A report by the Brookings Institution analyzed a sample 20,000 ISIS-supporting Twitter accounts, and sought to answer the question of how suspending accounts impacts the community. The data visualizations, included in the report, bring the message home. According to the authors, suspensions serve to focus ISIS supporters into a dense cloud: on the lower right of the image above.
The node-link diagram describes ISIS supporters from February 2015. The darker cloud shows the increased concentration of interactions after the suspensions, while the lighter cloud at the top left shows more scattered peripheral relationships at the same time. "As suspensions contract the network, members increasingly talk to each other rather than to outsiders," the report states. The upshot? Suspensions, of which Twitter made more than a thousand by December, may have unintended consequences, including cutting off ISIS supporters from beneficial social pressures on Twitter.
The report was written by author and analyst J.M. Berger (@intelwire) and data scientist and podcaster Jonathon Morgan (@jonathonmorgan), and commissioned by Google.
Air traffic data is great fodder for visualizations. Case in point, this lovely animation of a day of flights titled "North Atlantic Skies" by air traffic control firm NATS. (via Laughing Squid)
Here's a cool data viz project from GE that displays real-time conversational spreads about breast cancer on Twitter. (thanks, Laura Hollister)
John Nelson—the data visualization designer responsible for that global map of earthquakes I posted last week—has also made a strangely beautiful map showing every tornado to hit the U.S. between 1950 and 2011.
Part of what makes this map interesting is that it shows not only the touchdown location, but also the path of the tornado as it moved. Better yet, Nelson has several other related maps that break the data down in different ways. For instance, if you look at the tornado map broken down by seasonality, you can see a really amazing pattern, where what constitutes "Tornado Alley" appears to move northward over the course of the year. In December, January, and February, the bulk of tornadoes have been centered on south and south-central states like Mississippi, Texas and Kentucky. In peak tornado season—March, April, and May—the southern states are still affected, but the reach of the tornadoes has extended north and west. By June, July, and August, most of the tornado activity is happening in states like Michigan and Minnesota.
Another interesting thing I spotted on these maps: There's a hole in tornado activity centered on West Virginia. All around the state, there's a history of tornadoes. In the Mountain State, though, the number of tornadoes drops off precipitously. I'm really curious what's causing that, or whether it's a flaw in the data.
Compare tornado habits throughout the seasons
Compare tornado numbers by F-scale
Compare tornado history before and after the historically devastating 2011 season.
Watch an animation of tornadoes by year
This map of all the world's recorded earthquakes between 1898 and 2003 is stunning. As you might expect, it also creates a brilliant outline of the plates of the Earth's crust—especially the infamous "Ring of Fire" around the Pacific Plate.
But the real story—which Smithsonian points out and which was also the first thing I noticed—lies elsewhere. To put it colloquially: Holy shit, you guys, look at all those intraplate earthquakes!
Plate tectonics explains a lot of things, but it doesn't totally explain why earthquakes (and, in rare cases, extremely large earthquakes) happen in places far from the meeting point of two pieces of crust. There are a few possible explanations out there. We just don't know yet which one is correct.
One of the theories explaining intraplate earthquakes is based off the fact that the tectonic plates we know today have not been constant throughout Earth's history. Some of the places that are now "intraplate" were once right along fault lines. Others are at spots where continents began—and then failed—to split apart. All these things might leave behind spots of "weak" rock that's more prone to upheaval than the strong, intraplate rock around it. Studies in the 1990s found that 49% of all intraplate earthquakes happen near places like this. Of course, that leaves 51% of the shaking still unexplained.
Read more about this and other theories for why intraplate earthquakes happen.
Check out Smithsonian's write up, which includes a nice comparison between this image and a cartoon map showing the tectonic plates.
View the image, created by data visualizer John Nelson, on Flickr