Case Studies - Paper Prototyping: The Fast and Easy Way to Define and Refine User Interfaces - page 25

Figure 2.2: One of the paper prototypes used in testing. Not shown here are the colored dots on
pieces of transparency that were placed on top of the images to indicate the control points.
The Team Had Questions ...
In the early prototypes, the team had a number of fundamental questions about what users needed and
which design variations would work for them:
Did the overall information display work? Did the users really need to see both a detail and an
overview for each of the two images, or would just an overview do the job?
When users dug down into the details of an image, what were they looking for? How did they want
the control points to work? Initially, the team thought of control points as having two states:
manually selected and predicted (that is, placed as a result of an algorithm using data from
previous manual point selections).
Did users want to specify the degree of magnification, or would a simple "zoom by a factor of 2"
control suffice?
How important was it to make the interface consistent with similar tools that their users had worked
with?
... And Paper Prototyping Answered Them
The team learned that the four views were in fact needed and that users wanted a high degree of
control over the magnification—zooming by a factor of 2 each time is conceptually simple, but it doesn't
cut it when you're looking at large land masses. The team also learned that their "Fit to Window"
control, which some had argued was too vague, didn't trouble users, because when they needed to
manipulate the magnification they'd use the more precise tools.
Some interesting subtleties emerged in regard to the control points. The team realized that there were
really several states—and combinations of states—that the points could have. For example, a point
had different meanings depending on whether it had a corresponding match in the other image or was
awaiting a match and whether the match was predicted or manually selected. Although users liked
having the predicted points, they needed to distinguish them from manually selected points because
they often had to tweak the position of the predicted point manually. Once the team understood the set
of point states that users needed to see, they created a symbol for each one and a legend to help the
users learn what the symbols meant.