Effects of withholding information about implementation details on the design of a human-computer interface.

摘要

Dividing a task or concept into several levels of abstraction is a practice which permeates many fields of study in computer science. For example, software engineering concepts such as information hiding and data abstraction have recently been applied to human-computer interface design. The result has been a wide range of tools which separate the design of the human-computer interface from the design of the functional component. The typical means of communication between these design activities is a collection of functions which are called by the human-computer interface component and implemented within the system component. This collection of function calls provides a higher level of abstraction which facilitates the design of the human-computer interface.; This dissertation investigates human performance in solving a human-computer interface design task where various levels of abstraction are presented. It presents a model of how designer performance varies depending upon the amount of implementation detail presented to the designer. Two experiments have tested the validity of this model by presenting four levels of implementation detail to designers. The four levels of implementation detail are: (1) no information about implementation detail, (2) a description of a reasonable collection of system functions, (3) a description of a reasonable collection of system functions plus a brief description of how these functions are implemented, and (4) a description of a reasonable collection of system functions plus a complete description of how these functions are implemented and a brief description of how the file system is implemented.; The results from these experiments show that designers generally benefit from having access to information about lower levels of abstraction (up to level three above). However, designers who are presented the greatest amount of implementation detail (level four above) are overwhelmed by the volume of information, and their performance deteriorates.