Because every place worth going can be found on a topo map.
As with any art or science, one of the best ways to learn what works and what does not is to trace its history. For 900 years humans have been making graphical representations with various degrees of success, and for all of our time and technology we have not really done substantially better than Playfair’s time series from the early 1800’s. Certainly we can do more now, more data, more interactivity, more dynamism, but the basic principles of clarity are still at work. Tufte’s historical examples provide a narrative of evolving standards of excellence, and convey those principles, and the reasons behind them, better than a few bullet points at the end of a chapter.
I spent a fair bit of time pouring over the charts of cancer incidence, which, despite the flaws in the underlying data regarding differential reporting and geographic aggregation, are shockingly successful in elucidating overall trends. Our visual system is fantastic at both recognizing and then ignoring outliers on such graphs to better focus on broader trends. These maps, and the ones that followed, really made me consider the evolution of visualizations of direct analogies to physical reality (maps) to increasingly abstract representations of complex data. The earlier designs are obviously the most intuitive, and as we move further into abstraction we risk leaving the viewer behind us. It took statisticians 800 years to figure these sort of graphs out, and we ask our viewers to do the same in under a minute.
The theme present in both Tufte and Few was that of data ink versus non-data ink (I think Few made some other points, but they were trivial.) It is a concept that Tufte relates with precision, which I respect. The ideal of a successful representation is to directly and quickly engage the viewer. However, once visualizations move into the realm of abstraction (which is necessary in cases of massive displays of individual level multivariate data, a la my project) I feel like precision and simplicity are not directly linked, or rather, given an already complex system, a complex method of display may be required. It is very much like Few to give the advice to be as simple as possible, but we should also be careful to be no simpler than that. Interactivity, I believe changes a number of the rules of the game, allowing for multiple complementary views for different types of viewer, the possibility of effective redundancy, and also entertainment and engagement value: That the viewer has to be motivated to engage with the graph in order to get something out of it. By necessity all of these concerns may at times conflict with a desire for simplicity or lack of non data ink. While simplify, simplify, simplify! is a good motto for a designer, it would serve us well to remember that the first necessary step for a good argument is having someone listening to you. Without something eye-catching, the rest means nothing.
