Methodology II: Cognitive Dimensions and the Gulfs - Foundations for Designing User-Centered Systems - page 333

soup spoon or a dessert spoon (or both). Similarly, many database systems require
that you plan and commit to particular record structures and size limits before
entering any data or actually using the system. Until you know how the system will
be used, however, it is often difficult to finalize these decisions.
These limitations are examples of Premature Commitment. Computer pro-
gramming environments often require premature commitment such as declaration
of variables before working out how they will be used, and developing the system
in sequential order. Like many of the CDs, premature commitment will interact
with other dimensions. Premature commitment can contribute to viscosity, for
example, because the effect of the commitment is to make future changes very
hard to make.
The solution to the problem of premature commitment is usually to allow (and
provide support for) users to perform tasks in many orders (e.g., outlining tools in
many word processors allow the user to develop the outline and write the text in
any order). This support comes at a cost, however. Allowing users more freedom
often makes the system more complex to design and to build, and may contribute
to errors in program code. Paying attention to this trade-off, however, is com-
pletely consistent with the Risk-Driven Incremental Commitment Model that we
will discuss in Chap. 14 .
12.2.5 Hard Mental Operations
The final CD that we will examine here is the number of hard mental operations
involved, which will vary across systems. In GOMS, KLM modeling, and several
other task analysis methods, the mental operations are all assumed to be equiva-
lent, for convenience's sake. Psychology and human factors work show us that
some kinds of problems (and operations) are substantially harder for users to
perform, and that users thus prefer easier mental operations (although harder does
not necessarily mean slower).
The issue of hard mental operations can be illustrated by the isomorph of the
Towers of Hanoi problem, in which monsters manipulate balls according to a
formal monster protocol for swapping balls. The subjects in Kotovsky and Simon's
( 1990 ) study had monsters either follow a complex protocol to swap balls around
based on which size monster could hold which size ball, or a protocol that required
monsters to change size to swap balls around. Subjects found the latter protocol to
be much more difficult, possibly because we tend to think of size as a relatively
constant aspect of an object, so having to mentally change the size of an object is
more strenuous and error prone than applying simple rules to behavior. Similar
effects have also been found in mentally rotating objects, with larger objects being
rotated more slowly.
As another example, multiple negative facts can be very hard to disentangle
(e.g., the packet that was not lost, was not put out on the Ethernet). The concepts of
pointers and indirection in C and other programming languages also prove very