Kieras (1988), however, complains that the Card, Moran, and Newell pre-
sentation "does not explain in any detail how the notation works, and it seems
somevhat clumsy to use. Furthermore, the notation has only a weak connec-
tion to the underlying cognitive theory." Kieras offers a refinement, with the
Natural COMS Language (NGOMSL), and an analysis method for writing
down GOMS models. He tries to clarify the situations in which the GOMS task
analyst must make a judgment call, must make assumptions about how users
view the system, must bypass a complex and hard-to-analyze task (such as
choosing the wording of a sentence or finding a bug in a program), or must
check for consistency.