Browse Books

Reviewer: Fernando Berzal

Competitive engineering is mostly an extended and updated description of the key techniques the author already advocated in a previous book more than 15 years ago [1]. Many competitive engineering ideas, now dressed as powerful system engineering tools, were presented in that book, as sound software development advice. In fact, Gilb’s classic book has been acknowledged as “the first book with substantial chapters dedicated to [iterative and incremental development] discussion and promotion” [2], many years before the current hype over agile methodologies. Competitive engineering is also intended to serve as a handy reference for Planguage, a semi-formal structured natural language notation for communicating management objectives and system engineering requirements. Planguage, whose name derives from the union of “plan” and “language,” tries to promote communication in a clear and precise textual notation because many projects fail when communication fails. As a specification language, Planguage is designed to support all aspects of system engineering, with a strong emphasis on specification (both of requirements and design ideas) and project management. As a matter of fact, Planguage is more than a specification language. It comprises five core processes (some people might say best practices) that help projects to deliver timely results. The first process, requirements specification, covered in chapters 2 through 6, places a strong emphasis on the quantitative specification of quality attributes. As in his earlier work, Gilb focuses on measurable requirements. In his own words, “‘Bad numbers beat good words’: poor quantification is more useful than no quantification; at least it can be improved systematically.” Among many other things, Gilb also places emphasis on the importance of tracking the source of requirements and on the fact that requirements change and we must cope with this change. The second process, design engineering, covered in chapter 7, finds designs that match the requirements. At this stage, it is essential to distinguish between design ideas and design constraints (that is, requirements) in order to avoid the trap of letting designs masquerade as requirements. The third process, specification quality control, covered in chapter 8, is for defect detection and prevention. It revolves around a technique usually known as inspection in the software engineering field [3]. The term “inspection” was discarded by the author because of its connotations in other engineering fields. The fourth process, impact estimation, covered in chapter 9, is a reliable way of quantifying how effective design ideas are and, therefore, a practical method to numerically evaluate and choose design ideas. During the impact estimation process, evidence is required to support estimates and the inherent uncertainty in estimates is properly taken into account. The fifth process, evolutionary project management, also known as evo, is covered in chapter 10. It is a method based on the early and frequent delivery of results to project stakeholders. Each project is decomposed into a series of small delivery cycles and, within each cycle, design ideas are prioritized according to their current value for the project. This process is based on the fact that “your customers will be happier with an early long-term stream of their priority improvements than years of promises, culminating in late disaster.” Heavily influenced by the work of Walter Shewhart and W. Edwards Deming on statistical process control, its motto is “never perfect, continuously better” (page 219). The chapters devoted to these core processes all follow a standard format. They include examples, practical rules, process descriptions, key principles, discussions, case studies, and templates. They are full of interesting ideas, witty observations, and sound rules to be tailored to your specific situation. These chapters are not an easy read because of their condensed, highly structured, and somewhat redundant format; however, they are clear, precise, and, most of all, practical. A lengthy glossary, comprising more than 100 pages, completes this book. The glossary provides precise definitions and detailed explanations of most of the key concepts used in Planguage (the full glossary is online at http://www.gilb.com). Despite its many peculiarities and quirks, the author’s insistence on the consistent use of terms in Planguage is praiseworthy. However, even though the author considers the glossary to be “a central contribution” of his book, I found the methodological suggestions much more interesting and important from a practical point of view. Beyond the Planguage vocabulary and syntax details, which could easily be ignored, the author’s methodological insight makes this book very useful for organizations lacking a formal development process, for those individuals involved in process improvement efforts, and for software engineers in general. I only regret that the book had not been made more readable, giving it the potential to reach a wider audience; it could have been were it not intentionally written in a very condensed style to make for a handy reference to the author’s Planguage. Online Computing Reviews Service