To Engineer Is Human Page 3
The failure of the chair does not keep Goldilocks from next trying beds without any apparent concern for their structural integrity. When Papa Bear’s bed is too hard and Mama’s is too soft, Goldilocks does not seem to draw a parallel with the chairs. She finds Baby Bear’s bed “just right” and falls asleep in it without worrying about its collapsing under her. One thing the fairy tale implicitly teaches us as children is to live in a world of seemingly capricious structural failure and success without anxiety. While Goldilocks may worry about having broken Baby Bear’s chair, she does not worry about all chairs and beds breaking. According to Bruno Bettelheim, the tale of Goldilocks and the Three Bears lacks some of the important features of a true fairy tale, for in it there is neither recovery nor consolation, there is no resolution of conflict, and Goldilocks’ running away from the bears is not exactly a happy ending. Yet there is structural recovery and consolation in that the bed does not break, and there is thereby a structural happy ending.
If the story of Goldilocks demonstrates how the user of engineering products can be distracted into overestimating their strength, the story of the Three Little Pigs shows how the designer can underestimate the strength his structure may need in an emergency or, as modern euphemisms would put it, under extreme load or hypothetical accident conditions. We recall that each of the three pigs has the same objective: to build a house. It is implicit in the mother pig’s admonishment as they set out that their houses not only will have to shelter the little pigs from ordinary weather, but must also stand up against any extremes to which the Big Bad Wolf may subject them.
The three little pigs are all aware of the structural requirements necessary to keep the wolf out, but they differ in their beliefs of how severe a wolf’s onslaught can be, and some of the pigs would like to get by with the least work and the most play. Thus the individual pigs make different estimates of how strong their houses must be, and each reaches a different conclusion about how much strength he can sacrifice to availability of materials and time of construction. That each pig thinks he is building his house strong enough is demonstrated by the first two pigs dancing and singing, “Who’s afraid of the Big Bad Wolf.” They think their houses are safe enough and that their brother laboring over his brick house has overestimated the strength of the wolf and overdesigned his structure. Finally, when the third pig’s house is completed, they all dance and sing their assurances. It is only the test of the wolf’s full fury that ultimately proves the third pig correct. Had the wolf been a bugaboo, all three houses might have stood for many a year and the first two pigs never been proven wrong.
Thus the nursery rhymes, riddles, and fairy tales of childhood introduce us to engineering. From lullabyes that comfort us even as they sing of structural failure to fairy tales that teach us that we can build our structures so strong that they can withstand even the huffing and puffing of a Big Bad Wolf, we learn the rudiments and the humanness of engineering.
Our own bodies, the oral tradition of our language and our nursery rhymes, our experiences with blocks and sand, all serve to accustom us to the idea that structural failure is part of the human condition. Thus we seem to be preconditioned, or at least emotionally prepared, to expect bridges and dams, buildings and boats, to break now and then. But we seem not at all resigned to the idea of major engineering structures having the same mortality as we. Somehow, as adults who forget their childhood, we expect our constructions to have evolved into monuments, not into mistakes. It is as if engineers and non-engineers alike, being human, want their creations to be superhuman. And that may not seem to be an unrealistic aspiration, for the flesh and bone of steel and stone can seem immortal when compared with the likes of man.
3
LESSONS FROM PLAY; LESSONS FROM LIFE
When I want to introduce the engineering concept of fatigue to students, I bring a box of paper clips to class. In front of the class I open one of the paper clips flat and then bend it back and forth until it breaks in two. That, I tell the class, is failure by fatigue, and I point out that the number of back and forth cycles it takes to break the paper clip depends not only on how strong the clip is but also on how severely I bend it. When paper clips are used normally, to clip a few sheets of paper together, they can withstand perhaps thousands or millions of the slight openings and closings it takes to put them on and take them off the papers, and thus we seldom experience their breaking. But when paper clips are bent open so wide that they look as if we want them to hold all the pages of a book together, it might take only ten or twenty flexings to bring them to the point of separation.
Having said this, I pass out a half dozen or so clips to each of the students and ask them to bend their clips to breaking by flexing them as far open and as far closed as I did. As the students begin this low-budget experiment, I prepare at the blackboard to record how many back and forth bendings it takes to break each paper clip. As the students call out the numbers, I plot them on a bar graph called a histogram. Invariably the results fall clearly under a bell-shaped normal curve that indicates the statistical distribution of the results, and I elicit from the students the explanations as to why not all the paper clips broke with the same number of bendings. Everyone usually agrees on two main reasons: not all paper clips are equally strong, and not every student bends his clips in exactly the same way. Thus the students recognize at once the phenomenon of fatigue and the fact that failure by fatigue is not a precisely predictable event.
Many of the small annoyances of daily life are due to predictable—but not precisely so—fractures from repeated use. Shoelaces and light bulbs, as well as many other familiar objects, seem to fail us suddenly and when it is least convenient. They break and burn out under conditions that seem no more severe than those they had been subjected to hundreds or thousands of times before. A bulb that has burned continuously for decades may appear in a book of world records, but to an engineer versed in the phenomenon of fatigue, the performance is not remarkable. Only if the bulb had been turned on and off daily all those years would its endurance be extraordinary, for it is the cyclic and not the continuous heating of the filament that is its undoing. Thus, because of the fatiguing effect of being constantly changed, it is the rare scoreboard that does not have at least one bulb blown.
Children’s toys are especially prone to fatigue failure, not only because children subject them to seemingly endless hours of use but also because the toys are generally not overdesigned. Building a toy too rugged could make it too heavy for the child to manipulate, not to mention more expensive than its imitators. Thus, the seams of rubber balls crack open after so many bounces, the joints of metal tricycles break after so many trips around the block, and the heads of plastic dolls separate after so many nods of agreement.
Even one of the most innovative electronic toys of recent years has been the victim of mechanical fatigue long before children (and their parents) tire of playing with it. Texas Instruments’ Speak & Spell effectively employs one of the first microelectronic voice synthesizers. The bright red plastic toy asks the child in a now-familiar voice to spell a vocabulary of words from the toy’s memory. The child pecks out letters on the keyboard, and they appear on a calculator-like display. When the child finishes spelling a word, the ENTER key is pressed and the computer toy says whether the spelling is correct and prompts the child to try again when a word is misspelled. Speak & Spell is so sophisticated that it will turn itself off if the child does not press a button for five minutes or so, thus conserving its four C-cells.
My son’s early model Speak & Spell had given him what seemed to be hundreds of hours of enjoyment when one day the ENTER key broke off at its plastic hinge. But since Stephen could still fit his small finger into the buttonhole to activate the switch, he continued to enjoy the smart, if disfigured, toy. Soon thereafter, however, the E key snapped off, and soon the T and O keys followed suit. Although he continued to use the toy, its keyboard soon became a maze of missing letters and, for those that were saved fr
om the vacuum cleaner, taped-on buttons.
What made these failures so interesting to me was the very strong correlation between the most frequently occurring letters in the English language and the fatigued keys on Stephen’s Speak & Spell. It is not surprising that the ENTER key broke first, since it was employed for inputting each word and thus got more use than any one letter. Of the seven most common letters—in decreasing occurrence, E, T, A, O, I, N, S, R—five (E, T, O, S, and R) were among the first keys to break. All other letter keys, save for the two seemingly anomalous failures of P and Y, were intact when I first reported this serendipitous experiment on the fatigue phenomenon in the pages of Technology Review.
If one assumes that all Speak & Spell letter keys were made as equally well as manufacturing processes allowed, perhaps about as uniformly as or even more so than paper clips, then those plastic keys that failed must generally have been the ones pressed most frequently. The correlation between letter occurrence in common English words and the failure of the keys substantiates that this did indeed happen, for the anomalous failures seem also to be explainable in terms of abnormally high use. Because my son is right-handed, he might be expected to favor letters on the right-hand side of the keyboard when guessing spellings or just playing at pressing letters. Since none of the initial failed letters occurs in the four left-most columns of Speak & Spell, this proclivity could also explain why the common-letter keys A and N were still intact. The anomalous survival of the I key may be attributed to its statistically abnormal strength or to its underuse by a gregarious child. And the failure of the infrequently occurring P and Y might have been a manifestation of the statistical weakness of the keys or of their overuse by my son. His frequent spelling of his name and of the name of his cat, Pollux, endeared the letter P to him, and he had learned early that Y is sometimes a vowel. Furthermore, each time the Y key was pressed, Speak & Spell would ask the child’s favorite question, “Why?”
Why the fatigue of its plastic buttons should have been the weak link that destroyed the integrity of my son’s most modern electronic toy could represent the central question for understanding engineering design. Why did the designers of the toy apparently not anticipate this problem? Why did they not use buttons that would outlast the toy’s electronics? Why did they not obviate the problem of fatigue, the problem that has defined the lifetimes of mechanical and structural designs for ages? Such questions are not unlike those that are asked after the collapse of a bridge or the crash of an airplane. But the collapse of a bridge or the crash of an airplane can endanger hundreds of lives, and thus the possibility of the fatigue of any part can be a lesson from which its victims learn nothing. Yet the failure of a child’s toy, though it may cause tears, is but a lesson for a child’s future of burnt-out light bulbs and broken shoelaces. And years later, when his shoelaces break as he is rushing to dress for an important appointment, he will be no less likely to ask, “Why?”
After I wrote about the found experiment, my son retrieved his Speak & Spell from my desk and resumed playing with the toy—and so continued the experiment. Soon another key failed, the vowel key U in the lower left position near where Stephen held his thumb. Next the A key broke, another vowel and the third most frequently occurring letter of the alphabet. The experiment ended with that failure, however, for Stephen acquired a new model of Speak & Spell with the new keyboard design that my daughter, Karen, had pointed out to me at an electronics store. Instead of having individually hinged plastic buttons, the new model has its keyboard printed on a single piece of rubbery plastic stretched over the switches. The new model Stephen has is called an E. T. Speak & Spell, after the little alien creature in the movie, and I am watching the plastic sheet in the vicinity of those two most frequently occurring letters to see if the fatigue gremlin will strike again.
Not long after I had first written about my son’s Speak & Spell I found out from readers that their children too had had to live with disfigured keyboards. It is a tribute to the ingeniousness of the toy—and the attachment that children had developed for it—that they endured the broken keys and adapted in makeshift ways, as they would have to throughout a life of breakdowns and failures in our less than perfect world. Some parents reported that their children apparently discovered that the eraser end of a pencil fit nicely into the holes of the old Speak & Spell and thus could be used to enter the most frequently used letters without the children having to use their fingertips. I have wondered if indeed this trick was actually discovered by the parents who loved to play with the toy, for almost any child’s finger should easily fit into the hole left by the broken button, but Mommy or Daddy’s certainly would not.
Nevertheless, this resourcefulness suggests that the toy would have been a commercial success even with its faults, but the company still improved the keyboard design to solve the problem of key fatigue. The new buttonless keyboard is easily cleaned and pressed by even the clumsiest of adult fingers. The evolution of the Speak & Spell keyboard is not an atypical example of the way mass-produced items, though not necessarily planned that way, are debugged through use. Although there may have been some disappointment among parents who had paid a considerable amount of money for what was then among the most advanced applications of microelectronics wizardry, their children, who were closer to the world of learning to walk and talk and who were still humbled by their skinned knees and twisted tongues, took the failure of the keys in stride. Perhaps the manufacturer of the toy, in the excitement of putting the first talking computer on the market, overlooked some of the more mundane aspects of its design, but when the problem of the fractured keys came to its attention, it acted quickly to improve the toy’s mechanical short-comings.
I remember being rather angry when my son’s Speak & Spell lost its first key. For all my understanding of the limitations of engineering and for all my attempted explanations to my neighbors of how failures like the Hyatt Regency walkways and the DC–10 could happen without clear culpability, I did not extend my charity to the designers of the toy. But there is a difference in the design and development of things that are produced by the millions and those that are unique, and it is generally the case that the mass-produced mechanical or electronic object undergoes some of its debugging and evolution after it is offered to the consumer. Such actions as producing a new version of a toy or carrying out an automobile recall campaign are not possible for the large civil engineering structure, however, which must be got right from the first stages of construction. So my charity should have extended to the designers of the Speak & Spell, for honest mistakes can be made by mechanical and electrical as well as by civil engineers. Perhaps someone had underestimated the number of Es it would take a child to become bored with the new toy. After all, most toys are put away long before they break. If this toy, which is more sophisticated than any I ever had in my own childhood, could tell me when I misspelled words I never could keep straight, then I would demand from it other superhuman qualities such as indestructibility. Yet we do not expect that of everything.
Although we might all be annoyed when a light bulb or a shoelace breaks, especially if it does so at a very inconvenient time, few if any of us would dream of taking it back to the store claiming it had malfunctioned. We all know the story of Thomas Edison searching for a suitable filament for the light bulb, and we are aware of and grateful for the technological achievement. We know, almost intuitively it seems, that to make a shoelace that would not break would involve compromises that we are not prepared to accept. Such a lace might be undesirably heavy or expensive for the style of shoe we wear, and we are much more willing to have the option of living with the risk of having the lace break at an inopportune time or of having the small mental burden of anticipating when the lace will break so that we might replace it in time. Unless we are uncommonly fastidious, we live dangerously and pay little attention to preventive maintenance of our fraying shoelaces or our aging light bulbs. Though we may still ask “Why?” when they break, we already k
now and accept the answer.
As the consequences of failure become more severe, however, the forethought we must give to them becomes more a matter of life and death. Automobiles are manufactured by the millions, but it would not do to have them failing with a snap on the highways the way light bulbs and shoelaces do at home. The way an automobile could fail must be anticipated so that, as much as possible, a malfunction does not lead to an otherwise avoidable deadly accident. Since tires are prone to flats, we want our vehicles to be able to be steered safely to the side of the road when one occurs. Such a failure is accepted in the way light bulb and shoelace failures are, and we carry a spare tire to deal with it. Other kinds of malfunctions are less acceptable. We do not want the brakes on all four wheels and the emergency braking system to fail us suddenly and simultaneously. We do not want the steering wheel to come off in our hands as we are negotiating a snaking mountain road. Certain parts of the automobile are given special attention, and in the rare instances when they do fail, leading to disaster, massive lawsuits can result. When they become aware of a potential hazard, automobile manufacturers are compelled to eliminate what might be the causes of even the most remote possibilities of design-related accidents by the massive recall campaigns familiar to us all.
As much as it is human to make mistakes, it is also human to want to avoid them. Murphy’s Law, holding that anything that can go wrong will, is not a law of nature but a joke. All the light bulbs that last until we tire of the lamp, all the shoelaces that outlast their shoes, all the automobiles that give trouble-free service until they are traded in have the last laugh on Murphy. Just as he will not outlive his law, so nothing manufactured can be or is expected to last forever. Once we recognize this elementary fact, the possibility of a machine or a building being as near to perfect for its designed lifetime as its creators may strive to be for theirs is not only a realistic goal for engineers but also a reasonable expectation for consumers. It is only when we set ourselves such an unrealistic goal as buying a shoelace that will never break, inventing a perpetual motion machine, or building a vehicle that will never break down that we appear to be fools and not rational beings.