Skip to content
Published on

The Comfort of Admitting Mistakes — Psychological Safety Drives Performance

🇰🇷KO🇺🇸EN🇯🇵JA
Split View필사 모드
Authors

Introduction: The Words "It Was Me"

An incident review room. The payment system just went down for 30 minutes, and hundreds of transactions failed. Everyone is hunting for the cause. Then one engineer raises a hand and says, "It was the config change I deployed yesterday. I skipped the rollback step."

Saying those words takes courage. And whether that courage can emerge depends almost entirely on the atmosphere of that room. If blame and interrogation pour out after those words, no one will raise a hand at the next incident. If instead the response is, "Thank you for being honest. Let's look at how we can automate this step," that team gets stronger every time.

This essay is not a moral plea to "admit your mistakes." It first tries to understand why admitting mistakes is so hard, and then gets concrete about how to build the condition that makes it possible — psychological safety. And the central claim is this: psychological safety is not a matter of "being nice." It is a matter of performance.

Why Admitting Mistakes Is Hard

First, a fact we must concede. Hiding mistakes is not a defect but an instinct. The urge to avoid blame is etched deep by evolution — a trace of times when being cast out of the group was fatal to survival.

Inside organizations, we constantly engage in impression management. Amy Edmondson catalogued four images people want to avoid at work.

Image to avoid           So they stop doing this
--------------------    -----------------------------------------
Looking ignorant        Asking questions
Looking incompetent     Admitting mistakes, asking for help
Looking intrusive       Voicing opinions or concerns
Looking negative        Offering criticism or dissent

What this table reveals is a frightening fact: silence is always the safer choice. Not asking, hiding mistakes, swallowing concerns — these are always rational in the short term. The problem is that when these individually rational silences add up, the whole organization loses its capacity to learn.

The Cost of Silence

What happens when silence accumulates?

  • Small problems stay hidden until they become big accidents (buried bugs, ignored warnings).
  • The same mistakes repeat (because no one learns from them).
  • Good ideas die unspoken (out of fear).
  • The most honest person suffers most (raising a hand alone only to be blamed).

So admitting mistakes is not a problem of individual courage but a problem of the system. Organizations that rely on courage alone fail. Organizations that make honesty safe succeed.

Edmondson's Psychological Safety: Its Elements

Amy Edmondson of Harvard Business School has studied "psychological safety" since the 1990s. Her definition runs like this:

"Psychological safety is a shared belief that the team is safe for interpersonal risk-taking."

That is, the belief that you will not be punished or humiliated for asking a question, admitting a mistake, or voicing a concern. Strikingly, Google's large-scale team study "Project Aristotle" concluded that the single most important factor distinguishing high-performing teams was psychological safety. A safe atmosphere mattered more than the sum of smart individuals.

The Elements of Psychological Safety

Synthesizing Edmondson's work and follow-on research, psychologically safe teams have these four things.

  1. Inclusion and belonging: The feeling that I belong on this team and am accepted as I am. The most basic foundation.
  2. Safety to learn: The belief that it's okay to ask, to say "I don't know," to experiment, and to make mistakes. The zone where growth happens.
  3. Safety to contribute: The feeling that my opinions and ideas are welcomed and taken seriously.
  4. Safety to challenge: The belief that it's safe to question the status quo and propose better ways. The highest level.

Correcting a Misconception: Psychological Safety ≠ a Soft Atmosphere

Here we must address the most common misconception. Psychological safety is not "everyone be friendly, no criticism, low standards." Quite the opposite. Edmondson offers a matrix with psychological safety and performance standards as two axes.

                High performance standards
                      ^
        Anxiety       |    Learning
        Zone          |    Zone
   <----------------+----------------> High psychological safety
        Apathy        |    Comfort
        Zone          |    Zone
                      |
                Low performance standards
  • Anxiety Zone: High standards, low safety. People work in fear. They hide mistakes and burn out.
  • Comfort Zone: High safety, low standards. Pleasant but no growth. The "soft atmosphere" actually lives here.
  • Apathy Zone: Both low. No one cares.
  • Learning Zone: High safety and high standards. This is where the best performance and growth happen.

The point is clear. Psychological safety is not an alternative to high standards but the condition that makes high standards possible. Because it is safe, people can speak hard truths, take risks, and give each other honest feedback.

Blameless Postmortems

The best example of psychological safety implemented as a system is the blameless postmortem. This practice, cemented by Google's SRE culture and companies like Etsy, asks not "who did wrong?" when an incident occurs but "why did the system allow this mistake?"

Core Premise: Humans Make Mistakes

The philosophical starting point of the blameless postmortem is simple: well-intentioned, competent people make mistakes. Therefore, blaming and firing one person fails to fix the root problem. The next person will fall into the same trap.

Instead, the question changes.

  • "Why did that person make the mistake?" (X)
  • "Why did the system make the mistake easy and fail to prevent it?" (O)

A Blaming Review vs. a Blameless Review

Let me show the difference with dialogue examples.

Blaming review:

"Who approved this deploy? Why didn't you check? Be more careful next time."

The result of such a review: people get defensive, the real cause stays hidden, and at the next incident no one speaks honestly.

Blameless review:

"This deploy went to production without validation. At which step was validation skipped? What would it take to make validation mandatory or automated?"

The result of such a review: the real cause (the gap in the validation process) surfaces, the system improves, and people keep sharing honestly next time.

Components of a Good Postmortem

Section           Content
--------------    -----------------------------------------
Timeline          What happened and when (facts only, no judgment)
Impact            Who/what was affected and how much
Root cause        Dig to systemic causes with 5 Whys, etc.
What went well     What was effective in the response
Action items      Concrete, owned items to prevent recurrence

The "what went well" section matters especially. It signals through atmosphere that an incident review is a place for learning, not self-flagellation.

Leaders Showing Vulnerability First

Psychological safety is built from the top. No matter how much you say "it's okay to be honest," if the leader never admits their own mistakes, no one believes it. Action beats words.

The Power of a Leader's Vulnerability

The moment a leader says "I made the wrong call here," two things happen.

  1. A signal of permission: The message "it's safe to admit mistakes here" spreads throughout.
  2. The paradox of trust: Counter to intuition, a leader who shows vulnerability looks not weaker but more trusted. Brené Brown's research illustrates this "power of vulnerability" well.

Concrete Behaviors a Leader Can Practice

Behavior                 Example
--------------------    -----------------------------------------
Share your own mistake   "My misjudgment last quarter slipped the schedule"
Admit not knowing        "I don't know that. Let's find out together"
Welcome questions        "Good question. Anything else you'd like to ask?"
Protect the failer       Instead of blame, "It's okay, let's fix it together"
Ask for feedback         "What could I do better?"

That last one — asking for feedback — is especially powerful. When a leader sincerely seeks feedback about themselves, it signals to the whole team that feedback is safe.

The Power of "I Don't Know" and "I Was Wrong"

One thing high-safety teams share is that people say two sentences with ease: "I don't know" and "I was wrong."

Why "I Don't Know" Is the Expression of a Strong Person

Many treat "I don't know" as a confession of incompetence. In reality it's the opposite.

  • The person who admits not knowing is ready to learn.
  • The person who admits not knowing doesn't wreck the work with guesses.
  • The senior who admits not knowing gives juniors permission that it's okay not to know.

The most dangerous person is the one who pretends to know while not knowing. That bluff breeds bad decisions and hidden risks.

The Trust Created by "I Was Wrong"

The person who quickly admits being wrong is, paradoxically, more trusted. Because:

  • When they say "I'm right," you can believe it (since they admit it when wrong).
  • It shows they put truth above ego.
  • It steers arguments away from "who wins" toward "what's correct."

Dialogue example:

Colleague A: "I think this approach has a concurrency problem." Colleague B: "Hmm, I missed that part. You're right. How should we fix it?"

In this short exchange, B looks not weak but like someone who chooses a better outcome over pride.

Measuring Psychological Safety

There's a saying: "what you can't measure, you can't manage." Psychological safety, too, is not a vague feeling but something measurable. Edmondson proposed survey items to gauge a team's psychological safety. You can diagnose it by how much team members agree with these statements.

Diagnostic items (respond by degree of agreement)
-----------------------------------------
1. If you make a mistake on this team, it's often held against you (reverse-scored)
2. On this team, you can bring up difficult problems
3. People on this team don't reject others for being different
4. It's safe to take a risk on this team
5. It's hard to ask others on this team for help (reverse-scored)
6. No one would deliberately undermine my efforts
7. My unique skills and talents are valued and used

Items 1 and 5 are reverse-scored — lower scores mean more safety. Running this diagnostic regularly (e.g., quarterly and anonymously) lets you track changes in the atmosphere.

Reading Behavioral Signals

You can read psychological safety from everyday behavior even without a survey.

  • Who speaks in meetings? If only a few speak, it's a warning sign.
  • How often do you hear "I don't know" and "I was wrong"? If never, people are wearing masks.
  • Does bad news rise quickly, or stay hidden until it explodes? Fast reporting is evidence of safety.
  • How many questions does a new hire ask in their first week? A silent new hire has already read the room.

Building Safety Gradually: The First Step

Psychological safety doesn't appear overnight. It is the accumulation of small actions. Whether you're a leader or a team member, there's a first step you can take today.

Starting in a Single Meeting

Timing            Concrete action
--------------    -----------------------------------------
Meeting start     "Today, dissent is welcome too"
If someone's quiet "What do you think, Jordan?" — a direct invitation
On a good question "That's a really good question"
When a mistake     React first with "thanks for sharing this"
  surfaces
Meeting end       "Is there anything we might have missed today?"

The invitation of a quiet person — "What do you think, Jordan?" — is especially powerful. People who don't speak often aren't lacking opinions; they feel it isn't safe to speak. An invitation opens that door.

The Power of Keeping Small Promises

The foundation of psychological safety is trust, and trust accumulates from keeping small promises. If you said "I'll review it by tomorrow," reviewing it by tomorrow. This small consistency builds the belief that "this person does what they say," and that belief expands into "it's safe to be honest in front of this person."

Systems That Learn from Failure

Psychological safety isn't sustained by individual attitudes alone. You need systems that learn from failure systematically — building safety into structure rather than leaving it to atmosphere.

Mechanisms That Turn Failure into Learning

  • Blameless postmortems (covered above): Make them routine for every meaningful incident.
  • Error budget: Explicitly state a certain level of failure as an allowed budget. The pressure that failure must be zero actually suppresses experimentation and drives failures into hiding.
  • Failure-sharing gatherings: A light venue to share "what I broke this week." Normalizing failure.
  • Action tracking: Track whether postmortem actions actually get done. Otherwise reviews become ritual.

Distinguishing Types of Failure

For a balanced view, an important distinction. Edmondson stresses that not all failures should be praised equally. Failures come in kinds.

Failure type        Example                   Response
--------------    --------------------    -------------------
Blameworthy         Ignoring procedure,       Coaching, clear standards
                    carelessness
Complexity          Unforeseen interaction    System improvement
Intelligent         Experiment of a new try   Celebrate and learn

Note that "psychological safety" does not mean "any failure is fine." Deliberate negligence or repeated carelessness must still be addressed. The difference is that the way you address it is coaching and system improvement, not blame.

Anti-patterns: The Blame Game and Its Cousins

Finally, let's examine the anti-patterns that destroy psychological safety. Just noticing them prevents half the damage.

The Blame Game

The most destructive anti-pattern. A culture that, when an incident occurs, first hunts for "whose fault." The result is obvious: people hide risks, conceal information, and spend energy on self-protection.

Other Common Anti-patterns

Anti-pattern        Symptom                   Antidote
--------------    --------------------    -------------------
Blame game          "Whose fault is it?"      "Why did the system allow it?"
Hero worship        One person carries all     Distribute knowledge, pairing
Punitive reviews    Reviews are inquests       Blameless postmortems
Perfectionism       Failure never tolerated    Error budget, celebrate intelligent failure
Surface agreement   Nods in the meeting,       Actively invite dissent
                    grumbling after
False safety        Repeats "it's fine," no    Prove it through action, not words
                    follow-through

Beware "false safety" especially. If a leader merely says "be honest here" but then gets uncomfortable or penalizes whoever becomes honest, that gap destroys trust even more. Psychological safety is built not by declaration but only by the accumulation of repeated action.

Case Walkthrough: How One Incident Made a Team Stronger

To see how the principles above actually work, let's follow a hypothetical (but very common) incident.

Situation: A junior engineer ran a migration script incorrectly, leaving some data in the production DB with wrong values for 30 minutes.

Bad path (blame culture):

Manager: "Why did you run it directly in production? Who told you to do that?" Junior: (freezing) "I'm sorry... I didn't check."

Result: The junior shrinks, and the real cause (a tool design that makes production and staging hard to tell apart) stays buried. The next person who sees a similar risk also stays silent.

Good path (blameless culture):

Manager: "First, thank you for sharing it quickly. Because of that we caught it in 30 minutes. Let's look together. How did the production script come to look like staging?" Junior: "In the terminal the two environments look almost identical, so I didn't realize it was production." Manager: "Good find. So what if we color the production terminal red, or add a confirmation step to production migrations?"

Result: The blameless postmortem yields two actions — (1) a visual warning for the production environment, (2) a confirmation gate for migrations. The system improves, and far from shrinking, the junior becomes someone who reports quickly next time too. And the whole team, hearing this story, learns once more that "it's safe to speak honestly here."

The same incident, completely different outcomes. What made the difference was not technology but the first reaction.

Balance: Safety and Accountability Don't Conflict

The most common objection when discussing psychological safety is: "Won't that just make a soft organization where no one is accountable?" We must resolve this misconception precisely.

Psychological safety and accountability are not opposites but two independent axes.

                High accountability
                      ^
        Anxiety       |    High-performance
        Zone          |    learning zone
   <----------------+----------------> High psychological safety
        Apathy        |    Comfort
        Zone          |    Zone
                      |
                Low accountability
  • High accountability, low safety: the Anxiety Zone. People fear and hide mistakes.
  • High safety, low accountability: the Comfort Zone. Pleasant, but no results.
  • Both high: the high-performance learning zone. People honestly admit mistakes and take responsibility for fixing them.

The core insight is this. A blameless culture is not "a culture without accountability" but "a culture where the direction of accountability points from the person to the system, and toward future improvement."

The Right Way to Hold Accountable

  • Accountability that punishes the person (X): "Who did this?" → hiding.
  • Accountability toward improvement (O): "How do we prevent recurrence? Who owns that action, and by when?"

The latter also clearly assigns responsibility to someone. But that responsibility points toward future improvement, not past blame. This is real accountability and, at the same time, safe accountability.

Conclusion: Safety First, Performance on Top

Back to that incident review room. The engineer who raised a hand and said "it was me." Those words could emerge not because that person was especially brave, but because the room had made those words safe.

Psychological safety is not a matter of niceness. It is the foundation of performance. People must be able to say "I don't know" to learn, to say "I was wrong" to fix, to say "I'm concerned" to prevent accidents. And all of this leads to the best performance when it goes together with high standards.

Safety and standards are not an either/or. Safety is the floor that makes standards possible. When the floor is firm, only then can we build high.

Finally, this is not a job for leaders alone. Every time each of us says "it's okay, let's fix it together," the room where someone can raise a hand grows a little wider.

In One Sentence

If we compress this long discussion into a single sentence: when it is safe for people to speak the truth, the organization finally begins to learn. Truth includes bad news, the confession of not knowing, one's own mistakes, and differing opinions. When these truths flow without fear, small problems are caught before they grow, good ideas aren't buried, and the same mistakes don't repeat.

Psychological safety is the channel that makes that flow of truth possible. And that channel is built not by a grand program but by the first reaction you'll show today to a colleague's mistake.

Practical Checklist

As an individual

  • Have you recently said "I don't know" honestly?
  • Do you quickly admit it when you're wrong?
  • Do you respond to a colleague's mistake with "let's fix it together" instead of blame?

As a leader

  • Have you admitted your own mistake in front of the team?
  • Do you sincerely seek feedback about yourself?
  • When someone raises a concern, do you welcome it or get defensive?

As a team / system

  • Have you made blameless postmortems routine?
  • Are postmortem actions actually tracked and executed?
  • Do you distinguish types of failure (intelligent failure vs. carelessness) in how you handle them?
  • Do you pursue high standards and high safety at the same time? (Learning Zone)
  • Have you avoided the trap of "false safety"? (Words matching action)

References

Discuss on Twitter • View on GitHub