Empirical Evaluation of a Proposed Model of the Program Design Process

Abstract

Program design methodologies can be considered to be based on implicit cognitive theories. Therefore, studies evaluating the effectiveness of methodologies from a human factors perspective need to be conducted so that the results of these studies can be used in the design of IPSE’s. Structured programming, the most widely used methodology, has a very simple underlying cognitive theory. Because this theory has been shown to have limitations, the paper presents a richer theory of program design with emphasis on problem decomposition, based on the blackboard model, together with some supporting experimental evidence. A model of program designer behavior, based on the proposed theory, incorporates specific forms of sequential planning and island driving, which are steered by data and process oriented approaches. The model is expressed in terms of cognitive processes operating on long-term memory and communicating via information posted on a specific blackboard structure. The operational model, when applied to a particular programming problem, resulted in certain predictions; the most significant being that solutions based on primitive as opposed to abstract perceptions of problem structure are preferred. An immediate implication is that either of the preferred solutions require less effort to produce, or that they are the only perceived possibility. Supporting empirical evidence is advanced, from attempting the same programming problem. The implications for the man-machine interface of IPSE’s are considered, limitations of the experimental approach adopted are discussed and suggestions are made for further work.