Q & A: Learning From Mistakes

Megan Morr

Henry Petroski sees opportunity in catastrophe and all the small mistakes along the way. The Aleksandar S. Vesic Professor of civil engineering and a professor of history at Duke, Petroski has studied and written about failure analysis and design theory for more than twenty years—exploring the ways that engineering failures lead inventors to improve on the mistakes of earlier generations. He's also known for his histories of everyday objects, including the pencil, the paper clip, and the zipper. His latest book, Success through Failure: The Paradox of Design, based on a series of lectures he gave at Princeton University in 2004, covers a familiar theme.

How do you define the terms failure and success with respect to design?

Usually a design has stated goals. Success is when the design satisfies those goals. And failure is when it doesn't. A failure doesn't have to be an absolute collapse.

In your latest book, you talk about some catastrophic collapses: the destruction of the Tacoma Narrows Bridge in 1940; the fall of the World Trade Center towers in 2001. Can you give an example of how a design might fail less spectacularly?

Frank Gehry, the architect, designed the Disney concert hall in Los Angeles. Aesthetically, it's a great success. But one corner of the [polished stainless-steel] building reflected sunlight into a condominium complex across the street that was blinding and also heated up the condominiums by fifteen degrees. That would be an example of something generally working, but there's this unintended consequence. You could say that in the ideal world, that would have been anticipated. And in the future, it's another thing that designers will note what not to do.

Is there such a thing as absolute success in terms of design, or is there always something that can be improved?

I believe there's always something that can be improved. Look at the patent system. We keep getting more patents issued, and if you read a lot of those patents, they're really improvements on prior patents, prior designs. When people design something, even if it's a big team [doing the designing], they tend to develop a tunnel vision. They're looking toward a specific goal. Sometimes they focus on it so much that they forget things that later seem sort of obvious. It's easy to be a critic. But when a flawed design is out there for others to criticize, that's what they'll do.

And, according to your book, that criticism leads to improvement?

Right.

Has "failure" led to success in your own work?

I think I've learned a lot about writing from editors who have basically criticized what I've written or edited it or changed it. I'll ask, "Why'd you do that?" And very often you can have a dialogue with an editor, and he gives a reason that's quite good. If you don't get your ego caught up in it, you learn and you get better.

How did you begin studying failure analysis?

When I wrote my first book [To Engineer Is Human: The Role of Failure in Successful Design, 1985], my real intention was to understand what engineering is. I had been teaching engineering, I had worked as an engineer, and I was registered as a professional engineer. But I still thought that, philosophically, I wasn't sure what it was. People outside of engineering would say, "Do you engineers really understand what you're doing?" There were these plane crashes, there was Three Mile Island, there were other accidents happening. [So this was a] fair question. It got me started thinking about failure. If engineers really do understand what they are doing, how can you explain the fact that failures occur?

The premise of that first book—the idea that designs will inevitably fail and that success can come as a result of learning from mistakes—is very similar to that of Success through Failure, which was published in 2006. How has your understanding of engineering and the design process changed over that span of twenty-plus years?

In my first book, I concentrated almost exclusively on structural engineering—bridges, cathedrals. Even when I talked about children's toys, they were structural issues, like whether they were strong enough and why they broke. Since that time, I've come to think in more broad terms, not only about structural engineering, but also about ideas of success and failure in systems and in products. Not just, "Is something strong enough?" but "Does it work the way it was intended?"

In the book I talk about medicine bottles. You go to the drugstore and you get a prescription filled, and it's in this container with a cap you have to push down and twist [to open]. These things became required when a lot of children were dying of overdoses of medicine that they weren't supposed to be having access to. The idea was, let's make child-resistant packaging—simple as that. But sometimes you can make something almost too good, and it turned out that a lot of older people and arthritic people couldn't open these packages, so they would ask somebody to open them for them, and if they did get them open, they'd leave them open. So these older people would leave open medicine bottles all around, and their grandchildren would come and visit them—same thing was happening. This is not only a question of a machine working—or a design, if you will. But it's got to do with federal regulation and federal mandates. It's got to do with questions of interaction of technology and society, public policy, etc.

When you're studying the evolution of specific objects, like medicine bottles, what kind of sources are you looking at?

A lot of it is from patent literature. Patents provide several things. One is that they define the state of the art when they were written. The terminology is "prior art." It means, what's for sale today. And then the patent goes on, usually, to criticize it and point out all its faults and failings. And basically the inventor says, "I've got a better idea." Sometimes the patents go into wonderful detail and comparative detail. That becomes a principal source.

But then [I also use] the thing itself. You can buy different kinds of paper clips, different kinds of pins and needles. Very often, how they're advertised, how they're packaged also provides hints. It sometimes explains explicitly why this was invented or developed.

You've taken on the pencil, the paper clip, and the zipper in previous works. Are there any other objects that you're looking at now?

Yes. In fact I'm reading proofs for my next book, which is called The Toothpick. This was actually supposed to be a chapter in Success through Failure, and I was feeling pretty good about it. But then when I stepped back and looked at it, I said, "Boy, I don't really understand the history of this thing," and I had a hunch that, "I'll bet you it's an interesting story." And it turns out that it is.

The opening chapter is titled "The Oldest Habit." Basically anthropologists have found fossilized teeth almost two million years old that have curious grooves, striated grooves. And the speculation is that these were caused by the use of toothpicks that were very abrasive, and that were used over an extended period of time. Some anthropologists have conducted actual experiments and have reproduced these grooves.

It seems like you have a never ending supply of subjects to write about.

I haven't run out of ideas yet.