Making the Most of Your Mistakes

0:00This is Hidden Brain. I'm Shankar Vedantam. In 2009, British businessman Philip Davison Sebrey was celebrating his wife's 50th birthday in the Maldives when he got a phone call. The caller asked for a business meeting the next day at 8 a.m. Philip explained that that would be a little difficult, seeing as he was 4,500 miles away from work, on vacation. What are you doing away at a time like this? The voice at the other end of the line shouted.

0:30Your company is in liquidation.

0:35Philip thought it was a joke in poor taste. In an interview with Wales Online, he recalled that the caller assured him that it was no joking matter.

0:46Here's what happened. A British government agency had reported the demise of Philip's 134-year-old engineering company, Taylor & Sons. The government agency, known as Companies House, serves as a kind of registrar for British businesses. It said that Taylor & Sons, created in 1875, was being shut down. Turns out, a government clerk had made a typo. The company that was going out of business was Taylor & Sons in Manchester, not Taylor & Sons in Wales.

1:22Philip felt sick. His company had been doing well. It had some 250 employees. Within days, he later said in that interview with Wales Online, his contracts dried up. Orders were cancelled. Creditors demanded to be paid. The government agency did correct the mistake after some days, but a death spiral had taken hold. In time, Taylor & Sons actually did go out of business.

1:51Not all errors are so consequential, but some are deadly, and many have unpredictable effects. Wouldn't we all prefer that governments, organizations, and companies avoid making mistakes altogether? That's an understandable response. But it turns out that demanding no errors might be the biggest mistake of all.

2:17Flaws, flubs, and fallacies. This week, on Hidden Brain. Support for Hidden Brain comes from Land Rover. Even the boldest journey starts small, with a single decision to go somewhere new. Purpose-built and courage-driven, the Defender 110 combines on-road presence with off-road capability. It looks tough, because it is, with a raised hood and an exterior engineered for durability.

2:50Inside, capability meets comfort, with seating for 5 and the option for 7, plus refined finishes and thoughtful design. The Defender event is on now. Explore enhanced offers at LandRoverUSA.com. Support for Hidden Brain comes from ADT. Imagine you finally have a day off when your phone buzzes. A window's been broken at home. You're miles away. In one second, everything can change. ADT's security systems help keep your home safer with 24-7 monitoring.

3:21They can be professionally installed and keep watch from day one. Don't wait to prepare for an emergency. When every second counts, count on ADT. Visit ADT.com or call 1-800-ADT-ASAP.

3:38Support for Hidden Brain comes from Progressive. Insurance isn't one-size-fits-all. That's why drivers have trusted Progressive's Name Your Price tool for years. Just tell Progressive what you want to pay, and they'll show you coverage options that fit your budget. Visit Progressive.com to find a car insurance rate that works for you. Progressive Casualty Insurance Company and Affiliates. Price and coverage match limited by state law.

4:11To err is human. When people work on things, mistakes are inevitable. This is true in our personal lives, in our workplaces, and at the level of public policy. Not all mistakes are created equal, though. Some failures are trivial, while others can be disastrous. At Harvard Business School, Amy Edmondson studies how failures come about and what we can do about them. She has surprising insights into how organizations and people should think about the mistakes they make.

4:42Amy Edmondson, welcome to Hidden Brain. Thank you so much for having me. Amy, as a young scientist working on your first major research project, you spent months collecting data from doctors and nurses at two local hospitals. The stakes here were high. I understand that you were tracking medical mistakes? Yes. We call them adverse drug events. So that is when something bad happens to a patient that is deemed caused by human error.

5:13And so I would get a phone call from one of the physicians in the study who would say, there's just been an event, and so we're going to take a look at what happened. And so I would hop on my bike and ride down to the hospital, and I'd find myself in a conference room, and we would then sit around and hear from the perspective of different people who may have touched or been aware of the adverse event, and we try to truly understand what happened.

5:43So for instance, in one case, there was a patient that received a drug called lidocaine, and they were supposed to get a drug called heparin. Now, the two drugs were labeled similarly, and they were in the same location, even though they do very different things. Now, in this case, the lidocaine would not hurt the patient, but the absence of heparin might have led to real harm. It didn't. So these adverse events didn't always involve harm,

6:13but they always involved at least the potential for harm. Now, you became aware that some teams at these hospitals were making more errors, more mistakes than others. But besides tracking mistakes, you also were examining how teams functioned, and you found that some teams functioned better than others. Tell me about the components of teamwork that you measured. Well, I used a classic team survey called the Team Diagnostic Survey, and it measured such things as the quality of interpersonal relationships in the team,

6:47the team's own self-assessment of how well they were performing, the team's assessment of whether and the degree to which they had the resources they need to do their job well, especially interested in their assessment of the leadership of the team and how good was that leadership. And these are a set of factors that had been previously recognized as important to team effectiveness. And you were also looking at how much people cared about their work

7:18and whether people felt like everyone was pulling their fair share of the weight. Yes, and I think that assessment is partly captured in the quality of relationships measure, but there also was the measures of their satisfaction with the work, how happy they were in their job. And all of these measures tend to travel together. So in a good team, in a well-led team, they tend to be high on all of these factors. In a not good team, they're low on all of these factors. So you now have two sets of data,

7:50information on high-performing teams versus low-performing teams, and information on teams that make few mistakes or lots of mistakes. I mean, it seems quite obvious what the answer is going to be, but what was your intuition about how the quality of teams would relate to the mistakes that teams were making? Well, my intuition was that better teamwork would lead to fewer mistakes or adverse events for patients. This just makes sense. And the more I understood about the nature of patient care in a hospital,

8:23the more I realized how interdependent it was. First of all, it's 24-7 operations, so there's shift handoffs. And second of all, most patients are seen by multiple different caregivers through the course of their stay. And so the quality of the coordination and the collaboration ought to really determine the degree to which high-quality care is given. So, of course, I expected better teams would have fewer adverse events or mistakes.

8:55So, the day comes when you're ready to analyze all of the data. You have a small computer disk with all the medication errors made by each team. Paint me a picture of what happened that day, Amy. Well, I synced up the data on the disk with the data that I already had on the team properties in my computer and started to run the statistical analysis to connect those two data sources. And I just ran some simple correlations, and I saw right away that the p-value,

9:31which indicates, you know, whether or not the finding is statistically significant, I saw that the p-value indicated that my data were statistically significant. So, I was very excited. And then I looked more closely, leaning into the screen, and I realized that the sign was in the wrong direction. In other words, instead of seeing a relationship between good teamwork and low error rates, the statistical significance was saying there was a relationship between good teamwork and high error rates.

10:06How could that possibly be, Amy? Well, that was my question, and I sat there. I think I froze. I was upset, and I was scared, and so I sat there looking at it, and my first reaction was, I must have made a mistake. I must have put the data in incorrectly. So, I walked very carefully through everything I'd done. I redid it, and nope. Maybe my hypothesis had been a mistake, but I had not made a kind of data entry mistake.

10:37And so, there I was, staring at it again and again, puzzled. This just can't be true. 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. Maybe better teams don't make more mistakes.

11:09Maybe 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 that single item was profoundly, significantly correlated with the actual error rates.

11:39So, that meant that when people agreed with this item, making a mistake would not be held against you, the error rates were higher. That is not, you know, 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. So, you hired a research assistant to actually go to the hospital and observe firsthand how teams at the hospital were functioning,

12:11testing this hypothesis that better teams, in fact, were more willing to report the mistakes. And that's why they were showing up as having committed greater numbers of mistakes. What did the research assistant find? Well, first of all, I think it's important to point out that I did not tell him what the survey data said nor what the error data said. He only knew that there was a study of error going on. And I said, I just want your impression of what it's like to work in these units. I want you to observe them. I want you to interview them when they're on breaks.

12:42And just, you know, learn as much as you can about these different work environments. And off he went. And after a week or so, he came back. He said, you know, they really are very different places to work. Some of them, he said, his words were just far more open. And others, he said, again, his words were authoritarian in nature. In some units, people would say things like, if you make a mistake in this unit, you get treated like a two-year-old or you get put on trial.

13:15So, you really don't want to have made one. Or the nurse manager, who is essentially the boss of the unit, if you're a nurse, would get angry and treat you badly. In other units, even though they're only maybe across the hall or up a floor or two, he found people saying things like, well, in this unit, it's really easy to talk about mistakes because of what's at stake. You know, patient's health is at stake. So, of course, you're never afraid to speak up and you're never afraid to tell the nurse manager about what you see.

13:46I'm wondering how this insight started to change your thinking about the nature of mistakes and failure, Amy. Well, to begin with, I realized you can't learn from mistakes that aren't reported. When we think about organizations and teams and the goal of learning from mistakes and learning from failures, job one is to make sure we're actually talking about them honestly and openly. And so that started me thinking that maybe there are differences in work environments

14:19in what I called then interpersonal climate. And if the interpersonal climate differs, that would have real implications for people's ability to learn from mistakes and failures.

14:33Amy was starting to see that errors and failure are not always signals of disaster and dysfunction.

14:42When we come back, why failures are not created equal, how to tell them apart, and what we should do about different kinds of failure. You're listening to Hidden Brain. I'm Shankar Vedantam.

15:03Support for Hidden Brain comes from LinkedIn. Running a small business means every hire matters. A bad hire can cost you time, money, and momentum. A good hire, they can help grow your business. LinkedIn's new hiring pro screens candidates for you. So instead of sorting through applicants, you spend time talking to only the right ones. Get started by posting your job for free at linkedin.com slash HB. Terms and conditions apply.

15:33Support for Hidden Brain comes from Defender. Even the boldest journey starts small, with a single decision to go somewhere new. The Defender 110 is a vehicle built for those moments, for drivers capable of great things, whether they're headed toward uncharted territory or just a weekend away. The Defender 110 combines on-road presence with off-road capability. It looks tough, because it is, with an exterior engineered for durability. Inside, capability meets comfort, with seating for five and the option for seven,

16:05plus refined finishes and thoughtful design. It's also packed with intuitive tech, like 3D surround cameras with clear-sight ground view to help you navigate rough terrain, and the next-generation PIVI Pro infotainment system, designed to keep you informed, connected, and in control, no matter the path. The Defender 110 is naturally capable, expedition-ready, and built for those ready to move forward. Explore the Defender 110 at LandRoverUSA.com.

16:37This is Hidden Brain. I'm Shankar Vedantam. When airline pilots make grave mistakes, planes crash and people can die. When surgeons make mistakes, patients can bleed out on the operating table or have the wrong limb amputated. When you leave something in the oven too long, no one dies, but you'll have to eat burnt cinders for dinner. There is a reason parents, teachers, managers, and chefs try so hard to stamp out errors. Mistakes are costly, unpleasant, and dangerous.

17:11At Harvard Business School, Amy Edmondson studies the science of mistakes. She's discovered that we make a big mistake when we lump all failures into the same bucket.

17:23Amy, if you told the average leader of a company that you could completely eliminate all mistakes at her company, she would probably be ecstatic. But you say that we're making a mistake in the way we think about mistakes. Why is it problematic to try to completely eliminate all failure? You said it at the outset, to err is human. We are fallible human beings and we will always make mistakes. I don't mean we'll make mistakes all the time, but there is always the possibility that a mistake will occur.

17:55So a better approach is to think about how can we be set up to catch incorrect 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 failures stop, but that the failures 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

18:27that when you insist that we must have, you know, 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 was to celebrate failure. Fail fast, fail early was the motto.

18:58Was 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 R&D group. The, you know, let's adhere to the highest possible standards and try our very best to get everything right is how we want to run an operating theater.

19:29Right. If you're running an airline company, I'm not sure you want to tell your pilots, fail fast, fail often. Of course not. Imagine being the head of a factory making Toyota automobiles who decides, let's fail fast today. No, no, no, no. So rather than a blanket rejection of failure or a blanket embrace of failure, you say that we need to stop treating all mistakes as if they are the same. And you cite the social scientist Sim Sitkin

20:01who once made the case for something he called intelligent failures. What are intelligent failures, Amy? Intelligent failures are the undesired results of forays into new territory that are driven by a hypothesis and are as small as possible. And so in a sense, an intelligent failure is an experiment that didn't produce the result you had truly wanted it to produce.

20:31In some ways, what that implies then is that intelligent failures are almost always failures that take place on the frontiers of knowledge or discovery. Yes, so there's two kinds of frontiers. One is the frontier of knowledge or discovery in that we're talking about a place where no one has been before. And the other is the frontier that's just new for you. Let's say you pick up a new hobby. You decide to take a ceramics class. That's new territory for you and you can expect some intelligent failures along the way,

21:02even if it's not new to the world.

21:07I want to come back and talk at greater length about intelligent failure later in our conversation. You have a series of very useful insights into how people and organizations can use intelligent failure as an engine for growth and discovery. But it may be helpful before we do that to be able to spot two other kinds of failure. These are the kinds of failure we should, in fact, do our best to stamp out. Now, the first can superficially look like intelligent failure because these failures can also take place while people are engaged in complex tasks on the frontiers

21:38of human knowledge and discovery. In 2003, the space shuttle Columbia broke apart upon re-entry into Earth's atmosphere, killing the seven astronauts aboard. You conducted an in-depth analysis of what happened to the Columbia when it broke apart and it was something that you ended up calling a complex failure. Tell me the story of what happened, Amy. Well, the shuttle had completely combusted on re-entry into the Earth's atmosphere on February 1st, 2003.

22:09And it was later determined that the reason for that is something called a foam strike. Now, what happens is to get the shuttle out beyond the Earth's atmosphere, it takes off with the help of a solid rocket booster. And that has, that sort of has the energy to bring it out into space. And that is surrounded by insulating foam. And occasionally, little bits of that insulating foam would break off just because of the pressure of the launch. And sometimes, those little bits strike the shuttle

22:40and make little dents, you know, just little nuisance problems that would lead to maintenance later on to fix them up. But in this case, there was a rather large piece of foam that dislodged and hit the shuttle on a delicate spot, the leading edge of the wing. And so it unfortunately made a larger hole, a hole the size of a human head. Now, a hole that size in the shuttle as it re-entered the Earth's atmosphere allowed all the hot gases

23:12of the atmosphere in and led to instant combustion. Now, tell me a little bit about what happened in the days leading up to the launch because at least in retrospect, when people went back and did the investigation, they tried to, you know, follow the breadcrumbs and ask, could we have known what was going to happen before it actually happened? Well, yes. And in this case, unlike the even more famous Challenger incident, no one had any worries leading up to the launch,

23:43which was January 16th, 2003. But on January 17th, the day after the launch, an engineer named Rodney Rocha was looking at the launch video and he saw just a grainy speck on the screen that bothered him because he thought that grainy speck might be a foam strike. And the very fact that, he couldn't really make out what it was, but the very fact that he could see a speck at all suggested to him that the chunk,

24:13if it were a foam strike, it might be big enough to do real damage rather than just create a nuisance and a maintenance problem. And so that worried him. Now, that was about 15 days before re-entry. So theoretically, NASA had 15 days to kind of figure out whether there had been a real problem and if so, whether there was an alternate plan to the simple re-entry that was part of the schedule. In other words, could they find out, A, is there really a problem?

24:44And then B, if so, is there anything we could do about it in that 15-day window? But unfortunately, Rocha and his immediate colleagues were never able to get senior managers at NASA to take the problem seriously, to really believe that there was a problem.

25:05Was this partly because there had been other foam strikes that turned out to be fairly minimal, you know, routine maintenance kinds of issues? Exactly. So unfortunately, people at NASA had learned to equate the foam strikes that did happen with just maintenance. They're not a safety risk, but they had had so many of these little, tiny foam strikes that they didn't think it was worth looking into. And in some ways, this is sort of understandable,

25:36even though in retrospect, we know this was a mistake. I mean, if for years, shuttles have been returning safely despite these foam strikes, it's quite understandable how people could have become blasé to them. It's completely understandable. Their own experience had taught them that it was fine. I have enormous empathy for everyone who was a part of that shuttle program who believed it to be fine because I'm that way too. I'm a fallible human being who overly trusts my prior experience

26:07and often fails to be curious enough about, well, maybe this one's different. Ooh, let me look into it. Let me see what I can learn. So it's tragic, but there are no bad guys here. You point out that complex failures are often not the result of one big cause, but rather a number of small factors that line up perfectly in this perfect storm, as you called it. I understand that in the healthcare arena, these kind of complex failures are sometimes called the Swiss cheese model of failure.

26:38Explain that term for me, Amy, and explain how you use this analogy to analyze a case in which a young patient received a dangerous overdose of morphine. That's right. And the Swiss cheese metaphor comes from an error theorist named James Reason from the UK. And he uses this metaphor of Swiss cheese to try to explain the notion of complex failures. He says, you know, when your cheese has air bubbles in it, those are in a sense defects in the cheese, but they're not problematic until they line up

27:10and make a tunnel. Just rarely happens, but when it happens, then the error goes all the way through. So in the case of this morphine overdose, I was able to analyze seven factors contributing to the accident. So to begin with, there had been an overflow in the intensive care unit where most post-surgical patients go, and this boy had just had surgery. So he was sent to the regular medical floor, which has less specialized staff. So that's one factor.

27:40Now, that by itself would not lead to this kind of overdose. But unfortunately, there was a brand new nurse right out of school who was assigned to take care of him. And then there was an infusion pump that's used to deliver this pain medication, and it happened to be located in a rather dark corner, making it a little harder to see. And the nurse also hadn't done this kind of programming before, so he asked for a colleague to help. She stopped by to help,

28:11but she didn't do her calculations independently. She just looked over his shoulder and verified his. And then finally, the medication label was printed badly by IT and a little difficult to read, so that contributed to them not able to determine the concentration of the drug exactly right. And so all of those holes in the Swiss cheese lined up and let this overdose go through. Fortunately,

28:41it was noticed very quickly, and they called the physician and instantly delivered a drug to help correct the error. But it's the kind of story that is unfortunately common in healthcare, but especially in any complex system. And in some ways, Amy, it points out to me the importance of allowing people to speak up about problems that they're seeing and also to take a systemic view

29:12of problems rather than looking for, you know, the smoking gun approach to problems. That's absolutely right. And our tendency is to look for that smoking gun. Like we, our brains are used to looking for the single cause, the small part that broke rather than to back up and see how the parts are relating to each other and coming together in a way that created the failure or created the flaw. And it's a discipline to sort of realize

29:42that there's multiple factors and in order to prevent complex failures, speaking up is essential. People need to know that their voices are welcome because all you have to do is catching correct one of the many factors and you've prevented the failure.

30:10It is possible to implement procedures that ensure that errors are caught and corrected early. Tell me how this is done by the car maker, Toyota. Toyota is one of the best examples of doing this well and probably the best practice that is maybe even the most famous of the Toyota production system is something called the Andon cord and that is a literal cord that any team member is encouraged to pull whenever they see

30:41something wrong or even more importantly, whenever they see something that might be wrong. Now, many people think once you pull the cord, the line instantly stops. It doesn't. When you pull the cord, what happens is that a team leader comes quickly over and says, what do you see? And you explain. And the two of you together diagnose what's happening and most of the time it turns out that you can either fix it or recognize that there wasn't a real problem and the line keeps going.

31:12But one out of 12 times there is a real problem there, the line will stop and it won't start again until that problem is fixed. So that prevents the complex failure of some small problem moving on downstream and we're pouring good money after bad at that point. Because when you stop the assembly line, when you're stopping production, it is costly. But I guess in the long run, the benefit is that the production becomes a higher quality production and over time you're starting to stamp out more and more mistakes.

31:42Exactly. So yes, it's absolutely costly in the short term. If the line stops for a minute, that is literally the loss of one car sale. So you are allowing a frontline associate to cost the company several thousand dollars anytime they wish. And of course, as your question suggested, they understand that this is money well spent. This is not a cost, this is an investment. Because every time

32:12we can stamp out small problems along the way, we are less at risk for producing anything less than a perfect car with high quality that will serve that customer well for years and years.

32:28We've talked about intelligent failures and complex failures. In some ways, at the bottom of your taxonomy of failures are what you call basic failures. You encountered one of these failures when you went sailing a few years ago. Tell me what happened, Amy. Well, I had signed up for an alumni regatta at Harvard. And when I got there, I realized kind of to my horror that everybody else there was about two years out of school, not several decades as I was.

32:58But that was okay. I can do it. I'm a good sailor. You know, I'll do the best I can. So off I went and I was thrilled to not come in dead last in the first race. After the second race, there was a break where all the boats go back to the dock before the regatta continues. The dock was dead downwind, which means that if you're in a sailboat, the boom on the sailboat holding up the sail is as far out as it can possibly go. And if you're dead downwind

33:30and the wind shifts even just a little bit, the boom is at risk of flying over to the other side. So any experienced sailor, and I am an experienced sailor, knows that. So we all know that when you're heading dead downwind, you had better be vigilant because you are at risk. Now, in the Charles River where this regatta was, the wind is notoriously shifty. But the race was over. We're just heading into the shore, so I'm chatting with my crew,

34:00a little bit relaxed, and all of a sudden, the boom flies across the boat with a little wind shift, knocks me out. Next thing you know, I'm in the freezing cold Charles River. Oh, my God. You got knocked overboard. I got knocked overboard. It's early May. The water is probably 40 degrees. Fortunately, my crew, who's a fantastic sailor, quickly turns the boat around to come get me. I climb in the back, the stern of the boat, and then I just see it,

34:30all the blood everywhere in the hull that's coming out of my head. So it led to nine stitches in the side of my head, and it was a basic failure. It was a small moment of inattention, looking away, not paying attention, just being overly casual when, in a sense, I'm operating dangerous machinery. And in some ways, I think what I hear you say is that the situations in which we tell ourselves, you know, I can do this in my sleep, these are the situations in which basic failures can often happen.

35:01That's exactly right. Anytime you hear yourself or someone else saying, oh, I can do this in my sleep, watch out. You can't.

35:11I understand that a simple tool like a checklist can be very effective when it comes to preventing basic errors. We've actually talked about the power of checklists on an earlier episode of Hidden Brain. You've pointed out, Amy, that sort of just having a checklist is not enough. You actually have to do it mindfully. And you described the story of an airline crash that took place some years ago. Tell me the story of what happened. So, it was a freezing cold January morning in Washington, D.C. and Air Florida Flight 90

35:42was headed for Fort Lauderdale, back home, I guess. And unfortunately, they crashed into the Potomac River. Why did that happen? Well, it turns out, yes, they used the checklist and most of our listeners are probably aware that the checklist includes the item anti-ice. When the first officer said, as part of the checklist, anti-ice, the captain said, habitually, off. It's Air Florida. I think most of their flights

36:13would not be using the anti-ice machinery. The first officer went on to the next item on the checklist. That was a tiny but crucial and catastrophic mistake that led to this terrible failure. Because, in fact, you did want to de-ice the wings before taking off on that very cold day in Washington, D.C. That's exactly right. You wanted to de-ice the wings and take the time that is required to do that to have a safe takeoff.

36:50Amy's work on how to generate fewer basic and complex failures makes sense. We all want to see fewer errors in hospitals and in space shuttle missions. When we come back, in a world where failure is generally stigmatized, how to get people and organizations to do the hard thing, systematically identify places where they can fail intelligently?

37:13You're listening to Hidden Brain. I'm Shankar Vedanta. Support for Hidden Brain comes from Southern New Hampshire University. The brain isn't fixed. It's flexible. That means