Assessing The Software Process
Some configuration manager argues that the assumptions about software processes and software producing organizations underlying the maturity models may not apply in all cases. This may be particularly true for the CMM since this model is developed in close cooperation with large American software organizations. Reporting from an attempt to use the CMM in a research environment, Velden et al. found the CMM rather rigid and lacking explicit consideration of business goals. More generally, a model is always an abstraction of reality. Therefore, model-based assessments run the risk of identifying "poor fit" to the model rather than to the real problems of the organization. Finally, the information produced by the assessment may be insufficient to plan and implement improvements. The assessment reports describe problems, e.g., poor requirements management, but the underlying knowledge about existing processes needed to identify causes of problems and possible solutions is not included.
These arguments are in line with more general insights from the field of organizational learning and change - particularly Schon's observation that organizational change must be based on an understanding of the situation at hand rather than on predefined problems and solutions.
When the SPI project at Bruel & Kjaer was initiated, the improvement group soon realized that the specific circumstances in that organization excluded the model-based assessment approach. First, a Bootstrap assessment had been carried out approximately six months earlier, but no follow-up action to that assessment had been taken and the information available was not rich enough to plan a new improvement initiative. Second, the group did not want to conduct a new Bootstrap assessment too close to the previous one, since this might destroy whatever interest in the SPI program that was left in the organization. Finally, but not least important, the group realized very early that Bruel & Kjaer's project managers had a powerful culture that would impede any change initiative for which the project managers did not approve. The improvement group therefore decided to conduct a new assessment, but with a new purpose and with a new approach intended to do the following: Identify the problems in Bruel & Kjaer's software process as perceived by the project managers. Build an understanding of the causes of the problems. Support the development of a specific improvement strategy. Involve the project managers actively in the assessment as well as in the subsequent selection and implementation of improvement activities.
Furthermore, the new assessment should require only modest resources, have low organizational impact, and be straightforward to carry out. The problem diagnosis technique, described in the following, was developed to meet these objectives. We believe, however, as discussed above, that there is a general need for an alternative or supplement to the model-based techniques. PARA Research Method
This research is based on action research, which, according to Galliers, attempts to simultaneously achieve practical value to the client organization and contribute to theoretical knowledge. A number of explanations of action research cover our approach see in particular. The core of action research is presented as a group of researchers who join a real-world problem situation to help improve the situation. The action researchers must in advance declare a framework and a methodology so that reflection subsequently can yield research findings. The framework in this research consists of the ideas, models, and assessment methods as they have already been explained in the previous section.
|