Right Kind of Wrong Summary and key ideas

Right Kind of Wrong, by Amy Edmondson.

Learning from failure sounds straightforward, but people do not naturally turn toward it. When something goes wrong, they may delay examining it, soften the account, deny it, or blame circumstances and other people. Shame and embarrassment make disclosure costly: a mistake can seem like evidence of incompetence, and criticism can feel like a threat to one’s standing. Avoidance may protect someone briefly, but it can also block learning and make people less willing to take on challenges where they might develop.

Part of the difficulty is that failure covers different things. A failure is an outcome that falls short of a desired result. An error is an unintended departure from a specified standard. A violation is a deliberate departure from a rule. These distinctions matter because not every disappointing outcome signals the same problem. Nor does the size of a failure tell us how much it can teach. A costly failure may yield information that could not otherwise be gained; a small failure is not automatically useful. The question is not simply how bad the result was, but what kind of failure occurred and what can be learned from it.

Edmondson makes the stakes tangible with early attempts to repair children’s hearts using experimental procedures. Some patients died. The possibility of medical progress does not erase those losses, and the tragedy should not be treated as acceptable just because it might produce knowledge. The example shows why failure’s consequences deserve serious attention, while also raising the question of whether an outcome can teach something important.

The book begins to separate failures into three broad kinds. Intelligent failures arise in uncertain new territory, where people are trying to learn what works. Basic failures are mistakes in more familiar work, where preventable errors are possible. Complex failures involve multiple causes that can interact. These categories are not a complete test for judging any one event. They establish that “failure” is not one uniform experience, and that learning from an experiment differs from preventing a mistake.

The central puzzle began in Edmondson’s doctoral research. Studying teams in two hospitals, she expected better teamwork to mean fewer medication errors. Instead, the initial results showed a statistically significant association in the opposite direction: teams rated as better had more detected errors. Edmondson felt ashamed and feared telling her adviser. The result looked like a failure of her own research, not yet like a clue about how teams work.

She later considered another explanation. Perhaps the stronger teams were not making more mistakes; perhaps their communication helped them notice, admit, correct, and report mistakes more readily. If people fear being blamed, they may hide errors. A team can then appear to have fewer errors because fewer are reported, not because fewer occurred. In that case, the recorded error rate reflects both what happens and what people feel able to say about it.

That interpretation remained uncertain and drew skepticism because it challenged the aim of measuring actual error rates. Further evidence made it plausible. A research assistant who did not know the teams’ error rates, ratings, or Edmondson’s hypothesis observed and interviewed hospital units. He independently described some as open about mistakes and others as punitive; his assessment of openness closely tracked the detected error rates. A survey question about whether mistakes would be held against someone also correlated with higher detected rates. These clues did not establish the true number of errors in each unit. They suggested that less open teams might have had errors that remained unseen.

A NASA simulator study offered a related example of teamwork changing the number of errors a group could catch. Researchers tested twenty two-person cockpit crews. The fatigued crews had worked together over multiple shifts, while the rested crews were unfamiliar with one another. Fatigued individuals made more errors than rested individuals, but the fatigued crews who had worked together made fewer errors as teams than the rested, unfamiliar crews. Crew members apparently caught and corrected one another’s mistakes. The finding helped fuel crew resource management: individual fallibility did not disappear, but the team could compensate when people worked together effectively.

Edmondson uses the term psychological safety for a team climate in which people can speak up, ask for help, and discuss mistakes without expecting interpersonal punishment. It is a quality of the group, shaped by how its members respond to one another, not a fixed personality trait. Such a climate makes it easier to disclose a problem, hear an uncomfortable concern, and catch or correct an error. It can also support thoughtful experimentation where answers are not already known.

Psychological safety does not mean lowering standards or excusing poor work. Candor and accountability can coexist: people can be expected to meet standards and still be able to report a concern or admit an error. In fact, high expectations without room to speak up can make problems harder to surface. A psychologically safe team is not necessarily a comfortable one; candid discussion can be difficult. The point is that people can discuss both preventable harm and useful learning openly. Failing well begins by distinguishing between them—and making honest conversation possible.