Response to article "Improving Software Process Improvement" from John P. Durbin, Sept. 2002

The July/August article "Improving Software Process Improvement" by Reidar Conradi and Alfonso Fuggetta contends there is an "essential problem undermining most companies' SPI initiatives.." They say the problem is the selected SPI such as Software CMM, SPICE; they propose yet another framework (6 theses). They skipped over the crucial phase - Setting Goals - of Quality Improvement Paradigm (QIP) published by Victor Basili in 1994, which the authors used to launch their theses.

Given different goals, the resulting criteria for each SPI model will be different. In 1980's the trigger for Software CMM was the US government's goal of "high quality products" (compliant and reliable). The goal was not "product innovation" for retail demands. It is a SPI model for space shuttles used for many years, not cell phones used for a few months.

The first half of the article should have been focused on the "essential problem" of clarifying goals for software organizations selecting a SPI framework. Perhaps the title and focus should have been more like: "A SPI for Product Innovation".

John P. Durbin, Software Quality Engineer, Technical Staff

Response from Reidar Conradi and Alfonso Fuggetta, 1 Oct 2002

We refer to the letter by John P. Durbin. We believe that there is some significant misunderstanding.

As explicitly stated in the paper, we did not intend to propose "yet another SPI framework". The emphasis was on suggesting guidelines for evolving existing frameworks and, more important, for their use. To this aim, we formulated 6 pragmatic theses to summarize our industrial SPI experiences. These were never meant to constitute a full-fledged SPI framework.

Also, we are well aware that all SPI frameworks and ensuing SPI plans have a Plan-Do-Check-Act structure in the spirit of Total Quality Management. This means initial goal setting and planning, execution, final assessment of goals, and further replanning etc. etc. We will, of course, not abandon these principles. Indeed, we have used them in the experiments cited in the paper.

Our point was just that the goals of a planned SPI effort must primarily be aligned with overall company/business goals, not mechanically adhering to a predefined set of e.g. "key process areas" as in CMM. The reason is exactly the one suggested by Mr. Durbin: CMM was created for a specific purpose, and it is not appropriate to apply it as a general improvement framework. For most application areas and IT companies nowadays, the common over-focus in SPI work on "discipline and control" is likely to reduce the chance of success of an SPI effort. Thus "creativity" and the human and social factors must assume a larger role. Certainly, as mentioned by Mr. Durbin, there are other application areas or industrial sectors where "product refinement" is more dominant than "product innovation", and thus CMM may serve well.

In conclusion, the whole point of the paper was not to suggest a new approach or to ignore the relevance of CMM. The aim of the paper is to reinforce the idea that there are many different types of organizations and business models, and it is not appropriate to use an improvement approach based on incoherent assumptions and/or expectations.

- reidar and alfonso

Inserted 10 Dec. 2002, Reidar Conradi.