Writing a science-type book for general consumption is a tricky endeavor. Too technical and you lose the layman audience; too pedestrian, and the scientific audience fades. There's a lot to like in Henry Petroski's "To Engineer is Human" but this book takes off on tangents that are meant to illuminate, but only serve to exasperate.

The main thesis of the book is that engineering failures can teach us how to build better bridges, buildings, and machines. Petroski shines when he sticks to this theme, as he does in compelling detail over the Tacoma-Narrows bridge collapse, the skyway in the Kansas City Hyatt Regency Hotel, and the failure of the de Havilland Comet aircrafts in the mid-1950s. These accounts are rich in detail and reveal the highly specialized skill of forensic engineering, or, figuring out what went wrong.

But Petroski strains the reader's patience by belaboring the point in long passages about the roles and responsibilities of engineers to learn from mistakes and strive towards better designs. This is abundantly obvious throughout the book - why beat the drum over and over? Further, the author seems intent on showing his literal side, quoting mythology, Robert Frost, and Oliver Wendell Holmes. The purpose of these passages seem tenuous and forced.

As a final nit-pick, the paperback version I picked up had a photo of the space shuttle on the front and I was expecting a forensic report on what was arguably the worst engineering failure of the last 50 years. But the book was published before the 1986 disaster and there's only a glancing review in the "Afterword" section of the 1992 version that I picked up. In sum, the book would have been better suited to a long magazine article in "Scientific American" sans the moralizing.