Affording Design, Affording Redesign - Principia Designae: Pre-Design, Design, and Post-Design - page 88

People's mental representations of routes consist of links and nodes, where the links
are typically paths or streets, and the nodes are typically intersections or landmarks,
usually accompanied by names. Significantly, this same structure, and the same
elements, underlies people's descriptions of routes and environments (Taylor and
Tversky 1992 ; Tversky 2011b ; Tversky and Suwa 2009 ) . This suggests a phenom-
enon that appears across many domains: the same mental model drives descriptions
and depictions.
How do we know all this? That is, how do we study the sketches? We collect a
large number from individuals under standard conditions and code them. The coding
is both a priori—we have some ideas of what we expect to find and look for them—
and a posteriori—in looking through the sketches people produce, we see phenomena
that seem interesting and code those as well. The coding has to be reliable; that is, two
scorers following the same coding scheme have to agree. The coding takes into
account what isn't represented as well as what is represented. For what is represented,
it takes into account how it is visualized. Roads, for example, are usually represented
as single or double lines; the landmarks, intersections or buildings or distinctive
features of the environments, that appear are represented as vaguely-shaped blobs or
simply as names. In the distortion of angle and distance, we compare the real angles
of streets and distances to what participants produce. Similarly for omissions, we
compare what people put in their maps to what they could have put in. We count all
those cases and compute statistics on the counts. Then we generalize, making small
leaps from the data: to the notion of links and nodes; to the distortions of angle and
distance; to the essentials, from what is and isn't.
7.2.2 Systems Design
A current project is examining how master's students in a course in design of
information systems create and use sketches in the service of design (Nickerson
et al. 2008 ) . The design of information systems is elusive because systems consist
of concrete objects like computers and servers that have a real configuration in real
space. However, what is critical is not the real configuration but a functional
conceptual one; how the objects are networked. Euclidean distance among compo-
nents is not important; their network connectivity is. What is particularly hard for
students to grasp is a bus, a module or group of objects that are mutually
interconnected, but may have unique links into and out of the group. A bus is
essentially a hierarchical structure, but represented on flat paper. Throughout the
course, students solved a series of design tasks. Most design tasks gave a descrip-
tion of a system to be designed and asked students to make a sketch of the system.
They were also asked to make inferences from the system, for example, the set of
shortest paths through the system. Reasoning about paths of information flow is
crucial to system design. Both students' diagrams and their inferences showed that
especially early on, they had problems with the bus structure, indicated by faulty
sketches as well as errors of commission, specifically, listing paths that were not