Jones seeks to cover “all of the broad and most of the specific topics associated with starting a full corporate measurement program that encompasses productivity, quality, and human factors.” This objective places the book nearer to Grady and Caswell [1] than to Conte, Dunsmore, and Shen [2] in the spectrum of software measurement books. It makes a welcome and significant contribution to software measurement and economics.
Chapter 1, “Introduction,” succeeds as a summary of the major themes of the book: reasons for measurement, hard and soft data, function point analysis, differences in companies’ approaches to measurement, charts of accounts, and practical issues in the development of effective measurement programs in organizations. Chapter 2, “History and Evolution of Functional Metrics,” is true to its title. The author includes early work of A. J. Albrecht in l979 and various related efforts and enhancements since that time, including continuing work by the International Function Point Users’ Group and Software Productivity Research, Inc., which was founded by the author.
Chapter 3, “United States Averages for Software Productivity and Quality,” is a largely self-contained, function-point-based profile of US software development and maintenance. Data were obtained from 4200 projects between 1950 and 1990. Chapter 4, “The Mechanics of Measurement,” explains the methodology of conducting a baseline software measurement study for an organization, including characterizing an organization’s defect removal methods and the complexity of its source code.
Chapter 5, “Measuring Software Quality and User Satisfaction,” outlines five steps to software quality control, with an overview of various technical and organizational approaches to improving software quality. Appendices summarize rules for counting procedural source code, function points, and feature points, and present examples of a fully measured software project and a measurement baseline study (including charts from a corresponding executive briefing).
Most of my concerns about the book are minor and are resolved by an appreciation for the context of the book and its message. Readers should not expect to find (as they would in Conte et al. [2]) a comprehensive treatment of core software metrics subjects: reliability models, complexity metrics, and non-function-point-based effort estimation models. The five chapters read more like five reports that range over several related topics and reflect the author’s wide experience. References to particular companies and their successes with measurement are a refreshing contrast to the vague allusions often found in such books.
I occasionally found the order of presentation to be unexpected, although not greatly disruptive (for example, the reference to “all nine measurements” on page 37). Because the chapters are largely self-contained, I re-encountered subjects that I thought had been covered earlier, such as charts of accounts in chapters 1, 2, and 4 and defect removal in chapters 1, 2, 4, and 5. In fairness, however, the content is informative, quantitative, and worth reinforcing.
I found the characterization on page 204 of the five choices for responding to questions on a questionnaire as a “true statistical tool” to be curious; I would have expected the use of carefully crafted statements and a Likert-type scale. Similarly, because I am cautious about drawing conclusions from a comparison of two projects from the same organization, I was awestruck at profiles of productivity and quality on software projects (chapter 3) where the sample size is 4200, including 150 projects from the 1950s. The author thoughtfully discusses sources of error in the data and his approach for analyzing it, reflecting his extensive experience in dealing with project data.
A distinguishing feature of the book is its prevailing orientation to function point analysis, or, more appropriately, functional measurement. Readers will find useful function-point-based benchmarks for productivity and quality with respect to dimensions like application domain, language, and software development and maintenance activity. The charts of accounts, user questionnaires, measurement profiles, and baseline studies will help organizations start measurement programs and benchmark their productivity and quality.
The strength of the book, a milestone in the maturity of function point analysis, is its exposure of measurement strategies, results, and profiles based on real experiences across a wide range of companies. I am excited that the author has been willing to share years of consulting experience on the application of functional measurement. This book will accelerate the use of functional measurement as the metric of choice for characterizing the economics of software development in the 1990s.
