In my visualization class today, we had a guest lecture by Michelle Borkin, another Northeastern professor who works in the field of information and scientific visualization. She gave us a great overview of the foundational design aesthetics of Edward Tufte.
Whether you know him by name or not, you may be familiar with some of his principles. He writes extensively about “graphical integrity,” highlighting the importance of clearly labeling of data and cautioning against distorted or misleading axes. But, perhaps more fundamentally, the Tufte-ian mantra seems to be summed in one word: simplify.
Tufte advocates for removing as much extraneous ink as possible. Non-data ink should be minimized as much of possible; clearing away the clutter and letting the data speak for themselves.
Generally, his arguments make sense – there’s no need to create a 3D bar-chart just because Microsoft Office says that you can. But in this day of infographics and data journalism, Tufte’s style can seem rather…dull.
This has led to a great debate over chart junk: a topic so real it has its own wikipedia page. “Chart junk” refers to any element of a visualization which doesn’t explicitly need to be there – elements which may make the visualization more interesting, but which don’t directly convey the data. The term was actually coined by Tufte, who, as you may have guessed, was adamantly anti-chart junk.
Recent research, though, has shown that “chart junk” isn’t necessarily inherently bad. Infographics and other visualizations designed for broad public consumption may not have the precision of a scientific visualizations, but they are more memorable and impactful.
Is chart junk okay? The answer, I guess, depends entirely on the audience, the task, and the context.