Join 3,553 readers in helping fund MetaFilter (Hide)


Maps of Tweets: locals v. tourists, languages, and mobile devices
June 19, 2013 10:01 AM   Subscribe

Digital mapping startup MapBox teams up with social data warehouse Gnip to create some stunning visualizations of every geotagged tweet since September 2011.

There are three zoomable maps on offer: Locals & Tourists, Languages on Twitter, and Mobile Devices on Twitter. The continent-scale views are interesting global data but the real reward is zooming in to cities such as Tokyo, Chicago, or São Paulo to see neighborhood by neighborhood distributions of languages and tourist interest.

The maps were developed by Metafilter's own Eric Fischer and Tom MacWright. Eric's work has been discussed several times on Metafilter.
posted by Nelson (7 comments total) 20 users marked this as a favorite

 
MapBox is the coolest. Everyone should be communiating solely by MeatText.
posted by troika at 10:23 AM on June 19, 2013 [1 favorite]


Looks like some tweets from Area 51 (FAA Las Vegas sectional chart underlaid). I guess they allow personal cell phones there.
posted by crapmatic at 10:31 AM on June 19, 2013 [1 favorite]


Completely awesome.
posted by Emperor SnooKloze at 11:04 AM on June 19, 2013


These rock.

This type of data could probably be used to drive or at least inform a range kinds of policy decisions as well. The first thing I noticed when exploring the US map was how certain parts of the transportation infrastructure emerge from the map. I'm sure there's useful information in there that would complement existing data about large scale traffic patterns.
posted by Hairy Lobster at 1:47 PM on June 19, 2013


Thanks! I think it is actually the small scale transportation patterns that are more interestingly revealed: large-scale mobility is reasonably well known from census surveys and transit ridership counts, while there is very little formal data about pedestrian patterns. But where pedestrian counts are available, tweet counts near intersections correlate pretty well with them, so I hope they also generalize to other places where no one has counted.
posted by enf at 3:36 PM on June 19, 2013 [3 favorites]


I'm sure there's useful information in there that would complement existing data about large scale traffic patterns.

We're looking at a collated series of snapshots over time here. One could try to break this down into smaller time increments that could be "swept" over, to see if there are interesting or useful patterns of data traffic that associate with other events. A bit creepy, in light of recent revelations, but working with large datasets like this starts raising the possibility of broader explorations.
posted by Blazecock Pileon at 12:25 PM on June 20, 2013


Interesting choice to make each pixel only one color, winner takes all. Looks nice, but may give a skewed picture.

Also, I wish there was a better way to see what was in an area. The black background with black labels makes that really hard, unless I'm missing an option.
posted by smackfu at 1:43 PM on June 20, 2013


« Older Well, all you lazy butt people who couldn't manage...  |  My Imaginary Well-Dressed Todd... Newer »


This thread has been archived and is closed to new comments