Controlling Software Projects: Management, Measurement & Estimation

Tom Demarco should be required reading for anyone who is a software engineering manager. This book was one of the first in a long line of references for quantifying the craft of software engineering. But this book has not aged as well as, say "Mythical Man Month" or "Death March". Don't get me wrong - I am giving it 4 stars because it is still full of excellent material; you just need other, more 'modern' references to help give this book perspective. With the proper perspective, this book really helps explain "The more things change, the more they stay the same", something that is often forgotten in this fast moving field. The context that this book needs today is that of a given 'methodology', whether that be Extreme Programming, Scrum, Unified Process, or CMMi. If you don't have a model in mind, I would start with the book "Agile Software Development, Principles, Patterns, and Practices", by Robert Martin.

"You can't control what you can't measure", is how DeMarco opens this seminal work in software project management, and the primary focus of this book is how to develop an effective strategy for measuring software development costs and thus build a reliable means of estimating the cost of future projects. Most of this discussion is, in my opinion, too dated to be of direct use (DeMarco suggests, for example, that counting the now obsolete metric "frequency of compiles" as an indicator of software stability). But the true genius of this work lies in the periphery of his discussion--particularly insightful comments into what not to do when managing software development. Useful insights, lucid prose, and DeMarco's engaging wit keeps this book on the top of the stack on my desk, and I expect to revisit it often.

This is an excelent textbook about software projects estimation. The only problem with the book is the lack of more examples to learn the techniques discussed.