Humans strive to organize information so we can get our heads around it, understand it, and communicate it to each other. We do this by bringing order to the data. But too much information, even well-ordered, can overload our circuits and cause our eyes to glaze over.
One way we’ve learned to avoid such reactions is by using information graphics. But pie charts, bar charts, and graphs only go so far. Therefore graphic artists are constantly searching for new ways to visualize data and organize information, such as the circular information graphics that appear in the October and November 2012 issues of Scientific American (see below). Both of these graphics illustrate relationships and quantitative data. But one succeeds, in my opinion, while the other fails. Why?
What these data visualizations have in common is Martin Krzywinski, a Canada-based cancer genome researcher who has had a major impact on science visualization through his creation of Circos, software that converts data into circular formats. Diagrams created in Circos have become standard fare in scientific articles on genomics such as this. The circular format provides an excellent way to visualize genetic alignments and chromosomal relationships, as well as quantitative information.
Information graphic artists rapidly picked up on what Krzywinski’s software could do for them. An excellent application of Circos appeared in the New York Times online in 2007. In this piece, NYT graphics editor Jonathan Corum used Circos to produce a web-based graphic of the verbal interactions between politicians involved in the presidential debates. This graphic succeeds because it allows users to view the information selectively rather than all at once. Without this feature, Circos-generated graphics like this one can be so complex that they repel, rather than engage, their audience.
The mind’s eye can easily imagine a circle divided into between two and six equal parts. But try to imagine seven, nine, ten, or eleven divisions. The brain stumbles. When a circular information graphic attempts to communicate too much information, the result is that people will see complexity and nothing else.
In John Sexton’s October 2012 Scientific American article, “State of the World’s Science,” graphic artists Sven Laqua and Arno Ghelfi make the mistake of trying to do too much (see above). In their graphic illustrating the scientific exchange between 25 countries, some major trends are clear, such as the exchange between the U.S. and China, Germany, and France. The rest is spaghetti.
What works better is Jan Wilhem Tulp’s creation for the November issue of the same publication. In the Graphic Science department, Tulp presents a brilliant circular reinterpretation of data presented in a 2011 paper by Lorenzo Isella et al., regarding personal interactions in hospitals. While this looks at first glance like Circos, Tulp used a different approach that uses a method called hierarchical edge bundling. The data shows that compared to their colleagues, nurses have many more interactions with other sectors of the pediatric ward, making them more likely to transmit diseases (see below, before and after). This graphic succeeds because Tulp took the time and space to parse the information in print, in the form of secondary graphics, just as Corum had done online.
Clearly, Circos and circular presentations of data are a great way to organize information and will continue to influence visualizations. Yet there is no escaping our own limitations regarding what we can visually digest. It is a great thing, however, that software originally developed for visualizing genomes can be used to visualize data for everything from car sales to transitions in the workplace.
And if you want to see some mind-bending visualizations of the 2012 presidential debates, circle round to Krzywinski’s latest here.
I particularly like his Wingbag Index.by