The first half is a must read for anyone struggling with strategically improving a legacy codebase. I found the second half repetitive and less insightful but the first half was so good it warrants this book a permanent place on my shelf.