data flow browsing by tag


Review: Data Flow - Visualizing Information in Graphic Design

Tuesday, January 20th, 2009

My director handed “Data Flow: Visualizing Information in Graphic Design“  to me shortly before Christmas.  Being part of the visualization group, I’m always looking for new ideas in shiny, gelatin-covered books and magazines, hoping for inspiration in the form of informative, concise, and original visualization of interesting data.  I’m new to the field, relatively speaking, and so I’m still in a very active process of building my visual vocabulary.  I’ve got a strong eye for what I like, and what I like to imitate, though.  Tufte.  I love Tufte.

This book is about the same size, physically, as a Tufte book, and some, at least of the visualizations follow my favorite of his principles: trust the viewer with the data (present all the data you reasonably can), do not present information that is not data, and elegance and minimalism trump flashy.  Quite a number of the visualizations don’t follow these principles at all, but a number do, and for examples of those, I like this book.

What I don’t like about this book is that very little explanatory text goes with each visualization.  Now, if you’re unfamiliar with visualization, you might think that vis is supposed to be self-documenting.  That’s true up to a point in that things should be visually intuitive; you don’t use yellow to represent cold on the same graph where you use blue to represent hot, as a simple example.  But good visualizations of large datasets present so much data to the user that an extended legend is generally warranted.  This extended legend ought to include:

  • An explanation of what the dataset is.
  • An audience appropriate explanation of the domain.
  • How the dataset maps to the visualization.
  • Patterns that the researcher who commissioned the visualization has found or finds interesting.

Especially for this book, which includes hundreds of different visualizations from different domains, such legends would be helpful. I’m left as a reader without context, and while I can look at any page and think to myself, “That’s sure shiny,” I can’t look at visual elements and judge for myself whether a particular visual element was the “right” one for the data shown.  I’m not trying to review others’ work so much as garner for myself a visual vocabulary.  Here, I feel like I’m getting the words, but not the definitions.  A sort of Scrabble dictionary for data.

The other issue I have with the book is that it includes along with some very good visualizations, a number of infographics which seem to be designed to persuade rather than inform.  These visualizations feel more informative than they are, and lead the user to come to the conclusions that the graphic designer or her commissioner has already come to, rather than give them the chance to see things for themselves.

Yes, I think there is value in being visually striking, but for instance the graphic on page 155, “An Underground Disco Information Organization” gives me very little data with an excessive amount of context, which itself is presented like evidence.  Graphicdotal evidence, like anecdotal evidence, tells a convincing story, but it often occludes raw data, methods, and significant results.

So, my favorites from this book:

  1. Pie, pp. 38 - Graphing the mood of movies by analyzing frames for inherent color and plotting it on circles.  Ingenious.
  2. Data Visualization of a social network, pp. 60-64 - Different views of a social network show different connections.  I especially like the combination of the radial layout with the US map.
  3. War Report, pp 246 - Here, despite the sensitivity of the subject, the designer has chosen to leave his own presentation simple enough to let the data tell its own story.  The story that good raw data shown clearly tells is always more powerful than half-cocked data shown with graphicdotal markers.
  4. Sunlight Calendar, pp 202 - A complete calendar of the average sky hue at different times of day on each day of the year.
  5. Atlas of the North Sea, pp 105 -  Complex matrix visualization of the North Sea’s geography and significant environmental and geological factors.
  6. Silvertown Affect Map, pp 88 - Deceptively simple, and an elegant use of the data available to them.
  7. Cyberwar, pp 94-45 - Graphing a botnet attack
  8. Literary Organism, pp. 78 - A graph drawing of Jack Kerouac’s On the Road with a brilliant legend and visually appealing concept.
  9. The Shape of Globalization, pp 35 - Embedded circles to represent nested companies.
  10. Total Interaction, pp 48 - Combining pie charts with Florence Nightingale charts and anchored bubbles for a visually complex but informative graph.

So my review of this book is mixed.  Go to it for visually inspiring design and appreciation of how much data can be put onto a page at once without turning into noise.  Be wary of graphicdotal elements and techniques that are more persuasive than they are informative.  And write the authors and tell them to tell us more about the visualizations they collected!