Electronic Cocktail Napkin

PILOT EXPERIMENTS

We did several pilot experiments to explore whether people will make similar diagrams. If we build visual bookmarking systems where each person makes their own marks, we don't care that people all make the same kind of pictures. But to share bookmarks, or to make a diagrammatic index, then we need to depend on consistency in the diagrams people make. We can impose a system of diagrammatic index terms, just as text based indices do, but we'd still like to have some sense that people will more or less agree.

Drawing from Slides

Almost all diagrams made from a slide (left) included five main features.

In the first experiment, 50 undergraduates in the College of Architecture and Planning at University of Colorado were shown 10 slides of buildings and drawings and asked to make quick (30 seconds) diagrams to "help them remember or recall the image later". We found a high degree of consistency in the drawings. For example, given plan drawing (a), most sketches (e.g. b, c , d) included five main features - (1) parallel walls, (2) I-shaped stair module in the center, (3) top closure, (4) bottom U- or J- shaped element, and (5) a bottom closure. For each element, two or three alternative drawing representations were employed.

Drawing from Memory

Diagrams from memory fell into 3 main categories: boxes, spiral, and helix.

The second experiment involved about 20 designers, faculty, and graduate students at Georgia Tech and University of Colorado. They drew, from memory, three well known buildings: Frank Lloyd Wright's Guggenheim museum, Le Corbusier's Villa Savoye, and Le Corbusier's chapel at Ronchamp. Again, the drawings they made showed a strong consistency. For example, most people drew the Guggenheim museum in one of three ways: as a set of three or four stacked boxes; as a flat spiral, or as a curlicue representing a helix (the gallery space of the Guggenheim is in fact a helix).

Drawing to Illustrate Text

In the third experiment (Ellen's doctoral qualifying paper) 62 undergraduates at Colorado were given 6 text descriptions of architectural problems or responses (e.g. sunlight from the windows causes glare on computer monitors). They were asked illustrate the text with simple diagrams.

We found (1) The diagrams used a small vocabulary of elements; (2) there was a marked preference for making a plan or section drawing depending on the problem; and (3) people tended to add information to the diagram that was not included in the text. This suggests that people may be making a diagram based on a mental schema of the concept, rather than drawing directly from the text.

Next topic: Stretch-A-Sketch Back to Napkin Topics