Review of "Measuring the Software Process: Statistical Control for Software Process Improvement," by William A. Florac and Anita D. Carleton

Review of

Measuring the Software Process: Statistical Control for Software Process Improvement, by William A. Florac and Anita D. Carleton ISBN 0201604442

 Contrary to most books in computer science, this one has remained very topical to the modern world of software development. Even though it was published in 1999, the strategies put forward can be applied today. Furthermore, the content can be horizontally applied across a wide variety of disciplines.

 This is due to the fact that what is developed in this book is a set of tactics that can be applied to nearly every development process. The subtitle could have been “Statistical Process Control for Process Improvement.” While there are significant differences between software development and other creative processes, much of what is done in quality control is identical across disciplines.

 It all starts with determining if you have a system that has enough stability so that it makes sense to even attempt to measure it, at least in the statistical sense. There is a lot that can be done with statistics, but most of it is based on assumptions that what has happened so far is an accurate rendition of what will happen in the future.

 Once that is established, and doing that is explained, the next tactics are collecting and evaluating the data. Charts and other visual aids are used to not only explain trends, but also to demonstrate how one works and analyzes data that will naturally contain a lot of normal variation and some that is abnormal.

 This is not a book that one can simply hand to anybody and say, “We need to do this.” To understand and implement the content of this book it is necessary to have a basic understanding of statistical processes. On the positive side, the standard college course in basic statistics will generally be sufficient.

 One of the best sentences that sums up one of the problems with working with all such processes is the title of section 6.1 “How Much Data Is Enough?” Collecting data is like hiking across the American prairie. You know that you have to stop at some point, but you never know if the ideal spot is just over the rise you see in the distance.

 There is a famous quote that sums up many of the problems of effective quality control.

“Not everything that counts can be counted, and not everything that can be counted counts.”

After reading the book, this quote will be less applicable to your professional life.