Last week I got my copies of the book Making Software. What Really Works, and Why We Believe It, one the last books published by O'Relly Media. This book is edited by Andy Oram and Greg Wilson, who also edited other top-sellers books like Beautiful Code.
For me, this book is like a software engineering edition of Mythbusters. In software engineering, you often find claims about a particular methodology or technology, about how cool it is or how much it sucks. However, few have stopped to try to verify if those claims are true or just myths. This book does that, and not only that, it also teaches how to proceed to study such claims and verify whether there is a pinch of truth on them.
Ahmed Hassan and I have contributed a chapter about product metrics, that basically says that lines of code is a product metric as good as other fancier complexity metrics. Our conclusion: stop bothering gathering fancy product metrics, and just measure lines of code.
Other chapters address questions like test-driven development (is it helpful or harmful), pair programming (does personality affect pair programming?), design patterns (do they produce better software?), copy and paste (is it that bad?).
Some links to have a look before buying the book: