On yet another day at the visualization conference, we heard wonderful back-to-back talks given by Manesh Agrawal (UC Berkeley) and Barbara Tversky (Columbia University, Teachers College) on design principles for visual communication. Agrawal and Tversky have collaborated before and their talks complimented each other perfectly. Agrawal started by introducing a generalized procedure for getting the most effective visual representations…he described three steps:
1. Start by examining the most effective existing illustrations in that domain – what, would everyone, agree is an effective visualization (even if it’s just a part of the pending question or an older representation)
2. Analyze those examples to identify what was the task of the viewer and what techniques/methods did the renderer employ to emphasize/de-emphasize information in the visualization? How did she get across the most essential information?
3. Apply those principles to the visualization you are attempting to create.
Sounds simple, right? Well, maybe not. But once Agrawal showed us a few examples, we began to get a feel for it. Take for instance their work with mapping visualizations. Agrawal and his colleagues wanted to improve the computer-generated directions and maps (ala MapQuest or GoogleMaps) to make them more effective. So, following their own guidelines, they started by looking at existing maps and drawings that others agreed were most effective. They looked at tourist maps and hand-drawn maps to see what made them so effective. They realized that the hand-drawn maps stressed the turning points (the people who drew them thought of the route directions as a series of turns). The tourist maps did a good job of showing landmarks, items of importance to a tourist, and de-emphasized structures or locations of less importance. Precise geometry was less important in both the tourist and the hand-drawn maps. From their analysis, they determined that effective directional maps would exaggerate the length of short roads so that the turnings were more obvious, that the precise road shape would be simplified, that extraneous information would be left out, and that the turning angles would be regularized.
From this, they built a system called “LineDrive” – this diagram shown here (taken from a paper by Agrawal and Chris Stolte) shows a side by side by side comparison of a typical computer-rendered map (on the left), with a hand-drawn map (in the middle), with a linedrive map (on the right). You can see line drive in operation for yourself on http://mappoint.msn.com.
After that, Barbara Tversky took us on a journey through some of her work, analyzing the way people interpret and give instructions. She talked about “break points” and took us on a magical tour of graphic novels and comic books – Maus (by the amazing Art Spiegelman), Larry Gonick, and Scott McCloud’s wonderful book Understanding Comics (and if you haven’t seen McCloud’s TED talk, it’s definitely worth it!). She pointed out the way comics break up time and space and then connect them again with windows, frames, boxes, parentheses, visual anaphora (look that one up!), symbol foreshadowing, visual puns, and language.
So, the themes that emerged to me from this wonderful pair of talks? Consider carefully the many ways that visuals help us navigate our world. When creating new visuals, work to reduce the visual clutter, emphacize the parts/structures/places that are relevant, be aware of what is not seen, combine forms (photo, painting, line drawing, video) where it makes sense to do so, and always keep that critical question in mind – what is the main point of the visual? I also took to heart the reminder to learn from others (the masters who’ve drawn before us). Look to their work for patterns, techniques, and methods. And be sensitive to the delicate call and response between the visual and the viewer – what are we being asked to bring to the visual?