This book is somewhat mistitled -- it is more about software project estimation and management than about quality or metrics. But as the estimation practice lead for our software development group I found it insightful and valuable.
The heart of the book is an empirical function relating effort (cost), schedule, and system size, based on thousands of actual software projects. The projects cover a huge range of sizes and come from a variety of problem domains. This equation, together with a Rayleigh-distribution model of staff build-up during the construction phase of a software project, allows prediction of schedule and effort for a given system size, normalized productivity, and staffing pattern.
The value of this model is that it is shown by the authors to be applicable over a very broad range of problem domains and system sizes. It at least has a chance of modeling nonlinear effects such as the "mythical man-month" (total effort goes up fast as you increase the size of the team). And with only two free parameters (process productivity and staff-up rate) it is simple enough to be fitted to your own historical project data without a lot of re-analysis.
A lot of the book is devoted to graphical and pencil-and-paper application of the model, which is not really necessary since an implementation of the model ("Construx Estimate").
The only reason that I did not give the book five stars is that it is a bit dated -- it describes a waterfall lifecycle model, as was probably used by most of the projects in the calibration set. But I think it is still valuable as a source of quantitative guidance for software project estimation.
