Can Failure Be an End in Itself?
Science is constantly being revised, and failure is crucial to the process.
The following is an excerpt from Failure: Why Science is So Successful, by Stuart Firestein.
A real failure does not need an excuse. It is an end in itself. —Gertrude Stein
I have chosen this deceptively simple sounding statement, so typical of Gertrude Stein, to open this book because it gets so quickly to the heart of the matter. It challenges, right from the beginning, our idea of what a failure may be. What kind of a failure is Stein talking about here? What makes a “real” failure? Are there “unreal” failures, or lesser failures?
Like so many important words, failure is much too simple for the class of things it represents. Failure comes in many flavors, and strengths, and contexts, and values, and innumerable other variables. Nothing just stands alone as a failure without knowing something more about it. In the famous Encyclopédie of the French enlightenment, Diderot and d’Alembert (1751–1772) under the entry for erreur, which seems intended to cover failure as well, caution that there is no way to develop a general description or classification because erreur comes in so many forms. I started this project with what I thought were a few clear ideas about failure and its value in the pursuit of scientific explanations. What surprised me was how quickly those few ideas generated dozens of questions.
There is a continuum of failure, not just one narrow kind. Yes, there are failures that are just mistakes or errors, and they may often be no more than an unfortunate waste of time. There are failures from which you learn simple lessons: be more careful, take more time, check your answers. There are failures that can be taken as much larger life lessons: a failed marriage, a failed business venture; painful but perhaps character building. There are failures that lead to unexpected and otherwise unavailable discoveries: they often seem like serendipity, an accidental failure that opened a door you didn’t even know was there. There are failures that are informative: it doesn’t work this way; there must be some other way. There are failures that lead to other failures that eventually lead to some kind of success about learning why the other paths were failures. There are failures that are good for a while and then not—in science you might think of alchemy, a failure that nonetheless provided the foundations of modern chemistry.
Failures can be minimal and easily dismissed; they can be catastrophic and harmful. There are failures that should be encouraged and others that should be discouraged.
The list could go on. But I don’t want to get sidetracked into a lengthy polemic trying to define failure, which would surely fail. We’ll come upon all sorts of failures as we proceed, and we would do best to think of them as discoveries, not contradictions. Rather, I want to focus on the role that failure, in all its many identities, plays in science and how it contributes to making it such a successful enterprise.
Stein seems to be complaining about the common response to a failure—which is apology. Failure as mistake, unintended or unavoidable or because of some shortcoming that you are responsible for. Failure as the result of stupidity and naiveté that requires excuses and apologies. Why did you let that fail? Can’t you do any better than that? Or, perhaps less antagonistic but no less disappointing, failure as inevitable. Well, that wasn’t likely going to work. What did you expect? What a stupid thing to have even tried. And so forth. Stein, in that first simple sentence, identifies all these bad failures, useless failure, failures that demean failure.
Instead, how about failure that stems not from ineptitude, inattention, or incapacity. (True, even those occasionally turn out to reveal something unexpected and sometimes wonderful. But I wouldn’t depend on them. Sloppy indifference can get you only so far.) A real failure is different from all those that need or are accompanied by an excuse—because it needs no excuse.
So what are good failures? Ones that need no excuse and are an end in themselves? Not really an end in the typical sense—that is, not an end where you give up trying anything else. Rather an end in the sense of something new and valuable. Something to be proud of and therefore requiring no excuse, even if it was “wrong.”
Are there really such failures? Of course there are the mistakes we learn from, the errors that can be corrected, the failures that can be turned to success. But I’d like to take a chance here and venture that Stein meant something deeper than that. That she really meant meaningful failure. In the limit, this could mean that you might produce nothing but meaningful failures for your entire life and still be counted a success. Or at least never need to apologize. Is that really possible? What are these magical failures?
I have two possible answers. The first is that failures that are ends in themselves are interesting. Interesting is another word that one has to be careful about. It’s easy to use, but then it’s kind of vague and subjective. Is there anything that’s interesting to everybody? I doubt that. But if we take interesting as a descriptor rather than an identifier—that is, a quality of something and not necessarily a particular thing itself—then we can perhaps come to an understanding.
When the same Gertrude Stein was asked to write a piece about the atom bomb (shortly after its use in WW2 and, as it turns out, shortly before her death in 1946), she responded that it held no interest for her. She liked detective stories and that sort of literature, but death rays and super weapons were not that interesting because they left nothing behind. Someone sets off a bomb or some weapon of mass destruction that kills everybody and ends everything. So what’s to be interested in? Certainly better if it didn’t happen, but if nothing is all you’re left with, then who cares? So maybe it’s what’s left that could make something an interesting failure. Good failures, we could call them Stein Failures, are those that leave a wake of interesting stuff behind: ideas, questions, paradoxes, enigmas, contradictions—you know what I mean. So that’s one kind of successful failure I’m pretty sure about.
Here’s the second idea. Is it the actual failure that is the end in itself? Or is it the willingness to fail, the expectation of failure, the acceptance of failure, the desirability of failure? Can you imagine making failure desirable? Can you imagine aiming at failure? Can you appreciate making failure your goal?
You can if you have the right idea about the word failure—what I hope to convince you is the scientific version of failure. It is more than a stupid error, more than a shortcoming on your part, more than a miscalculation, more even than a chance to improve. Yes, more even than failures as life lessons. I know we all believe that a failure can be valuable if you learn something from it. After all, that’s what we call experience. But how about a failure that does not aim at later self-improvement? How about failures that really are an end in themselves?
In this sense virtually all of science is a failure that is an end in itself. This is because scientific discoveries and facts are provisional. Science is constantly being revised. It may be successful for a time; it may remain successful even after it has been shown to be wrong in some essential way. That may seem strange, but good science is rarely completely wrong, just as it is never really completely right. The process is iterative. We scientists hop from failure to failure, happy with the interim results because they work so well and often are pretty close to the real thing.
Newton was famously wrong about two little things—time and space. They are not absolute. Gravity is not explained by the attraction between the centers of massive bodies, although it looks that way and can be usefully described that way. To the extent that we can explain it at all, it seems to be best understood, for now, as an emergent phenomenon of mass creating curvature in space. An imperfect but useful analogy is the way a heavy bowling ball on a mattress causes a depression and things placed on the mattress tend to fall toward it, as if they were being attracted to it. But Newton’s failure in that one regard, even though it seems like a fundamental part of the theory of gravity, is not at all fatal to the success of his work. His equations quite accurately describe action at a distance between two bodies—sufficiently well to calculate how to dock a rocket with a space station orbiting some 250 miles away and moving at a speed of 17,000 miles per hour.
Nonetheless, there was a nagging inconsistency in Newton’s model over what appeared as two different kinds of gravity. This inconsistency was what needled Einstein so much that he was ready to take a most unintuitive, illogical perspective. Although it’s not exactly how Einstein thought about it, these two kinds of gravity are most easily experienced as the loss of gravity—weightlessness. One of them can be felt as distance from a massive body (the weightlessness experienced in outer space), and the other is due to acceleration (the weightless feeling you would have in a rapidly dropping elevator). They seem to be from two different and unrelated causes—the mass of a nearby body and the force resisted by inertia, or acceleration. Two hundred and fifty years later Einstein essentially corrected the failure of that part of Newtonian mechanics by showing that in the correct inertial frame, one that does not assume absolute time or space, the two kinds of gravity are the same.
Granted, it turned out to be a rather major correction, requiring a Copernican-sized shift in our point of view. But as with Copernicus it didn’t require throwing everything else out. We continue to live our everyday lives in a Newtonian world where space and time seem sufficiently absolute, just as we continue to live most of our lives in a pre-Copernican world where the sun “rises” and “sets.” That oversimplifies the story a great deal (see Notes), but the point is that Newton was successfully wrong and it was the very failed part of his model that led to Einstein’s remarkable insights. Pretty good work.
A failure can be even less successful—that is, wholly incorrect—and still useful. An example from biology might be the longstanding principle known as “ontogeny recapitulates phylogeny.” This tongue twister of a phrase, coined in 1866 by the “father of embryology,” Ernst Haeckel, is simply a slightly bizarre attempt at making a complicated concept memorable by forming a jingle about it. It means that over the course of its development an embryo in the egg (or uterus) appears to go through all the stages of evolution of that organism. For example, mammals early in embryonic development have what appear to be gill-like structures, making them look a bit like fish. These structures eventually develop into our jaws and other muscle and bone groups of our heads and throats but have nothing directly to do with respiration, as gills do for fish. In fact, Haeckel’s concept is completely wrong, even though it held sway for decades and led to many advances in embryology. Not only is it wrong about embryology, it is wrong about evolution. We didn’t evolve from fishes (or apes for that matter); we shared a common ancestor that evolved into both of us, in the case of fish some 500 million years ago, and in the case of apes only about 85 to 90 million years ago.
Nonetheless, this failed ontogeny-phylogeny concept gave rise to important ideas about how development proceeds in clearly established stages, and that structures do evolve from earlier forms, possessing a common ancestry even if a contemporary divergence. Haeckel’s work was painstaking and actually started the branch of science we today call embryology. In particular, he introduced comparative anatomy and development—that is, the notion that we can learn a great deal by making comparisons across species. This showed crucially that not only were species related but that their development proceeded in a similar way along certain principles. The value of this “failure” to modern biology cannot be overestimated. On the other hand, it remains damaging in that there are many people who still believe in it because they were taught it as schoolchildren. You remember, the silly business about having had a tail when you were an embryo.
You could object that Newtown’s and Haeckel’s failures eventually led to successes and were not therefore really ends in themselves. I think that’s too much to ask of failure. Failures like these not only lead to greater insights, they often lead to very unpredictable insights. They force us to look at a problem differently because of the particular way in which they failed. This could be considered the case with Einstein’s recognition that Newton’s little failure was actually a fundamental misconception about time and space. We expect success to lead us to even greater success. What may not be so obvious is that failure can do the same.
These then are what I would call the failures that need no excuse, that stand shoulder to shoulder with success. They are the packing material, the innards, of science, and not giving them their full due is to miss more than half of what science is about and how it works. The big job I hope to do here is to remedy that.
Excerpt from Failure: Why Science is So Successful, by Stuart Firestein. Oxford University Press, 2015. Copyright © 2015. Reprinted with permission.
Stuart Firestein is the author of Failure: Why Science Is So Successful (Oxford University Press, 2015) and a neuroscience professor at Columbia University. He is based in New York, New York.