What Can We Learn About Social Networks From Traffic?

[originally published in Centrality, 15 April 2005]

Recent network research into traffic patterns and flow may shed light on the “flow” in social networks:

[from NewScientist.com by Kate Ravilious]

Traffic should flow best in cities when only a limited number of roads lead to the centre. This counter-intuitive finding could allow planners to prevent gridlock by closing roads rather than building new ones.

It comes from a new way of thinking about complex networks developed by Neil Johnson, Douglas Ashton and Timothy Jarrett at the University of Oxford, UK. The researchers began by approximating a complex city network to just a ring road and a number of the arterial roads that cross at the centre.

They then worked out how the average time for journeys changes as the number of roads increases. When the model assumed that there was no congestion in the centre, average journey times got shorter as the number of roads increased.

But that changed when the researchers modified the model to delay any journey that passes through the centre. With a small number of roads, journeys initially became faster as extra roads were added to the network. But beyond a certain number, adding more roads increased average journey times rather than cutting them. The optimum number of roads depended on just how much extra delay there was to journeys passing through the centre.

Natural networks

The model mimics effects that have been observed in cities in which short cuts generate more traffic and so increase congestion, the researchers say. But this is the first time anybody has managed to model the effect mathematically.

The same process of analysing the costs associated with moving across a network could help solve a long-standing problem in biology: why some natural networks are centralised like cities, whereas others are decentralised like the internet.

Making a comparison to social networks would of course require additional research, but I wonder if we don’t see the same sorts of traffic congestion there?

Those who occupy positions at the “center” of social networks — those who are most central to the network — naturally are involved in more social interactions: requests for introductions and guidance, and the like. Creating more connections to such individuals does not decrease traffic in their “neighborhoods” — on the contrary. As a result, it may take longer for a request for an introduction to take place if that request passes through very connected individuals, since they have more interactions of that sort going on.

But the traffic analogy fails in the social context in the final analysis, since all roads to the center of town are candidates when you driving, and the primary consideration in general is minimizing the time of your drive. However, the primary consideration in the social setting is not timeliness, but instead the likelihood of gaining some social leverage, and that is based on the strength of relationships and the reputations of the individuals involved. So while there is “flow” in social networks, it doesn’t really work like traffic in cities, since people naturally tend to minimize speed over effectiveness in social networking.