Episode Summary
Executive Summary: This Hidden Brain episode explores the science of failure with Harvard Business School professor Amy Edmondson. It challenges the notion that all mistakes are bad, introducing a taxonomy of failures: basic, complex, and intelligent. Through examples from healthcare, space exploration, and business, Edmondson argues that organizations and individuals should not aim to eliminate all errors but instead learn to distinguish between harmful failures and those that drive innovation and growth.
Main Topics: The Taxonomy of Failure: Edmondson categorizes failures into three types: basic (due to inattention), complex (multiple factors aligning), and intelligent (experiments in new territory). Each requires a different response. The Importance of Psychological Safety: Edmondson's early research in hospitals showed that better teams reported more errors, not fewer, because they had a climate where mistakes were not held against individuals. This led to the concept of psychological safety. Intelligent Failure as a Driver of Innovation: Intelligent failures occur in new territory, are hypothesis-driven, and are as small as possible. Examples include Thomas Edison's light bulb experiments and blind dates. They are essential for learning and progress. Complex Failures and Systemic Thinking: Complex failures, like the Columbia space shuttle disaster, result from multiple small factors aligning. The Swiss cheese model illustrates how defects in systems can line up to cause catastrophic outcomes. Preventing Basic Failures with Checklists and Mindfulness: Basic failures stem from inattention. Checklists are effective but must be used mindfully, as shown by the Air Florida crash where a checklist item was skipped. The Toyota Production System and the Andon Cord: Toyota's andon cord allows any worker to stop the production line when they see a problem. This prevents small issues from becoming complex failures and invests in quality.
Key Arguments: Demanding error-free performance leads to underreporting of mistakes, not fewer mistakes. Failures should be categorized to determine the appropriate response: basic failures need prevention, complex failures need systemic fixes, and intelligent failures should be encouraged. Intelligent failures are valuable experiments that provide information and drive innovation, even if they don't achieve the desired outcome. Psychological safety is crucial for learning from failures; teams must feel safe to speak up about errors. Complex failures are rarely caused by a single factor but by multiple small defects aligning; addressing any one can prevent the failure. Checklists are powerful tools but require mindful execution to prevent basic errors. Organizations should invest in small, controlled failures (like simulations) to learn without catastrophic consequences.
Data Points: Percentage of hypotheses that fail in Jennifer Heemstra's lab: 90% - Heemstra tells her students that 90% of their well-informed hypotheses will end in failure, which is a normal part of scientific discovery. Number of ways Thomas Edison found that didn't work for the light bulb: 10,000 - Edison famously reframed his failures as discoveries, saying he found 10,000 ways that don't work. Number of stitches Amy Edmondson needed after a sailing accident: 9 - Edmondson's basic failure while sailing led to a head injury requiring nine stitches. Number of factors contributing to a morphine overdose case: 7 - Edmondson analyzed seven factors that lined up to cause a dangerous morphine overdose in a young patient. Number of employees at Taylor & Sons before the typo error: 250 - The company had 250 employees before a government typo led to its eventual liquidation.
Pivotal Quotes: "I haven't failed. I've just found 10,000 ways that don't work." — Thomas Edison (quoted by Amy Edmondson): Edmondson uses this quote to illustrate the mindset of intelligent failure, where each failure is a valuable discovery. "If you make a mistake in this unit, it's held against you." — Amy Edmondson (quoting a survey item): This single survey item was profoundly correlated with actual error rates, showing that psychological safety affects reporting. "When you insist that we must have error-free performance or error-free lives, the main thing that happens is not that error goes away, it's that you stop hearing about it." — Amy Edmondson: Edmondson explains the counterproductive effect of demanding zero errors.
Implications: Listeners should rethink their relationship with failure. Instead of fearing all mistakes, they should learn to distinguish between harmful errors and valuable learning opportunities. Fostering psychological safety in teams and embracing intelligent failures can drive innovation and growth in both personal and professional contexts.
From the Episode
Thomas Edison was dogged in his pursuit of the inventions that he knew or he believed were possible and he knew could really change lives. So he's, of course, credited with inventing the incandescent light bulb. And he tried literally thousands of different materials and techniques that one after another, they didn't work. They had all sorts of problems, but he didn't give up. He just kept going. And when a lab assistant, I think attempting to be supportive, said, Gosh, all these failures must be very hard for you, sir, he responded, failures? I haven't failed. I've just found 10,000 ways that don't work. Yeah, so this is the very definition of intelligent failure here. Yeah, he completely understood the concept of intelligent failure. You have a goal, you've done your homework, you've used the knowledge you have. from your prior experiences, from available literature, from everything you can get your hands on. And then you reach that point where there is no way to go forward other than by trying it out, acting, having an experiment. And then lo and behold, again, you were wrong. But fortunately, you're wrong behind closed doors or you're wrong at a small scale. You haven't caused real harm. You haven't blown anything up.
Can't be true, right? It can't be that better teams with higher quality relationships, more able and willing to collaborate effectively, to coordinate clearly, could have more, not fewer, adverse drug events. And I suddenly had a thought. You know, maybe better teams don't make more mistakes. Maybe they're more able and willing to report them. And I suddenly remembered that I had inserted an extra question that wasn't in the original survey that is stated as follows. If you make a mistake in this unit, it's held against you and it's rated on a seven-point scale from, you know, not at all to a great deal. And it turned out that single item was profoundly significantly correlated with the actual error rates. So that meant that when people agreed with this, Making a mistake would not be held against you. The error rates were higher. That's not a perfect proof, but it certainly suggested that there was something in the climate of the team that would make it easier to speak up about and report error.
That a mistake will occur. So, a better approach is to think about how we can be set up to catch and correct mistakes before they cause harm. And, in some ways, from the hospital study that you were mentioning to me earlier, when you send a signal that failure is not going to be tolerated, what happens then is not that the failure stop, but that the failure stop being reported. Exactly. It's one of these profound insights that I think way too few. Leaders, or even just people in families, recognize that when you insist that we must have error-free performance or error-free lives, the main thing that happens is not that error goes away, it's that you stop hearing about it. There's another approach to failure. In some ways, this is the polar opposite to the first approach. And this idea is popular, or at least used to be very popular, among tech entrepreneurs in Silicon Valley. And this To celebrate failure. Fail fast, fail early was the motto. Was this a better approach than leaders telling employees that failure was unacceptable? I don't think it's possible to say one is better than the other. They are better for different contexts. So the fail fast, fail often is a fantastic approach for a laboratory or for an RD group. The, you know, let's adhere to the highest possible standards.
About Hidden Brain
Why do I feel stuck? How can I become more creative? What can I do to improve my relationships? If you’ve ever asked yourself these questions, you’re not alone. On Hidden Brain, we help you understand your own mind — and the minds of the people around you. (We're routinely rated the #1 science podcast in the United States.) Hosted by veteran science journalist Shankar Vedantam.