Can there be Science of Cities?

This was the question behind a recent conference at the Santa Fe Institute.  Geoff West , Luis Bettencourt, and Jose Lobo, who are well-known for their work on scaling laws for complex systems including cities, were the hosts.  My own response to the question was:  “I think I know what ‘of’ means, but ‘science’ and ‘cities’ are both difficult concepts.”  And indeed we struggled with both of these words for 2 1/2 days.  What would a science of cities do? and What is a city anyway?  Over the coming weeks we will have several participants from the conference reporting their views on what was overall an extremely simulating event.  Here is my personal overview and conclusion.  You can find the agenda and participants on the conference wiki  and the talks themselves are here.


The two key words – science and cities – were the focus of much of our debates.  It is well-known that there are may definitions of what constitutes a city.  Is it the legal territory or the boundaries of the transportation systems or the integrated economic area or the dense urban core and its hinterland of suburban and rural communities?  How big is a city? is a tricky question.  Even within a given area, there is a related question of scale – what is the appropriate scale for understanding the city’s economy, its flows of traffic, its public health, its crime, its water supply, and so forth.  Michael Batty of the Centre for Advanced Spatial Analysis at University College London gave a very perceptive talk via Skype that illustrated this and the often fractal nature of development as a city grows.

A science of cities should be expected to be both descriptive of known cities as well as predictive of the properties of new or unknown cities.  To large extent the work on scaling laws provides very impressive abilities to predict levels of income, the number of Thai restaurants, levels of crime and many other aspects of cities within a given national economy.  But this is very much an economist’s or a statistician’s view of a science of cities.  It tells us little about the evolution of any specific city.  For example even though the scaling laws are highly accurate in predicting changes in many attributes, such as numbers of patents issued, versus the size of the city, they do not tell us how the number of patents issued in a specific city will evolve as the size of that city grows or shrinks.  Moreover, this approach is too remote from how the city actually works to help us with understanding and diagnosing “good” or “bad” cities.  The scaling laws cannot us tell how water is used, where it comes, from and where it goes to.  Nor can they tell us how the inhabitants makes decisions about their use of the city’s facilities.  There is certainly room in such a science for the statistical view, but I came away still believing that we also need a field science, just as Jane Jacobs insisted on street-level observation as her source of truth.  which brings us back to the question of scale.

Two experimental methods at widely differing scales were reported by Tim Gulden  and Carlo Ratti.  You may recall the article entitled “The World is Spiky” written by Richard Florida in response to Tom Friedman’s book “The World is Flat”.  As one of Richard Florida’s graduate students at the time, Tim was responsible for generating the work maps with those spikes, which are derived (in a rather tricky way) from satellite images of Nightime Light.  What Tim has since demonstrated is that Nightime Light is in fact a very good proxy for economic activities of a “city” or region.  Indeed the boundary of the illuminated region is itself another definition of a city.  At a much finer level of scale, Carlo Ratti, director of the MIT Senseable City Lab, showed how social networks can be revealed from the Call Description Records produce whenever we make a telephone call and also that there are is a highly degree of locality in these networks.  If fact they are also yet another kind of definition of a city.  Smart Cities in general are providing many new forms of instrumentation often at a very fine granularity.  Michael Batty’s talk included a range of scales and he spoke of the Seven Laws of Urban Scaling; although he did not speak of it at this meeting, Michael’s group does a lot of analysis of individual journey segment in the systems of Transport for London.  

My own hopes for a science of cities would be a mapping of how information flows within the city, how it is communicated – sometimes in digital networks and sometimes leaping between individual inhabitants who may not even know one another.  How do these flows affect the creativity of a city?  Its ease of use?  Its “discoverability”?  The effectiveness of its transportation systems and its public safety?  Ultimately I would like to understand how the inhabitants make their decisions about how to use the many services that the city offers.  We can glimpse some of this, not least in Michael Batty’s work, but much of it remains hidden.  I compared the fascinating work on scaling laws to trying to understand human physiology by studying public health statistics.  One can learn many interesting things from public health statistics, but it will never get us to the level of physiology.

Nonetheless, this was a fascinating experience, and our debates could have gone on much longer.  In the coming weeks we will publish a series of articles in this blog on more specific topics with the very broad range that we covered and I hope that you will see their relevance to the work of the Urban Systems Collaborative.


Colin Harrison