PROGRAMMING : Facts and Fallacies of Software Development by Robert Glass

(hope I got that name right :)
Having read this book I have a lot of respect for Glass, who I haven't really encountered before. He has a lot of history and a lot of experience behind him. All the stuff in this book is spot on, but it's not particularly contentious (as he wishes it was) and most of it is quite old-hat. Maybe managers might get something out of it, but they've never listened to us before so why would they start now.
The two major things I got out of this book were

• a 25% increase in problem complexity results in a 100% increase in solution complexity
  This is actually the clincher and can be used to back up pretty much every other claim in the book. In fact this is such an important statement i think that it should be tatooed on all developers foreheads.
  
• Current research into computer science is pretty crap
  In largely consists of speculation and hype, rather than actual investigation and experimentation. He put this so well that I was somewhat inspired to help sway the scales, but not until I'm rich :)

He also said some other interesting things (55 facts and 5+5 fallacies all up, most of which aren't just filler :), and encouraged computer science courses to teach code reading before writing.
I guess I did get a bit out of it, but I really feel that it could have been summarized in 55 pages of interesting writing, rather than 200.