ABSTRACT

One area of growing importance to ergonomists, and an area of specialism, is that of human-computer interaction (HCI). While the commercial trend is toward increasing the role of computing technologies in the workplace, accompanying this trend is an increase in design problems that concern the specification and evaluation of software applications. GOMS, or the Goals Operators Methods and Selection rules method (Card et al. 1983), was developed at an early stage in the history of HCI research, and is a method with enduring appeal. When considering ergonomic contributions to the system development process, contributions may be considered along two dimensions; the stage at which the contribution is made, and the nature of the contribution. System development may be characterized in terms of a set of stages: requirements capture; design specification (conceptual and detailed); implementation; and evaluation (including testing and maintenance). The nature of a contribution, at any stage, may be either analytic or empirical. Therefore, when considering the stage of requirements capture, an ergonomist may either survey users to empirically elicit expressed requirements; or generate requirements analytically, following reflection on the domain, tasks, and so forth. Using this scheme of characterization, GOMS may be described as a method for analytic evaluation. To understand the enduring appeal of GOMS, one needs to adopt a slightly more complex view of the stages of system development. While the product under development will necessarily be evaluated, or at the least tested to ensure it works once implemented, specifications (for that product) at earlier stages of development may themselves be evaluated. For example, an ergonomist may generate two interface design specifications and wish to evaluate them both, to know which will be more effective, before the time-consuming and expensive stage of implementation is undertaken. Only by adopting a suitable method for early evaluation can such a design decision be made reliably. Given that at early stages of development device specifications alone exist (the process of implementation comes later), historically

analytic methods have been required for early ergonomic contributions. (The recent rise of early prototype construction however, increasingly enables empirical contributions to be made at early stages of development.) GOMS was one of the first methods to support the early evaluation of design specifications, this being possible because GOMS supports the “prediction’’ of human-computer performance, rather than the “description’’ of empirically observed interactions. There are few methods in HCI which are predictive, fairly accurate, and which can be carried out quickly on the metaphorical “back of an envelope.’’ Hence the resilience of the GOMS method within HCI.