To Forgive Design: Understanding Failure by Henry Petroski (review)

As he explains in an autobiographical chapter of To Forgive Design, Henry Petroski entered the history of technology indirectly, as a specialist in a field between physics and engineering called continuum mechanics, which he was originally hired to teach at Duke University. At Argonne National Laboratory he had become a group leader in the still-new field of fracture analysis. A historic failure like the ones he was studying gave him what proved to be a fortunate break: declining funding after the Three Mile Island nuclear incident in 1979 encouraged his move back to academia.

The present book revisits the themes of Petroski’s first, To Engineer Is Human (1985), with a great range of fresh examples. Failure has a special place in technology, according to the author. The only way to test a design thoroughly is to use it. All design is a compromise; every structure and device may have latent weaknesses that the best theoretical tools can’t reveal, especially since the conditions of use change. We usually discover these flaws only through failures. Mostly these are small problems leading to incremental improvements, but tragedies motivate some of the greatest innovations. Engineers may be responsible for disasters, but they are not necessarily blameworthy; one chapter is called “Things Happen.” Engineers’ duty is to investigate failure and develop means to prevent the same one in the future, yet these modifications may result in new failures.

History and memory are essential for engineers: for example, the research of British academics Paul G. Sibly and Alistair C. Walker in the 1970s revealed a thirty-year cycle of bridge failures, in which disasters encouraged development of newer and apparently safer designs, which a new generation of engineers advanced on an increasing scale, until a latent weakness (often coupled with unforeseen conditions) resulted in a new catastrophe, starting the cycle over again. Petroski believes that the cable-stayed bridges popular since the 1970s may be vulnerable; retrofitted safety features like cable ties and shock-absorbing dampers may be encouraging a false sense of confidence such as that which possessed engineers in the late 1930s regarding suspension bridges. The collapse of the Tacoma Narrows Bridge in 1940—fortunately with no human casualties—revealed that civil engineers had been neglecting aerodynamic forces.

Memory, Petroski argues, often must go back far beyond thirty years. The substitution of two offset rods for a single continuous steel rod supporting the elevated walkways of the Kansas City Hyatt Regency Hotel resulted in the deaths of 114 guests in 1981; it might have been prevented, Petroski suggests, had the engineers been aware of a story told by Galileo of how adding an additional support to a marble column in storage led to the column’s fracture as the supports at each end sank into the soil.

Technological, social, and political changes and user behaviors also can disrupt the best engineering calculations. Bucksport, Maine’s Waldo-Hancock suspension bridge of 1931, originally acclaimed for its economy and beauty, aged prematurely not only because of saltwater corrosion of its cables but because of an unexpected postwar growth in heavy truck traffic. The grandest of 1930s bridges, the Golden Gate, has remained structurally admirable, but because of its beauty and site has become fatally attractive to potential suicides. Even the Golden Gate began to sway potentially hazardously under a heavy pedestrian load at its fiftieth anniversary. Petroski warns of the manifold perils of today’s exuberant footbridge innovations.

Petroski also points to the complex network of organizations that manage the largest engineered objects, for example the Deepwater Horizon drilling rig in the Gulf of Mexico that caught fire and sank, releasing tens of thousands of barrels of petroleum a day. The problem was partly in the rig’s original design but mostly in the failure of organizations to learn from smaller incidents to avoid shortcuts and false economies under pressure. A similar story could be told of the space shuttle disasters.

To Forgive Design offers no methodology, beyond professional...