To Engineer is Human
By Henry Petroski
St. Martin's Press, 1985
Petroski is known for detailed accounts of the material word -- perhaps best exemplified by The Evolution of Useful Things. To Engineer is Human is his first book, and it is somewhat more literary than many of his later books. It still contains lots of good information and interesting history, but the meat is hidden behind a good helping of engineering philosophy.
While this is a book about engineering, and particularly structural engineering, it takes a good amount of time to get into the meat of the subject matter. There are some excellent technical chapters (although nothing taxing to the lay reader), but there are nearly as many poems, personal anecdotes, and especially, constantly, extended analogies. Petroski is transparently convinced that his subject matter is both very important and deadly dull, and in the early chapters and he struggles to make his deep and important points meaningful to the reader.
Which is great if you are an English major who doesn't really care about bridges, but torture if you were actually curious about structural engineering. But he does want certain points hammered firmly home, and hammered they are. So what are his points? Primarily, that good design evolves over time; that society provides much of the selection pressure; and that these facts are very familiar to all of us through our interactions with everyday objects. Unfortunately, his way of convincing us that things are familiar to us are pages of descriptions of our everyday lives and frequent references to literary works.
But he does get past that, and we get to hear about bridge failures, airliners breaking up, the problems with designing nuclear reactors and buses, and famous structural failures (and a few successes) of all sorts. While I was seriously considering giving up on this book during the first few chapters, it picked up speed about chapter eight and was well worth it after that point. It does read primarily like a history of engineering, simply because it was written in 1982, and his modern examples are almost 40 years out of date. This is not a bad thing, and he had originally intended for it to be a review of the last ~100 years into modern times, so the historical slant is largely intentional. Overall, this is as much a feature as a bug.
I highly recommend this book to anyone interested in the history of structural engineering, but I also recommend being ready to skim over -- or perhaps even skip -- the first eight chapters. The fact that I am recommending this book after trashing the first 80 pages should serve as an indicator of how interesting the later 150 pages are.