The Lingering Power of Geography Over How We Communicate
Modern communications technology was supposed to obliterate space, enabling us to encounter and talk to each other far beyond the boundaries of the backyard fence or the neighborhood pub. Eliminate the barriers and costs of communicating across longer distances – as cell phones and the Internet have – and, the theory went, our social circles would no longer have to be bound by geography.
"In the 1990s, this was a very popular suggestion," says Stanislav Sobolevsky, a researcher with MIT's Senseable City Lab. "Now we’re more and more seeing that that's not happening."
We now know, for instance, that many people on Twitter still interact with other users located within the same metropolitan region. And a new study from Sobolevsky and colleagues at the Senseable City Lab and the French telecom giant's Orange Labs adds yet more proof of the power of old-fashioned geography over how – and with whom – we communicate.
Following up on earlier research mapping communities with British telecom data, the researchers analyzed millions of landline and cell phone calls made within seven countries: the U.K., France, Italy, Belgium, Portugal, Saudi Arabia and the Ivory Coast. They developed computer algorithms to detect networks among the nodes of people who communicated most regularly with each other by phone. And the networks that emerged turned out to resemble not only cohesive local communities, but communities strikingly defined by the same borders familiar to us on traditional maps.
This means that even in our supposedly border-less world, most people in France or Italy or Saudi Arabia are still calling their relative neighbors on their cell phones (the U.K. data was based on landlines, and all the other countries on cell phone calls). Put another way: You can actually map local divisions within a country (think metro regions or states within the U.S.) simply by looking at these networks of phone calls.
In Belgium, for instance, just 3.5 percent of all of these communications cross the border between the regions of Flanders and Wallonia.
The below maps from the study, just published in PLOS ONE, show France, the U.K., and Italy. Regional calling networks, as they were identified by the study, are shown in the left column, depicted in different colors. Actual administrative borders are drawn in black lines onto the map, with impressive overlap.
If you were to partition France into the regional geographies where people communicate the most among each other, you would get a country divided almost exactly along political boundaries. In the right column, the regional networks are subdivided even further into local communications networks. Again, in the case with France, these hyper-local networks coincide almost perfectly with local borders.
"Delineating geographical regions with networks of human interactions in an extensive set of countries" by S. Sobolevsky in PLOS ONE.
This exercise tells us that geography still matters, even in the digital age. But it also suggests that geographers or politicians might look at data like this when re-drawing maps. In some cases, historic geographies that have long since been erased from maps lie hidden in our modern communication patterns.
Sobolevsky and his co-authors looked more closely at the intriguing case of Portugal. The (c) panel below shows the current regional administrative boundaries in the country, overlaid atop the communications networks detected in the study. The (d) panel shows the same picture, but layered with historic regional borders dating back to the 1930s that were later abandoned. Those old borders seem to more closely trace today's communities as people define them through their calling patterns. The (e) panel at right shows a proposed map for redrawing the country's administrative regions, from 1998. That proposal was defeated in a referendum. And this picture suggests at least one reason why:
"Of course, there are always remote connections, remote communications, remote locations," Sobolevsky says. "But the majority are still local ones."
What's most intriguing about data like this is its power to define what "local" means, and where exactly it's located.
Top image shows Belgium and Portugal.