stringent speed and accuracy instructions—in the context of Fitts’ law, which relates target
acquisition time T to the geometry of the task, T quantifies the minimum time it takes to reach
the specified target.
Finally, we leave aside navigation in inherently three-dimensional information worlds,
whether designed for work applications (e.g., Card et al. 2004) or games (e.g., Lewis &
Jacobson, 2002). We focus on the case of a user navigating a document, a flat surface
displaying text or (still or animated) graphics. Such documents are pervasive in current
computer applications such as text editors, spreadsheets, atlas viewers, photo editors, etc. In
fact, as in Fitts’ original paradigm, we will often reduce our documents to a single dimension.
2.3. Fitts’ Paradigm in Graphical User Interfaces
2.3.1. Mapping the Hand and the Document onto the Screen
In GUIs the target-acquisition problem involves four basic elements and their inter-relations.
The elements are the user’s hand (typically attached to, and, in this sense, confounded with
some input device), the cursor, the view, and the document.
Visualizing the
Hand
Visualizing the
Document
Screen
Hand
Document
Figure 1. The double mapping scheme of graphical user interfaces.
As illustrated in Figure 1, pointing in a GUI is indirect: the hand, which designates, belongs
to a different world than the document, which contains the objects to be designated. An
encounter can only take place in screen space, which requires a bilateral mapping scheme.
First, the user’s hand is represented on the screen by a cursor. Hand motion is translated into
cursor motion through a display-control mapping function. The slope of this function
determines the scale of display space relative to motor space.
Second, the document is represented on the screen through a view. The view displays a part of
the document called the selection, using a viewing function. Figure 2 illustrates this mapping
between the view and the document in the one-dimensional case. Both the documents (on the
x axis) and the screen (on the y axis), appear as intervals. The slope of the viewing function
defines the visualization scale, while its intercept defines the offset. View pointing (see
below) consists in controlling these two parameters: varying the slope changes the viewing
scale and is called zooming, while varying the intercept changes the offset and is called
scrolling or panning.2
2 The term scrolling was introduced in HCI along with the classic WIMP interface. The term panning appeared
more recently, coupled with the zooming facility in multiscale interfaces. Although their connotations differ, we
need not distinguish the two terms here.