This is a picture of the world, as connected by Twitter, created by Eric Fischer. It shows where people travel–and, what’s more, who they communicate with all around the world. Thus, in one map, you can see where people’s physical communities are, and their virtual ones as well.
Here’s how it works: Fischer tracked all of the information of Twitter accounts with geo-locations enabled between May 17 and September 1, 2011. Using this information, he could track to whom people were speaking (by looking at the geotags of the people to whom they were @replying) and where they were going (by looking at changes in their geotags). On the map, the @replies are rendered as purple lines. The travel is rendered as green. White lines are places connected by both travel and talking.
The connections are aggregated, and the brightness of these resulting shows how popular different connections are. The United States glows white as people travel and talk to one another. China, where Twitter is blocked, remains largely dark. Many other big cities are easily identifiable, including the bright spot surrounding Jakarta (a surprising number of people tweet in Indonesia).
While all of this is easy to explain, setting up the map so that these connections were legible took a great deal of ingenuity and information design.
The problem is line density, explains Fischer. If you draw straight lines between each point on the map, it becomes hard to tell which points are connected to one another because the straight lines end up passing over intervening cities. Normally you’d use a technique called Force-directed edge bundling to group paths together, “but I wasn’t able to figure out how to make it work when you have millions of locations.”
Instead, Fischer deformed the map, creating an area cartogram based on the density of geotags. Dense areas were made smaller, and sparse areas were made larger. By drawing straight lines to connect points through the deformed space, and then transforming it back to geographical reality, lines in dense areas were pulled apart and lines in sparse areas were pulled together. This gives the map its characteristic wispy routes and makes it easier to follow lines between distant cities (for example, notice how the lines from San Francisco to London curve up through northern Quebec, bypassing Chicago and New York).
“I’m still not entirely happy with the results,” says Fischer, “but I think this version looks better than with the other types of density blurring and area calculation I tried.” The Flickr set has some of his earlier experiments.
All of this ties into a body of work around using big data to create interesting maps. You may remember Fischer’s maps of cities that used Flickr data to work out where locals and tourists were taking pictures and other maps that used census data to figure out how segregated various cities were.
Fischer’s aim with these projects is to find opportunities to understand where people choose to spend their time and why. From there, we can begin to think about why people stay in some areas and quickly pass through others. Fischer mostly thinks about these issues at the city level, “but it is still fascinating to have a view into which places in the world are most closely related to which other places, either by actual travel between them or by membership in the same virtual communities.”
Check out our previous posts about Fischer’s work: