ON MAY 8, 1842, A TRAIN loaded with passengers from Versailles to Paris in the wake of a birthday celebration for King Louis Philippe derailed unexpectedly, crushing passengers in a horrific accident that killed over fifty people — one of the earliest mass casualty train wrecks. During the investigation into the crash, experts determined that a front axle had suddenly fractured, though there was nothing inherently wrong with either its manufacture or its use. It was not the first time this phenomenon had been noted: three years earlier, the French engineer Jean-Victor Poncelet had struggled to describe how certain kinds of metal springs, in theory perfectly sound, could sometimes break without warning. Searching for a word to describe this malfunction, Poncelet wrote that these metal springs were susceptible to “fatigue.” Other engineers soon adopted this term: sometimes metal, like people, just got tired and gave up.
Poncelet, like others of his time, was borrowing from the lexicon of nineteenth-century medicine, which had recently developed the concept of neurasthenia, or nervous fatigue. Originally a physical ailment, neurasthenia had lately become a catchall term for malaise, apathy, and other amorphous symptoms: a new epidemic of tiredness, an inexplicable breakdown in a worker’s ability to be productive. The term “metal fatigue,” in those years before stress fractures could be measured and studied, suggested an hysteria of engineering, as mysterious to experts as psychological hysteria itself.
It’s no accident, of course, that these two kinds of bedeviling fatigue first presented themselves as problems in those early decades of industrialization. As George Beard wrote of neurasthenia in 1881, “The chief and primary cause of this development and very rapid increase in nervousness is modern civilization, which is distinguished from the ancient by these five characteristics: steam power, the periodical press, the telegraph, the sciences, and the mental activity of women.” The logic of early industrial capitalism meant pushing the limits of what a body was capable of doing; nervous fatigue was, in many ways, the obvious and natural reaction to the increasing mechanization of the body. Just as psychologists labored to ameliorate these effects through talk therapy and pharmaceuticals, so too did engineers work hard to solve the problem of metal fatigue. In the decades after the Versailles derailment and other high profile disasters, engineers like August Wöhler made great progress in describing and predicting metal fatigue, introducing the concept of a “fatigue limit” as a means of guaranteeing that metal wouldn’t be pushed past its functional threshold and lead to disaster.
And yet, in the decades since Wöhler’s fatigue limits were established, the list of disasters due to metal fatigue has continued to grow, including dozens of plane crashes and train wrecks — nearly all of which, in theory, should have been preventable. After each such crash, the National Transportation Safety Board (NTSB) or a similar governmental body does an analysis and makes recommendations for safety improvements. And yet these improvements and recommendations buy us, at best, a few more years: inevitably, the same failures occur, and the same foreseeable tragedies dominate the headlines once again.
Why haven’t we learned to prevent these accidents, when their causes are so well understood? This is the question motivating Henry Petroski’s To Forgive Design: Understanding Failure — not simply Why do things fail?, but Why do they continue to fail, after we’ve supposedly learned our lesson and should know better?
Petroski, a structural engineer and professor at Duke University, has written on engineering and design for the past three decades, but he is probably best known among the general public for his brilliant histories on minor, ordinary objects, including The Pencil: A History of Design and Circumstance and The Toothpick: Technology and Culture. Petroski’s intended audience here is a more specialized one: he speaks primarily to engineers, for the simple reason that much of this book is geared towards preventing future catastrophes like those he discusses in the book. This focus can sometimes lead to an assumption that the reader is already well versed in engineering history (the collapse of the Kansas City Hyatt Regency’s walkway in 1981 is cited throughout without much context), and thus his descriptions can leave one wanting for more colorful details or more of a storyteller’s grace. Nonetheless, To Forgive Design remains a largely accessible, important contribution to the growing library of failure.
Whereas Petroski’s first book, To Engineer is Human: The Role of Failure in Successful Design, focused on failure in engineering and how to avoid it, it generally stayed away from the human element. Twenty-five years later, To Forgive Design focuses on people’s interactions with the machines they build. Understanding metal, glass, and concrete only goes so far, since the things built from them are all imagined, designed, built, and maintained by humans; as such, they are doomed to fail. “All things,” Petroski writes, “and especially systems in which people interact with things, fail because they are the products of human endeavor, which means that they are naturally, necessarily, and sometimes notoriously flawed.” Our metals, ourselves. The metaphoric connection between structural engineering and human psychology made almost two centuries ago by Poncelet is more than just linguistic, since how we build stems from who we are.
Petroski offers an increasingly depressing litany of disasters and their preventable causes. The 2,800-foot span of the Tacoma Narrows Bridge was designed as a state of the art rival to the George Washington Bridge in New York (3,500 feet) and the Golden Gate Bridge (4,200 feet), but because it was designed for two lanes of traffic instead of six or eight, it was unusually flexible; after its opening in 1940, “cars driving over the bridge could see the traffic ahead of them rise and fall and thereby pass in and out of view as the road undulated up and down as if it were floating on waves of the sea.” After only four months the bridge collapsed in high winds, resulting, luckily, in only one animal fatality (a dog trapped in an abandoned car). Commuters of the Point Pleasant Bridge that connected West Virginia and Ohio were less fortunate. Overly optimistic engineers had designed the bridge in 1926 to hold cars weighing 1,500 pounds (the weight of a typical Model T), not expecting the weight of the average car to triple by 1967, when the bridge suddenly collapsed, killing 46 people. The flaw that led to the collapse was later found to be a crack measuring 1/8 by 1/4 inches, located in a chain link in the bridge inaccessible to visual inspection. In both cases, tried and true methods failed unexpectedly due to slight variations in design and the imaginative failure of engineers to anticipate the result of these variations.
In addition to bridges, Petroski discusses construction crane collapses, building failure, and the notorious 1986 explosion of the space shuttle Challenger, which resulted from a faulty O-ring seal — a problem that some designers were well aware of in advance of the shuttle’s launch. And while the O-rings were redesigned for later flights, NASA seemed to learn nothing about the failure of its larger operational culture, leading, seventeen years later, to yet another shuttle loss (Columbia) as a result of yet another well known problem (heat tile damage from falling insulation material during lift-off) that was once again erroneously presumed to be trivial.
Lining up all these horror stories makes clear that, while the causes for each were slightly different, none were anomalies. Why do they reoccur with such depressing regularity? Or, as Petroski puts it,
[I]f technology itself is cumulative, with each incremental improvement adding to our store of engineering knowledge and achievement, then why shouldn’t all newer bridges and oil rigs be better than their predecessors? In particular, why should any ever fail? And why shouldn’t a younger generation of engineers know more and be wiser than the older?
The answer, he explains, lies beyond the easy culprits of hubristic designers, corrupt contractors, careless workmen, and cost-cutting bureaucrats (though these have all played vital parts in various disasters, to be sure). Rather, design failure and engineering catastrophe stems from how we view the world and our own sense of history. In one particularly revelatory example, Petroski discusses a certain exploratory oil well being drilled by a semi-submersible rig in 1979 in the Gulf of Mexico that lost pressure and blew out, leading to a massive oil spill of several million barrels of oil. The well was called the Ixtoc I, and the drilling rig the Sedco 135-F. It’s an event that few remember today, even though it eerily foreshadows the Deepwater Horizon catastrophe of 2010. Why was the Ixtoc I forgotten? It turns out it wasn’t; in the wake of that explosion, the industry tightened regulations, added new precautions and tightened fail safes, and made significant progress with regards to safety and environmental care. But in time, as Petroski points out, and “with a growing record of successful drilling for and extraction of oil from Gulf waters, oil rig and well operations grew lax, and this produced the kind of climate that set the stage for the Deepwater Horizon explosion and subsequent environmental catastrophe.”
This was not entirely unexpected, Petroski claims: “It was no accident that these two unfortunate events occurred about thirty years apart, for that is about the span of an engineering generation and of the technological memory for any industry.” We do learn from our mistakes, but the problem is that all we pass on to the next generation are the lessons, not the mistakes themselves, nor the process by which we learn from them. Learning is a dynamic, fluid, and active process, not a one-time fix. Furthermore, we seem to be incapable of passing along the urgency of a past disaster to a newer generation. “The problem,” Petroski explains,
is exacerbated when the engineer is a generation removed from the inception of a design method or structure type. Without a knowledge of the fundamental design assumptions that were made during the development stage — assumptions that implicitly if not explicitly limit the applicability of the method or structural type — younger engineers can be effectively designing blind in uncharted territory, all the while thinking they are just following a well-worn path.
In other words, we can only progress as far as the memory of our last failure.
Most books on failure take one of two approaches. The first can be found in any of the glut of business-guru/self-help books out there: Ralph Heath’s Celebrating Failure: The Power of Taking Risks, Making Mistakes, and Thinking Big and Charles C. Manz’s The Power of Failure: 27 Ways to Turn Life’s Setbacks Into Success, to name only two of many. As their subtitles suggest, these books aren’t really about failure; they’re about success. Heath’s book bills itself as “the definitive how-to manual for leaders seeking to embrace the power of failure as a learning tool to improve their organizations and achieve ever-greater goals,” while Manz teaches us the management mantra “Fail to Succeed.” Failure, for these writers, is only ever a means to greater and greater success, a way station on the journey to riches and happiness.
At the other end of the failure literature spectrum are those who take their cue from Samuel Beckett’s line, “fail more, fail better.” Beginning with the queer feminist collective LTTR’s 2004 publication “Practice More Failure,” and continuing through Judith Halberstam’s 2011 The Queer Art of Failure, these artists and writers have embraced failure as an end in itself, an alternative of the cult of success intrinsic to American capitalism. Arguing that “success in a heteronormative, capitalist society equates too easily to specific forms of reproductive maturity combined with wealth accumulation,” Halberstam instead celebrates failure as a means “to escape the punishing norms that discipline behavior and manage human development.” Here, failure is a cool “fuck you” to success, a mantle to be claimed by freaks, punks, queers, weirdos, and anyone else rejecting the mainstream goals of marriage, kids, a fulfilling job, and a house in the suburbs.
On first glance, Petroski’s book would seem to belong squarely in the former, more practical and success-obsessed category — after all, his goal is to keep bridges from falling and planes from crashing: hardly a countercultural undertaking. But understanding failure, for Petroski, means not using it as a springboard to succeed, but flat-out rejecting the failure/success dichotomy. In the world of engineering, he suggests, it may be more prudent to reject the concept of “success” altogether. He proposes, as an example, an alternate history of the Titanic: suppose for a moment it hadn’t met that iceberg on April 15, 1912 — what then? “The more times the Titanic crossed and recrossed the ocean,” he imagines,
the more confident the ship’s captain, owner, and potential passengers would have become in its extraordinary seaworthiness. Competing steamship companies would likely have wanted to emulate the Titanic’s success, but they would also have wanted to make distinguishing changes that they believed would be improvements, whether for technical, economical, or commercial advantage. Larger, faster, and more opulent ocean liners would then likely have been designed and built. To make them more competitive financially, the newer ships would have been made with thinner hulls and carried fewer lifeboats. After all, the design of their new ship was based on the unsinkable, unsunk, and thus eminently successful Titanic.
For engineers to think of their work in terms of its eventual success, Petroski concludes, is dangerous. This is the central argument of To Forgive Design; because all engineering and design has hidden flaws that we can only understand after the fact, we cannot afford to think in terms of success at all. This is not a gesture of pessimism or nihilism, but an invitation to perpetual vigilance. As one engineer quoted in the book puts it, “every success sows the seeds of failure. Success makes you over-confident.”
Another title for To Forgive Design, then, might be Against Over-Confidence. And, while Petroski’s book may primarily be about engineering, in this it speaks to a much larger swath of human endeavor. It may be the case that we need to rewrite human history not as a forward march of progress but as a string of once and future failures. This entails a willingness to mine the past for its myriad dead ends, mistakes, calamities, and embarrassments — and a willingness to keep them in a constant dialogue with the present. It means recognizing that history offers not one simple thing to teach us but a cacophony of contradictory things all at once, and that the work of learning from the past involves the slow, sympathetic sifting-through of this cacophony. Such a task is neither easy nor small, to be sure, but then, failure belongs neither to business gurus peddling the next big success, nor to the queer and the cool, but to all of us.