CityReads│Do we all live in “urban villages”?
Markus Schläpfer, Luís M. A. Bettencourt, Sébastian Grauwin, Mathias Raschke, Rob Claxton, Zbigniew Smoreda, Geoffrey B. West, Carlo Ratti,2015. The scaling of human interactions with city size, J. R. Soc. Interface, 12 (103) Source: http://senseable.mit.edu/urbanvillages/
Do you thrive within the crowds of a large metropolis or do you prefer the close community spirit of village life? Analyzing mobile communication data reveals how a person’s social network changes when moving from a small town into a big city. A team of researchers from both the MIT Senseable City Lab and the Santa Fe Institute worked with Orange Labs, British Telecom and Raschke Software Engineering to study how cities affect our social relationships.
The statistical relationship between the size of cities and the structure of the network of human interactions at both the individual and population level has so far not been studied empirically in detail. Early 20th century writings suggested that the social life of individuals in larger cities is more fragmented and impersonal than in smaller ones, potentially leading to negative effects such as social disintegration, crime, and the development of a number of adverse psychological conditions. Although some echoes of this early literature persist today, research since the 1970s has dispelled many of these assumptions by mapping social relations across different places, yet without providing a comprehensive statistical picture of urban social networks. At the population level, quantitative evidence from many empirical studies points to a systematic acceleration of social and economic life with city size. These gains apply to a wide variety of socioeconomic quantities, including economic output, wages, patents, violent crime and the prevalence of certain contagious diseases. The average increase in these urban quantities,Y,in relation to the city population size, N,is well described by superlinear scale-invariant laws of the form Y=Nβ ,with a common exponent β ≈1.15.
Recent theoretical work suggests that the origin of this superlinear scaling pattern stems directly from the network of human interactions - in particular from a similar, scale invariant increase in social connectivity per capita with city size. Such conjectures have not yet been tested empirically. In this paper, we explore the relation between city size and the structure of human interaction networks by analyzing nationwide communication records in Portugal and the UK. The Portugal data set contains millions of mobile phone call records collected during 15 months. The UK data set covers most national landline calls during 1 month.
With respect to Portugal’s mobile phone data we demonstrate first, that this individual based interaction network densifies with city size, as the total number of contacts and the total communication activity (call volume and number of calls) grow superlinearly in the number of urban dwellers, in agreement with theoretical predictions and resulting from a continuous shift in the individual-based distributions.
Second, we show that the probability that an individual’s contacts are also connected with each other (local clustering of links) remains largely constant, which indicates that individuals tend to form tight-knit communities in both small towns and large cities.
Third, we show that the empirically observed network densification under constant clustering substantially facilitates interaction-based spreading processes as cities get bigger, supporting the assumption that the increasing social connectivity underlies the superlinear scaling of certain socioeconomic quantities with city size.
Additionally, the UK data suggest that the superlinear scaling of the total social connectivity holds for both different means of communication and different national urban systems.
Human interaction networks can be inferred from billions of anonymized mobile phone records. This study, focusing on Portugal and the UK, reveals a fundamental pattern: our social connections scale with city size. People who live in a larger town make more calls and call a larger number of different people. The scaling of this relation is 'superlinear,’ meaning that on average, if the size of a town doubles, the sum of phone contacts in the city will more than double – in a mathematically predictable way. The small excess of above unity implies a substantial increase in the level of social interaction with city size: every doubling of a city’s population results, on average, in approximately 12% more mobile phone contacts per person. This implies that during the observation period (15 months) an average urban dweller in Lisbon (Statistical City,N=500000)accumulated about twice as many reciprocated contacts as an average resident of Lixa, a rural town (Statistical City,N=4000,see figure).
Figure.Human interactions scalesuperlinearly with city size
(k=the number of reciprocated mobile phonecontacts, N=city population size; C=the fraction of mutually interconnectedcontacts)
Surprisingly, however, group clustering (the odds that your friends mutually know one another) does not change with city size. We examined the local clustering coefficient, Ci ,which measures the fraction of connections between one’s social contacts to all possible connections between them. A high value of Ci (close to unity) indicates that most of one’s contacts also know each other, while if Ci = 0 they are mutual strangers. The clustering coefficient averaged over all nodes in a given city, remains approximately constant with Ci ≈0.25 in the individual-based network in Portugal. It seems that even in large cities we tend to build tightly knit communities, or ‘villages,’ around ourselves. There is an important difference, though: if in a real village our connections might simply be defined by proximity, in a large city we can elect a community based on any number of factors, from affinity to interest to sexual preference.
Figure.The average clustering coefficientremains unaffected by city size.
Figure. The chance that two phone contacts
from one person know each other is notaffected by city size
By mapping society-wide communication networks to the urban areas of two European countries, we were able to empirically test the hypothesized scale-invariant increase of human interactions with city size. The observed increase is substantial and takes place within well-defined behavioral constraints in that i)the total number of contacts(degree)and the total communication activity (call volume, number of calls) obey superlinear power-law scaling in agreement with theory and resulting from a multiplicative increase that affects most citizens, while ii)the average local clustering coefficient does not change with city size.
Assuming that the analyzed data are a reasonable proxy for the strength of the underlying social relations, and that our results apply to the complete interaction networks, the constant clustering is particularly noteworthy as it suggests that even in large cities we live in groups that are as tightly knit as those in small towns or ‘villages’ . However, in a real village we may need to accept a community imposed on us by sheer proximity, whereas in a city we can follow the homophilic tendency of choosing our own village – people with shared interests, profession, ethnicity, sexual orientation, etc. Together, these characteristics of the analyzed communication networks indicate that larger cities may facilitate the diffusion of information and ideas or other interaction-based spreading processes. This further supports the prevailing hypothesis that the structure of social networks underlies the generic properties of cities, manifested in the superlinear scaling of almost all socioeconomic quantities with population size.
The findings of this research help elucidate the role of cities as accelerators of human interactions, and the effectiveness of urban social space in the diffusion of ideas and information. Eventually, they can help us understand a broader spectrum of social phenomena, from crime to the spread of diseases.