InfoVis 1997 - Test of Time Award

The Structure of the Information Visualization Design Space

Card, S.K.; Mackinlay, J.D.

Stu Card's memorable acceptance speech--   Turtles and Rabbits v.8-2017.0904-1950

I sometimes think of papers as either fussy theoretical turtles or flashy visualization rabbits, that is, rabbits with the latest visual hop and flash -- you know: the kind that, if left at a conference unsupervised, multiply into many copies of the same idea. Turtles, by contrast, start slow, but over time, a few find their place in the sun among the thinning community of rabbits.  That slow, plodding work pays off for the long term when it creates abstractions and theories that allow us to connect our little islands of knowledge.

The paper under discussion is a turtle. This can be seen by the fact that we waited 20 years to discuss it, and that that delay is the reason it deserves a prize. The paper’s line of theory starts from the French cartographer, Bertin. Bertin created a theory of graphical meaning by categorizing the data side according to whether its variables were Nominal, Ordinal, or Quantitative and the visual side by retinal variables like position, color, and size. Mapping between the two had to be consistent. Jock Mackinlay extended the theory and showed that it could be formalized and computed with. The present turtle tries to go another step, aligning the theory with studies of human perception and increasing the descriptive properties. In the scheme of this paper, Jock and I reorganized the properties into a big table with four parts: (1) Data, (2). Controlled Processing, (3). Automatic Processing, and (4) Interactive Techniques. The result is a view of the structure of the design space that tries to connect the literatures on visualization, psychology, and design. By filling in the cells of this matrix, it is possible to describe and compare just about all visualizations that exist.

There are three reasons for betting on turtles. The first is to aid evaluation studies by relating the studies to structure. The second is to generate new ideas for designs. For years in the research lab, our group was able to just march down the wings of this structure, inventing visualizations as we tripped over the theoretical structures. The most spectacular example is that this is the theory that in various versions was behind the system and company Tableau. The third reason is that developing abstractions lets us grow the theory and connect it with other islands of knowledge, developing new technical disciplines. Thus, the abstractions and theories now link cartography to Information Visualization and on into support for Visual Analytics and Data Journalism. This will hopefully continue until finally we don’t have to depend just on lucky rabbits anymore, and it’s turtles all the way down.