Software Process Modeling: Theory, Results and Commentary

As an emerging technology, the effectiveness and potential impact of process improvement efforts have been debated, but not fully tested or validated. At the very core of this technological evolution is the idea that the quality of a software product is highly dependent on the quality of the process used for its development. Branching from this central idea are three categories of innovations: process models, modeling techniques, and the novel concept of process maturity assessment. The purpose of this paper is to examine the current state-ofthe-practice regarding software process models, modeling techniques, and assessments. Particular attention is directed at the most recent process improvement initiative, Humphrey’s Personal Software Process. 1. Software Process Models All engineering disciplines depend on reliable processes to create products. The term process as used herein conforms to its accepted definition as a written description of a course of action to be taken to perform a given task. It follows that a software process is as a set of activities, methods, and practices for the activities associated with the production and maintenance of software. Many scholars [16] contend that unlike the more mature, professional engineering disciplines, software engineering is deficient in the areas of scientific theory, mathematical modeling, proven design solutions and rigorous quality control. These deficiencies are clearly not from lack of effort--the origins of software process models can be traced back as far as software itself. Among the earliest software process model to gain widespread acceptance was Royce’s [36] waterfall model, considered to have been first utilized by IBM in development of the eminently successful System/360. 1060-3425/98 $10.0 Alternative models of the software development process have continued to evolve. Examples include: prototyping, operational, concurrent, reuse, and spiral models. The most recently proposed solutions to the software crisis, software process improvement models, represent a major innovation—perhaps better described as a revolution than just another evolutionary step. In contrast to the ease with which new process models are accepted, the adoption of process improvement innovations has been slow. In fact, Balzer [5] stated that process improvement problems continue to overshadow the progress made. Major obstacles to software process improvement include the attitude and culture of the people involved in the software development process. Whereas adoption of any new process model can be accomplished by individual developers, successfully implementing new software process improvement processes requires top management support, a culture of accepting change as a way of life, and significant financial investment [10]. If any these elements are missing, the software process improvement effort is sure to fail. Some of the software process improvement models currently in use include: the Software Engineering Institute (SEI) Capability Maturity Model (CMM), the International Standards Group for Software Engineering's Software Process Improvement and Capability Determination (SPICE) model, and the Quality Improvement Paradigm (QIP). This paper will focus on the CMM, which has had a profound impact on the global software industry, in spite of mild criticism by many experts [7, 22], largely due to a lack of emphasis on technology or automation. Detailed discussions of SPICE [13] and QIP [6] are readily available along with descriptions of other approaches such as: Model-Based Process Assessment [26] and the Software Productivity Research assessment approach [22].