The goal of the authors’ research was to improve the management of software development. The paper begins with the common lament that software demand far exceeds current development abilities and then follows with the much less common suggestion that the failure to improve management practices is a major cause for the persistent shortfall. The two main parts of the paper are an explanation of the development of a complex model of the software project management process and a discussion of the implications of the model.
Any experienced developer will appreciate the graphical explanation of the development of the authors’ model from a simple six-component model to a complex model with a dozen components. This evolution is based on common project management dilemmas. What follows is a research-oriented explanation of the construction of a model that integrates software development with planning, control, and human resource management. Researchers will appreciate the careful methods and choice of citations. Developers will find the final model structure appealing for descriptive purposes, even without reading the validation steps. But is the model useful beyond description? The authors make three claims for the value of the model beyond descriptive validity: (1) it uncovers dysfunctional consequences of current management policies; (2) it supports management decisions; and (3) it explains software project phenomena like the 90 percent syndrome. To support their first claim, they provide an interesting simulation study in which a judgmental safety factor is added to original COCOMO formula projections of project effort. This interesting application of their model could be copied by project managers for other questions of interest. The basis for their second claim is an example of how the model can justify quality assurance activities. The basis for their third claim is an illustration of how Brooks’s law (adding manpower to a late software project makes it later) applies only to the final stages of a late project.