Try this for size: imagine a city with a population of ten million people that has widely deployed renewable energy sources, that has implemented micro-markets for consumer-produced energy, and let’s say that these micro-markets are cleared every ten or fifteen minutes. Imagine further that this city has a high penetration of plug-in electric vehicles (PEVs), many of which need re-charging every day for local commuting. Finally suppose that the electricity distribution networks have three tiers of interconnection from high voltage to medium voltage to residential voltage. Each node in this distribution network has on average four peer connections, and serves 1,000 consumers, hence there are about ten thousand nodes. So, every fifteen minutes the flows of electricity in this network have to be re-balanced to meet the micro-market contracts and to re-charge the PEVs under the constraints of link capacities and with enough safety margin to avoid knock-on failures if a link fails. Roughly speaking,  this involves re-balancing around one million flows every fifteen minutes. We could also add energy production and consumption forecasts every fifteen minutes based on cloud movements (real clouds, not IT clouds).

That’s what I would call a complicated problem. Quite a challenge to get your head around, but well within the abilities of our data jockeys, especially when we consider that there will be a large degree of day to day similarity. We can solve this problem. Try something a bit harder.

Imagine a city in which some hundreds of thousands of people live in slums. For example, Kibera in Kenya. Suppose that a well-meaning agency wishes to offer these slum-dwellers a higher quality of life and constructs a new suburb offering low-cost apartments, roads, schools, police, water, sewage, and waste services. Building such a new suburb is business as usual. What could go wrong?

A lot, as it turns out. What is lost in building this new suburb are not only the layout and functions of the existing infrastructure, but above all the social networks or infrastructures that enable a young child to walk safely to buy rice. Or the arrangements a mother has made to have her son looked after by a neighbor while she goes to work and the reciprocal services that she provides in turn. Modern housing is in principle a good thing, but what needs to be relocated is not only the people, but the Social Ecosystem they have constructed. In Kibera many people decided to return to their former homes.

While the flows of energy in a utility network, as imagined above, may number in the millions, the interconnections within such social networks may be an order of magnitude or two higher. Worse, they are largely invisible, even in a “smart slum”. Our data jockeys have nothing with which to work. I will call this a ‘complex problem.’

In the first article in this series, I argued that cities are complex, urban ecosystems that exist at multiple spatial and temporal scales and that do not permit the kinds of decomposition or systems engineering on which technology is based. Because of this, until we have a deeper understanding of what really makes the city a living entity, our progress on smart cities will be inherently superficial and of limited impact. My interest from the perspective of smart cities lies in how a city works at time scales of seconds to days, at spatial scales of a meter to 100 km, and at the social scale of individual people. How does the operation of the city – what people do and the decisions they make – relate to the Natural Ecosystem, the Infrastructure or Built Ecosystem, and the Social Ecosystem?  I do appreciate that longer time scales and larger spatial scales must also be considered.

In this article I ask: if we want to develop an understanding of how cities work, how would we go about this? The following are some approaches known to me for developing such understanding.

Geoff West, President and Distinguished Professor of the Santa Fe Institute, and Luis Bettencourt, director of the Mansueto Institute for Urban Innovation at the University of Chicago have applied their research on scaling in biological systems to the scaling of urban systems. These Scaling Laws emerge from Complexity Theory and represent the exponential scaling of some factor, for example energy consumption per capita, over cities with population sizes varying by several orders of magnitude. Studies of cities in the United States show that some factors exhibit sub-linear scaling, such as an exponent of 0.85, for example, energy consumption per capita. Others exhibit a super-linear scaling, such as an exponent of 1.15, for example, GDP per capita. These results imply that there are common network structures among cities and offer predictions of, for example, the number of kilometers of streets or the number of violent crimes to be expected in a city of a given size. This work is wonderfully descriptive of the evolution of urban ecosystems, but I feel it is too distant from what happens among individual people to provide the theory of a city I seek, although it should provide strong constraints on the predictions of such a theory.

Michael Batty, Emeritus Professor of Planning at the Bartlett Institute of University College London, has led work, also emerging from Complexity Theory, that provides theoretical support for the development of street plans in cities. Such patterns of urban development reflect the Natural Ecosystems, such as topography, as well as the Infrastructure Ecosystem, such as existing transportation systems, and the Social Ecosystem, such as property prices. While it is fascinating to understand how specific cities have evolved, the time scales here are remote from smart cities, but again can provide constraints on predictions from a theory of smart cities.

Geoff West also likes to ask: “When will we see the Hamiltonian of a city?”  A Hamiltonian can be a representation of the total energy balance of a closed system, possibly a city. It is most valuable in studying systems with many degrees of freedom, which can ultimately become chaotic. In the context of cities, unlike physical systems in which energy is conserved, the Hamiltonian of a city might represent the simultaneous conservation of various factors, some of which may be fungible, such as energy, environmental quality, material, knowledge, wealth, and others as agents, including individuals, organisations, built infrastructure, and natural infrastructure go about their daily lives. This is a deeply scientific approach, but I fear that today it lies beyond our abilities.

This latter approach might be likened to the concept of an economic equilibrium in which actors seek to maximise their individual satisfaction in competition with one another. Equilibrium is achieved when no individual actor can improve his or her satisfaction unless one of the others changes his or her strategy. Satisfaction here may relate to wages, to access to capital, to commuting time, to habitation, to sexual partners, to environmental impact, or any another need that involves competition. It implies rational and irrational decision-making on a very large scale. It is reasonable to assume that a real city normally operates far from such an equilibrium and to hope that, by introducing smart technologies to increase the flow and decrease the latency of information, the city may come closer to such an equilibrium. Nonetheless, beyond those fairly large caveats, this has the feel of a pragmatic approach for which we are likely to be able to collect data. The insights from applying this and other tools from Game Theory could be in the form of the strategies employed by individuals, groups, and organizations to maximise their satisfaction. With these insights, we could then begin to ask at various spatial and temporal scales how smart city technology could reduce the “friction” that impedes the approach to equilibrium. I am not aware of anyone adopting this approach, but would be delighted to hear of such work.

While the latter approach may lead us to theories for the Infrastructure and Social Ecosystems, it does not sufficiently involve the Natural Ecosystem. We will need also to understand the dependencies of the city on Ecosystem Services and the impacts produced by the city on the Natural Ecosystem, which is itself highly complex and involves both short and long time scales and local and global spatial scales. Marina Alberti, Professor of Urban and Environmental Planning at the University of Washington, has written about applying ecosystem methods to the evolution of cities in her book “Cities That Think Like Planets”, which I am currently trying to absorb.

Discovering this understanding of cities is a grand challenge and may take as long or longer than developing a complete understanding of the human body. In fact, since cities are becoming the platforms for almost all of human life, it will be a Theory of Everything. This is hard to contemplate, just as a century ago it must have seemed impossible that we could ever understand the human body. Although I argued in my previous article that mountains of data and powerful analytical tools are not a sufficient substitute for understanding, they are a necessary resource. Just as medical science has benefitted from dramatic advances in tools to study the human body, so we may hope that IoT and smart city technology will provide the raw materials for a science of cities.