Recently one of my colleagues, Igor Nikolic, was fortunate to be invited to speak at TEDxRotterdam where he gave a great talk on how we at TU Delft use insights from Complex Systems Theory in order to deal with complex socio-technical systems.
The movies used in this presentation come from my work in exploring how visualization techniques can help us to better understand complex systems. Below the behind-the-scenes details regarding the creation of these is discussed for those interested in making their own visualizations, or understanding the data sources used.
The movies with the spinning globes were built using the Shapes3D library for the Processing visualization package, based on the example here. The Shapes3D
library is great and I was able to create the visuals with a surprisingly small amount of code. However, I wasn’t able to figure out how to anti-alias the frames, so I just blurred them slightly with the with the imagemagick utility to get rid of the graininess. This seemed to make things look more realistic, just as the atmosphere slightly blurs the surface of the earth.
The first movie uses as a base image the classic earth at night images by NASA, which shows how ubiquitous electricity consumption is around the globe.
The second movie overlays the NASA image with data from the Carbon Monitoring for Action site (in red dots), which contains information on locations, emissions and power output for over 50,000 power plants around the world. What’s interesting with this movie of power plants, is that you can see the different strategies that countries have undertaken in power generation. For example, countries like France do not seem to have many power plants, but they also rely on large-scale nuclear power plants. Other countries like Denmark and Germany seem to have a much more distributed portfolio of power plants.
The next movie was built using the excellent Prefuse visualization package, and is meant to demonstrate the concept of intractability, which has been defined as "problems that can be solved, but not fast enough for the solution to be usable"(1). The white line represents the path that you actually take through life, and the nodes represent decisions made which can alter your path. Every time you make a decision, you leave behind a whole realm of future possibilities that you could have lived (all the red branches are where you married someone else, pursued a different career, etc.). This means that the future is hard to predict because of these combinatorial explosions. The fastest way to predict the future is to just live through it.
What’s interesting about this movie is that the frames are actually played backwards from the order in which they were rendered, due to the intractable nature of the layout algorithm used. In other words, the only way to make this movie was to wait for the future to happen and then play backwards the path we used to get there. To explain, this figure uses a force-directed layout, where nodes push away from each other and edges bring them together like springs. Essentially, this is a physics simulation. The problem is that there’s no way to easily predict/calculate where to add the nodes to the outside so that the forces are evenly balanced so that the inner nodes aren’t pushed around, causing the whole figure to tremble in a rather distracting manner. If you place the new node too far out, then it will drag the inner nodes outward. If you place it too near, it will push the inner nodes further inwards. To get around this problem, I first drew all the nodes and waited 30 minutes for all the forces to balance out and reach an equilibrium. From there I started subtracting nodes randomly from the outside, until only a single node was left.
This last movie shows the evolution of the wiki at TU Delft since late 2004. Just like the intractability visualization above, this was done in Prefuse and uses a force-directed layout. This layout works particularly well for complex network structures since it allows for the visualization to self-organize, whereby tightly interconnected pages migrate to the center, while nodes with few links are pushed to the outside. For the colors of the nodes, red indicates inactive pages, while yellow means that edits or views are happening on that page.
If you look closely at this movie, you can see breaks for weekends and summer vacations. Blobs at the outside are likely notes that people made during conferences, or classes where students kept their notes and reports online. This wiki actually started on a computer under Igor’s desk, and once it became a part of the university ICT infrastructure in early 2009 you can see a significant flurry of activity. Much of the activity on the outside is likely Google and other search engines indexing the site, while on the inside we see the adoption of the wiki by new groups across the university.