Some of my purist friends contend that data is important but not so much how they are represented. I enjoy huge troves of data, but is also a strong believer that representation matters. We humans, for better or worse, are generally good at some things rather than others; and by expressing information in modules for which we have the best sensitivity we give ourselves a better fighting chance.
An example came up with my illustration work for Wikipedia. The request was the usual one, to convert a low-quality raster image into a high-quality vector format; the subject is the seating chart from a flight disaster. The image is as follows; can you tell apart which passengers were unharmed, and which ones were seriously injured?
As an illustration, this is not effective in conveying the information. The most confusing part is perhaps the visual similarity between “serious” and “none” – we (I?) just can’t see at a glance which one belongs where. Some other choice of patterning would have been much more helpful, for example, a gradation of line density that correlates with severity. My revised version follows:
The clear advantage I have here is the use of color, which the original government document have no access to. With colors being accessible, I could tap into the schema we culturally associate with danger and safety, and additionally used both saturation and brightness to separate aircraft configuration from seating injury. Most people probably don’t need to be able to read the legend to figure out that, at least in this incident, seats at the front are correlated with fatality.
Enhancements of this type are highly contextual, and require understanding what is supposed to be expressed: trivial in this particular case, but less so with more technical subjects. After the thesis I’m looking to do up a large series of chemistry-related illustrations, and I’ll think-aloud as that happens.