Martin van Creveld: The Transformation of War: The Most Radical Reinterpretation of Armed Conflict Since Clausewitz, Free Press 1991.
This was was pushed on me by Eirik Newth, on the theory that I would be interested. He was right. Van Creveld shows how the traditional, Clausewitzian concept of war (as a fairly regulated game between two clearly identifiable nations, a “continuation of politics by other means”) breaks down when the fight is not about land, but survival, and when the power differential between the two warring parties is too great, the nature of war will change towards terrorism and other forms of “low-intensity” conflicts, increasingly also targeting political or military leaders rather than their fortifications.
His definition of war is interesting: It is not a “real” war unless both parties are putting their lives at risk. Soldiers attacking a weak or unarmed enemy are committing an atrocity, which harms them morally and in the long term renders their cause unjust. Here is an excerpt of this in my opinion pivotal point (p. 175):
Another very important reason why, over time, the strong and the weak will come to resemble each other, even to the point of changing places, is rooted in the different ethical circumstances under which they operate. Necessity knows no bounds; hence he who is weak can afford to go to the greatest lengths, resort to the most underhand means, and commit every kind of atrocity without compromizing his political support and, more important still, his moral principles. Conversely, almost anything the strong does or does not do is, in one sense, unnecessary and, therefore, cruel. For him, the only road to salvation is to win quickly in order to escape the worst consequences of his own cruelty; swift, ruthless brutality may well prove more merciful than prolonged restraint. A terrible end is better than endless terror and is certainly more effective. By way of an analogy, suppose a cat and mouse situation. Its very size precludes the mouse from tormenting the cat, though it is capable of driving him crazy–a different matter altogether. The cat, however, must kill the mouse at once. should it fail to do so, then its very size and strength will cause its actions to be perceived as unnecessary; hence–had it been human–as cruel.
There are a number of other points as well – about the role of women in war, for instance: The smaller the group, the more directly active the women, but as soon as things become organized and regular, the women are relegated to support roles, so as not to spoil the game for the men.
This book was written in 1991 and does, of course, not say much about the Iraq conflict, but it is, according to Wikipedia, required reading for US Army officers. It should be.
On the negative side, the writing is uneven and a bit repetitive – I found myself longing for a table or some sort of timeline of wars, and fewer restatements of the main hypothesis. But in most places, it shines.

Kevin Kelly’s very useful blog Cool Tools has an entry about the
Gladwell, M. (2005).
Keegan, J. (2003). 
Coburn, P. (2006).
The Search is better written – it is a mix of a corporate biography and a discussion of how search capability changes society. The language is tight – though sometimes cute, as in the phrase "the database of intentions" about Google clickstreams and archived query terms – and there is a thread (roughly chronological) through the book that allows most people who have been online for a while to nod and agree on almost any page. John Batelle has
Ambient Findability looks at search from the other side of the coin – how do you make yourself findable in a world where search, rather than categorization, is the preferred user interface? For one thing, you have to make your whole web site findable, make it accessible and meaningful from all entry points. Morville fills the book up with drawings and pictures on almost every page, comes off as a widely read person, but I am still looking for a thorough expansion of the central message – or at least some decent and deep speculation on personal and organizational consequences. It is more a book popularizing information science than a book that wants to tell a story, and it shows.
Steve