A study of the problems of man-computer dialogues for naive users

The success of an interactive computing facility will depend, to a large extent, upon the effectiveness of the man-computer dialogue which it supports. Comparatively little work has been directed towards the design of effective dialogues for situations in which the 'man' is a 'naive' user i.e. a person without training or experience of computer procedures. Thus the aim of this project has been to produce a series of specialised guidelines for designers of dialogues for naive users.

An examination of the literature reveals that published dialogue guidelines tend to be of a general purpose nature and therefore cannot be applied directly to specific situations. Furthermore, as each set of recommendations is based upon a limited range of experience, authors opinions appear to contradict or be in need of further qualification.

At a practical level, a survey of computer games, intended to be self-explanatory and therefore suitable for naive users, bears out the widely held feeling that the dialogue interface is often a poorly considered aspect of interactive program writing.

Pilot studies highlight the need for experimental work into man-computer dialogues to be carried out under conditions conforming as closely as possible to a 'real world' environment.

The main study focuses upon the general public as users of a local information system developed and installed in Leicester's Information Bureau. Monitoring the public's usage of and reactions to the system has enabled a series of dialogue guidelines for public information systems to be produced. A review of the literature provides supplementary recommendations.

The influence of dialogue recommendations on the software writing community is considered. Less than half of a sample of application programmers are found to refer to material of this kind. Follow up interviews indicate that the concept of a dialogue guideline is too narrow and should be broadened to cover all types of dialogue design information. This would render it more applicable to differing design situations. For designers who do not refer to published material, it is suggested that .sound principles can be communicated via trained experts and the use of library subroutines supporting dialogue creation. An example is considered of a routine to process textual inputs.

A number of paths for future research are described concerning the development of experimental methodology suitable for testing man-computer dialogues, an evaluation of the proposed strategy for communicating dialogue design principles and the application of new input/output techniques to public information systems. It is also suggested that the likely social consequences of computerised information facilities should be determined.