Browse Books

Reviewer: James W. Moore

In the annals of software engineering classics, few have been more influential than Winston Royce's 1970 presentation, “Managing the Development of Large Software Systems: Concepts and Techniques” [1]. Reacting to the then-prevalent code-and-fix development cycle, Royce described a systematic and interacting set of focused activities. Simplified and called the “waterfall” by others, the model has served as the baseline life cycle model for three decades despite criticism. The story is ironic because Royce's prescient presentation anticipated most of the improvements suggested by reformers. Twenty-eight years later, his son, Walker, a formidable software manager himself, has written the book. Drawing liberally on important work by the elder Royce, Barry Boehm, and others, this volume provides the materials that a manager needs for implementing an enlightened but systematic approach to software development. This book is organized into five parts. An overview of current practice (68 pages) introduces the three manageable factors of software cost—volume, rework, and automation. A description of processes and a framework for management (66 pages) provides a basis for a reorganization of the software life cycle based on the spiral model, risk management techniques, and a reference set of phases and milestones. Summaries of some of the disciplines crucial to managing a project (96 pages) address the controlled evolution of the plans, estimates, and other artifacts of the development process. A look toward future improvements (34 pages) considers practices currently being applied by innovative managers. Case studies and backup material (136 pages) include useful material on cost estimation, metrics, and process improvement. In addition, a glossary, list of references, and index are provided. The text is intended for software project managers. Royce does not deal with the technical artifacts of software development but rather with the economics, organizational approaches, and metrics needed for a successful project. Although some sections of the book are specifically labeled as “pragmatic,” the treatment is pragmatic throughout, drawing heavily on Royce's experience in successfully managing major projects. The advice is sometimes contrarian—for example, Royce feels that peer inspections are overemphasized. The text is to be applauded for its focus on the management of change. Fifteen years ago, the conventional wisdom could have been summarized as “freeze the requirements, and stick to it.” Royce understands, though, that any project involving significant effort must intersect with its users' needs as they exist as some point in the future. Rational management in the face of change necessarily involves measurement of change, and his approach to measurement is oriented to that purpose. Readers should be advised that the approach taught by this book is not straightforward. Implementing the approach includes mastery of a number of seemingly identical but different concepts: “stages” versus “phases,” “artifact sets” versus “releases,” and “iterations” versus “increments.” The terminology is complex, but so is the job to be done. My criticisms of the book are few and minor. A three-page discussion of the Airlie Software Council appears to have been forced into the flow of the book for no apparent reason. The book provides gratuitous references to the Unified Modeling Language (UML)—a concession to the needs of the employer rather than the needs of the reader. Most readers will easily overlook these flaws, though.