How will we define cities in the next century when most populations will feel the inﬂuence of globalization?
In summer 2002, I was driven from Nanjing to Shanghai through Suzhou, a distance of about 250 kilometers. I was struck by the fact that we journeyed through almost continuous urban development where the countryside was turning into a new kind of city: a “desakota,” as Terry McGee once coined the term, a mixture of town and country—an urban sprawl indigenous perhaps to China itself, but symptomatic of the fact that by the end of this century, nearly everyone will be living in one kind of city or another.
At the time of writing, I have just retraced my steps but this time simply traveling from Suzhou to Shanghai. Nevertheless, I remember Suzhou in 2002 as a town of perhaps 1 million, but still enough of a village built around canals in its core—this is the Yangtze delta region—reminiscent of the old China. In the intervening years, it seemed to me that Suzhou had reached 2-3 million, but I was told quite ﬁrmly that Suzhou is now 11 million and rising. Simply add that to the Shanghai municipality population of 23 million and you already have more people living there than in any of the large city agglomerations that are currently ranked number one in the world, such as Mexico City or Tokyo, which are said to be about 25-30 million.
My journey was through a remarkable urban sprawl that was more a series of adjacent and linked clusters of high-rise development all running into one another, rather different from anything that resembles Phoenix or Atlanta, the archetypical examples of sprawl in America. In many cases, this is the Chinese city of the future, just as the continuous urban development that we have in western Europe at much lower densities and with very little high rise is our future. London is just over 8 million people; Amsterdam is 1 million. The wider London region is at least 15 million, probably nearer 25 million, while in the Netherlands as a whole it is simply impossible to draw city boundaries for a population of nearly 17 million, which in any case spreads across the border into Belgium.
Defining a City
In an age when we will all live in cities, defining city size will be entirely problematic, as it has already become in many parts of the world. What this means for some of the seemingly long-lasting signatures of urban development such as our city-size scaling laws is unclear. If I can define a city like London as 8 or 15 or 25 million, then we can manipulate the rank-size curve in such a way that it no longer has any meaning with respect to the organization of the hierarchy of cities.
When we are all living in cities, there will of course still be a distinct hierarchy because this is the signature of the competition between cities which, in fact, appears to be becoming even more intense. But physical form is what we have used for many years to define the city, and this is no longer the measure of what a city actually is. If it takes 40 minutes on the bullet train from Suzhou to Shanghai, then where does Shanghai end and Suzhou begin? If you look at the top-50 cities in terms of the rank of their population over the last 2,500 years, it is Suzhou that has been in the top rank for more years (2,150) than any other city; although in terms of the current population rankings (11 million), it left the top 50 over 150 years ago.
Clearly it has now returned, but the fact that it is not recognized as such is simply an admission that the current ranks bear no relation to what is happening on the ground. It is an admission too that we can no longer define city boundaries with any certainty. I can now draw “plausible” boundaries that would easily put the Shanghai-Suzhou at the top of the hierarchy and London in the top 20, but there are many plausible boundaries. That is the problem.
So, how do we define a city? And how will we define cities in the next century when most populations will feel the inﬂuence of globalization, for this in essence is one of the major reasons why we can no longer define the extent of a city?
Indeed, it is tempting to think of cities being composed of networks of linkages across the world that reﬂect trade, social contacts and even knowledge that has become global through access to information on the Internet. The “cloud” is recognition that information is global in its import and that it no longer matters where it is physically located.
Much of what we now do does not relate to the place we inhabit which, in some senses, is becoming independent of the activities that support and sustain us both economically and globally. Networks are, of course, the icon of our age and the real challenges in science are related to how multiple networks are coupled together in diverse ways, interacting with one another through diffusing processes. Networks of genes, email, brains, migration, computers—you name it and it will have some network characteristic—are all being studied in new ways, and cities are no exception.
A Sense of Place
But there is a difference in that our sense of place is not really network-based. Although most places and the activities and populations within them now depend on others, once we recognize that such networks are spatially disjointed, we lose the recognition that pertains to clusters that we have defined as cities for the last 5,000 years. The challenge, therefore, is to ﬁnd characterizations of cities—or rather urban clusters—that span large distances and times, while at the same time rooting these notions in the traditional idea that places are defined locally by contiguous conﬁgurations of development. In short, we need new ways of linking local to global.
London is just over 8 million people; Amsterdam is 1 million. The wider London region is at least 15 million, probably nearer 25 million, while in the Netherlands as a whole it is simply impossible to draw city boundaries for a population of nearly 17 million…
The usual way has been to aggregate the populations into small zones that provide enough variation across a city as conceived of as a contiguous physical development to examine its spatial heterogeneity. However, spatial contiguity is broken in a global world, for populations depend on networks related to all the activities they engage in—work, social interactions, entertainment, education, health and so on. Plotting networks at an individual or more aggregate level is difﬁcult, and there is no way of really reconciling this variety to represent it in a spatial context.
In short, we have not yet come up with aggregate varieties of network that portray variations across space (and time) in a sufﬁciently robust manner as to provide a good summary of how cities are now composed of multiple populations and activities that depend on a continuum of linkages from local to global.
We still need to show this continuum in one place at the level of the physical city, and this probably means we need to map the phenomenon of the future city in the same way as we have done traditionally. One simple way would be to ﬁx on some local—global set of linkages and provide an average distance—a kind of global accessibility for every population in every space. This might literally be measured as a distance or reach that is an index of how the activities of each population depend on other activities and populations at a distance from the location of the population in question.
To make this idea operational, we really need a local—global footprint for every member of the population and this might contain several sub-footprints dependent on the particular activities that an individual is engaged in. For example, someone working in ﬁnancial services might be supplying activities whose demand is truly global and whose inputs are equally global in terms of the inﬂow of cash to support the product. This would be in contrast to a worker providing services locally but whose inputs also contained some measure of more global activities.
In short, each individual in terms of work would have a footprint determined by the spatial extent to which their activities depend on inputs in different places, as well as providing outputs that determine consumption and production at different places. The location of these places would determine the footprint in terms of some measure of distance or accessibility.
This, of course, is no more nor less than a highly disaggregate spatial input—an output model that is a set of accounts related to where the inputs and outputs for any individual in the population come from. However, such models have rarely been examined spatially in the past, largely due to lack of data, but we may be able to get somewhere for aggregate populations by making assumptions about where their work originates from and is destined for in terms of a global measure of distance.
For example, we might portray the footprint of individuals from their work perspective as containing different percentages of local, regional, national, continental and global reach by noting what they produce and applying a national estimate to each in terms of their work location. We could then add up these values for aggregate populations living or working in each zone of the city and produce a composite index by weighting the values according to an average distance measure for each of these ﬁve scales and summing them.
The larger the score the greater the global reach of the aggregated zonal population, and plotting these would give a sense of the diversity of the city in terms of work and residential locations with respect to the extent to which the city was globalized. If one had this type of measure for several cities, one could then ﬁlter the data for each level and produce composite mosaics of the global, continental, national, regional and local city.
To launch this idea, we need data on the precise locations where the inputs and outputs of products produced by an individual are actually located. Let us first define the footprint for an individual in terms of the products which that individual consumes or produces. These inputs and outputs can be summed over the relevant set of locations in zone in each city and then weights applied to determine a composite index for each input and output. This would give the relevant input or output reach for particular products in the aggregated zone.
Now imagine collecting all the zones in all the cities for each input and output. We can then construct a spatial proﬁle of the typical city input or output with respect to its global reach by aggregating values between different ranges of reach. We can do this for all inputs, all outputs, for inputs and outputs, and for these different aggregations with respect to work or residence. We can in principle take the zones from each city that satisfy these different ranges and map these as mosaics or composites but, frankly, this has little meaning. A much better plot would be to look at the pattern of cities as some sort of frequency distribution, rank ordered by the number of zones from which it would be easy to see patterns of city types.
We urgently need a new definition for cities that goes beyond physical extent, one that takes into account how cities ﬁt into the global picture which is built from what we know about networks at every scale.
But we also need to retain a measure of spatiality and ideally adjacency, as the way we interpret cities is still strongly physical. A sense of place is essential and, even if cities are composed of many parts all spatially distinct in terms of their global reach, our earliest and most formative inﬂuences are likely to be local rather than global.
*[An earlier version of this article was originally published by Environment and Planning B: Planning and Design.]
The views expressed in this article are the author’s own and do not necessarily reflect Fair Observer’s editorial policy.