It's said that you can't judge a book by its cover, but that's no excuse for buying Knuth's "Things a Computer Scientist Rarely Talks About" without even looking inside, as I did. Based on that impulse-buying experience, I fear I may be the kind of fool the famous have to thank for the fact that they can pretty much just crap in a box and be guaranteed to sell it.
It's not often you buy a book from the CS shelves and end up wondering "Am I any better than a teenage girl who buys Britney Spears perfume?"
I can't even claim I wasn't forewarned. Years ago I bought, read, and passed on to a friend a copy of Knuth's earlier "3:16". So I was fully aware of the perp's prior. But hope triumphed over experience and I assumed that this time he'd be talking about esoteric CS or mathematical topics. Like "Selected Papers", only less famous.
But mainly Knuth wanted to waste 200 pages of once-lovely trees, $20 of my money, and several hours of my precious reading time on, well, nothing really. Nothing stood out as I read it, and nothing much has stuck with me. I remember him describing the religious symbolism in some of the calligraphy in "3:16", which for me finally explained why there were so many weak works in that book: the effort of shoehorning in symbolism can compromise anyone's work. (Interestingly, Knuth describes how he made a "correction" to one artist's work because he didn't get it; I remember that as one of the uglier examples, but uncorrected it actually works.) I also remember him commenting that the ratings for "3:16" on a certain well-known patent-abusing bookseller's web site were all either 1 (the lowest) or 5 (the highest).
The website ratings for "3:16" go to the heart of the matter. The problem with books like this, or, to be more accurate, the authors of books like this, is that they're only really good at preaching to the choir. You might expect that the rigorous Knuth of the "Art of Computer Programming" complexity analyses might avoid this trap, but he doesn't. If you don't already find mysticism interesting, you won't after reading this book. Really, this book is interesting only in the same way that Buckminster Fuller's "Critical Path" is interesting, only it's rather less funny because Knuth's crackpot side is so much more commonplace than Fuller's: the friend I gave my copy of "3:16" has (to an outsider) the same beliefs as Knuth, but I'm pretty sure I've never known anyone quite like Fuller.
If you're a stoner or a drunken student, you might like this book. Then again, if you are either of those things, you probably get enough of this kind of blather every time you sit down with your friends. Sound and fury, signifying nothing. It's not even as forehead-slappingly "let them eat cake" wrong-headed as C.S. Lewis' "Mere Christianity" (a book I usually think of as in many ways the opposite of Aldous Huxley's "Brave New World") because, as far as I could make out, Knuth and his friends don't really say anything.
Funnily enough, I've read Star Wars fiction that gave me more to think about.
I promised myself a while back that I'd stop writing reviews of books and films that weren't directly related to computing, but since I've just saved you $20, I have a couple of suggestions of what you might spend it on. If you want (often religious) craziness tranformed into art, try "A Scanner Darkly". For once this adaptation really does Dick justice, even to the extent of poking gentle fun at him (or taking on his style so completely that the invented bits were totally convincing). If you're more of a book person, you could read the original book, or you might prefer T.E. Lawrence's "Seven Pillars of Wisdom", about Lawrence's time fighting in the desert in WWI. It took me two months to read, but it was worth every page. And it coincidentally contains some of the most interesting writing on religion I've read in years.
Plus shit gets blown up every couple of chapters, and you can't buy quality like that.