Growing interest in the study of networks is an opportunity to question the very nature of geographic space, the most fundamental yet least well-defined concept in our discipline.
Networks in geography
Networks are all the rage in geography. In recent years, a growing number of geographers have been seeking to better understand the relationships between geographical space and social networks.
This spatialization of social networks has recently taken a more formal turn, with the development of spatial social network analysis or SSNA.
This exciting field of research analyzes the extent to which spatial elements, such as urban density, transport infrastructure or international borders, affect social relationships, and vice versa.
It is different from actor-network theory and most qualitative approaches developed by the relational turn in the social sciences.
SSNA postulates that there is a fundamental relationship between our tendency to live and work close to each other, which geographers call propinquity, and our tendency to interact with others who share common attributes, or homophily.
To use a well-known analogy in network science, birds of a feather flock together, yes. But because they roost in the same locations and follow the same migration routes, they may end up with similar feathers.
This short essay suggests that the most productive way to benefit from recent advances in network science is to develop a common language that builds on the core concepts of both geography and network analysis.
Space subdivided
Thus far, the most popular approach to spatialize social networks is to identify several types of space. Since space is socially produced, as Doreen Massey rightly argued, some geographers have argued that we should compare “absolute” space with “relational” or “network” space.
This approach is used in medical geography, where the spread of a disease in a population can be explained by neighboring locations and network neighbors.
By comparing the density of social ties between people spreading the disease in both dimensions, one can determine whether physical or social proximity is more prevalent.
While the idea of dividing space into different types is appealing, there is no consensus among geographers as to how many types there should be, what they should be called, and what their inter-relationships are.
For Nigel Thrift, geographic space can be either an empirical construct, a set of connections, a representation, or a place. On the other hand, for George Hepner and Richard Medina, there are five categories of space: physical, social, virtual, perceptual and hybrid.
Space undivided
An alternative approach is to consider that one learns less about geographic space by dividing it than by investigating how its different components are socially produced, notably through social networks.
The main source of inspiration for this idea comes from the work of Denis Retaillé, who introduced the concept of mobile space to the geography literature 30 years ago.
Unlike most geographers inspired by the relational turn, Retaillé did not ask himself what kind of space was produced by the mobility of West African nomads, traders and jihadists. For him, space as a whole is mobile.
Instead of considering movement as a flow between fixed locations, the mobile space approach argues that movement is the organizing principle of geographic space.
In this respect, this theory overturns the paradigm of spatial analysis, based on the primacy of fixed points, lines, and areas, from which the whole structure of space is derived.
Retaillé’s approach has led me to envision that the geographer’s job is to understand how social networks interact with spatial features to produce the cities, borders, and markets we experience on a daily basis.
Only when we clearly understand what our “fundamental stuff” is produced, and made of, can we envision how geographic space as a whole interacts with other human dimensions, particularly social networks.
A common language
The next step is to find a common language between geography and network science. In an unpublished article, Steve Radil and I described the possible foundations of such a dialogue.
We argue, for example, that the two definitions of place used in geography have a correspondence in network science.
The way we think of places as unique localized settings connected to other places – ‘places in networks’ – is described in network science as centrality, or the structural importance of a node.
Conversely, our definition of places as a location made through people’s actions – ‘networks in place’ – corresponds to groups of densely connected nodes, or communities, for network scientists.
Similarities can also be found between our concepts of distance, scales and power, and those developed by network scientists.
Importantly both networks and geographic space are not static – they can grow change and shift. Understanding such changes is currently under way, through dynamic social networks analysis, or DSNA, which we used to model conflicts in Africa for example.
Next steps
Although encouraging progress has been made in recent years, the spatialization of social networks remains an unfulfilled task.
In this relatively new field of research, technical approaches aimed at quantifying networks are still more represented than conceptual reflections on the nature of geographic space.
In this respect, spatial social network analysis reflects a more general evolution in our discipline, marked by a relative dearth of theorization of space in favor of the ever more advanced use of analytical tools.
Instead of dividing geographical space into several types, more effort should be made to understand how our most familiar concepts can be reinvigorated by the relational focus of network scientists.
Ultimately, the growing focus on networks in geography should lead to more theorizing of space in our discipline.
Acknowledgments
I would like to thank Steve Radil and Lacey Harris-Coble for their comments on an earlier version of this blog. Photo by Honey Yanibel Minaya Cruz on Unsplash.
By Olivier Walther, 4/3/2025.