, then planners may actually see anything (or at least, a the results of a number of order parameters) develop before their eyes. By changing a single lamppost, they may end up changing not just the entire microscopic course of development of the city, but also its macroscopic order parameter. However, once again, Portugali's demonstration game does show macroscopic similarity between game runs (p. 257); perhaps planners may be freer to place lampposts than they thought.

Part IV Synergetic Cities includes the only serious mathematics in the book. It begins by applying Haken's pattern recognition approach (used successfully in a synergetic computer) to levels of populations within the city worlds. It is good to have a taste of the type of solutions that might be applied directly to non-linear differential equations for FACS-style evolutions. There are problems, such as the states (cultural identity, etc.) must currently be unambiguously labelled whereas the later FACS cities allow evolution of new population states (indeed, it must be assumed these states are identifiable), and as Portugali observes, the current mathematics does not include inflow and outflow of populations. The book then goes on to consider decision-making in city planning as a pattern-recognition task. It is exciting material to read. The work of the decision-maker is considered as a synergetic inter-representation network (SIRN), that is, an interaction between the individual (internal) mind and the (external) collective memory of the city which produces plans. Unfortunately, it is a shame that Portugali does not consider the SIRN in the wider context of other ideas concerning collective memory and society/city, for example, those of Halbwachs (1992) and Rossi (1984), which would have added some depth to the discussion. There are also caveats to the method. To what extent is decision-making simply applying schemata? Does such a method allow creative solutions? And so on. That said, the SIRN approach leads to fascinating experiments with multiple 'planners', currently in preparation by Portugali.

Part V Self-Organization and Urban Revolutions is a brief section on connecting self-organisation principles to the idea of urban revolutions, in relation to the work of Childe and Lefebvre. In it, Portugali proposes that trying to identify 'cause' should be abandoned, since the trigger of a revolution may be the proverbial butterfly. It is a grand claim, and apparently flawed. 'Cause' may cover many things. Although the butterfly may be the trigger, the group of prevalent 'causes' should be taken into account. Portugali even does this himself, when he investigates, for example, the effects of the immigration of Jews to Israel from Russia in City-1. The immigration wave is a cause, while Portugali seeks to find the effect of such a cause. Thus, claiming that there are no 'causes' is disingenuous; the point is really that hunting the trigger may be impossible.

In conclusion, Self-Organization and the City is a stimulating book, a collection of various strands of Portugali's work, with many ideas poking through a grandiloquent baldachin. If you are considering the implications of formal self-organisation to agent-based systems, then it is imperative to read it. If you are interested in the interaction of societies and space, there are sufficient original 'perceptions' to make it worth exploring.

---

References

GIDDENS A. 1984. The Constitution of Society: Outline of the Theory of Structuration. Polity, Cambridge.

HAKEN H. 1978. Synergetics: An Introduction, second edition. Springer-Verlag, Berlin.

HALBWACHS M. 1992. On Collective Memory. University of Chicago Press, Chicago, IL.

HELBING D., J. Keltsch and P. Molnár 1997. Modelling the evolution of human trail systems. Nature, 388:47-50.

O'SULLIVAN D. 2001. Graph-cellular automata: a generalised discrete urban and regional model. Environment and Planning B - Planning and Design, 28:687-705.

ROSSI A. 1984. The Architecture of the City. The M. I. T. Press, Cambridge, MA.

WRIGHT S. 1969. Evolution and the Genetics of Population, Volume 2: The Theory of Gene Frequencies . University of Chicago Press, Chicago, IL.

Return to Contents of this issue

© Copyright Journal of Artificial Societies and Social Simulation, 2002