A while ago, Wired reached out about putting together a fun data visualization for Inform, an exhibition in the Adobe Digital Museum (ADM). Thomas Goetz, Executive Editor at Wired, was the curator.
I worked with Tim Leong to obtain a massive data set of ranked Twitter users - when I say massive, I mean gigabytes of many compressed text files. Within the data set, Twitter users are scored along various dimensions of reputation: trust, friendliness, interestingness, etc.
After going through an ETL process, I started tinkering with different ways to represent the dataset. While I could have implemented charts or graphs, I really wanted to do something different, something weird. When I took a step back I started seeing Twitter as this ecosystem where organisms (users) act and react to behaviors (tweets/retweets) from other organisms. I thought it would be interesting to think of each Twitter user as a uniquely colored and shaped single-celled organism, or as I call it tweeture.
The visualization was initially sketched in Cinder, an incredible C++ framework. In Cinder I wrote a minimal amount of code the generate hundreds of different looking tweetures. As a constraint, I wanted to generate each tweeture by chaining together simple shapes. The generated tweetures were stored in unique folders with an image and the respective algorithm. I then manually went through and determined the tweetures that looked appealing. Coming up the sketch was fairly straight forward. The size, flocking behavior, and animated behaviors are all a function of a tweeture's rank.
While I enjoyed hacking in Cinder, I had to move the piece to Processing so that it could run in the context of a browser. All told, I'm happy to see how fast it came together in such a short period of time. Additionally, it's a great feeling to have my work next to my data visualization heroes, like Aaron Koblin and John Maeda.
You can view the exhibition or a stand-alone version of Tweetures.