Skip to content
Published on

Learning from Mistakes — Resilience and Antifragility

🇰🇷KO🇺🇸EN🇯🇵JA
Split View필사 모드
Share
Authors

Opening — The gap between a mistake and a verdict

Most of us learned early that mistakes are dangerous.

A wrong answer got marked in red.

A spilled glass drew a sharp voice.

By the time we are adults, the reflex is deep: hide the error, explain it away, move on quickly.

The trouble is that this reflex is almost perfectly designed to prevent learning.

You cannot learn from something you refuse to look at.

This essay is about the other path.

It is about treating a mistake as information rather than as a sentence handed down about who you are.

Along the way we will meet a few ideas that have aged well.

We will look at Nassim Taleb's notion of antifragility, at the blameless post-mortem that grew out of engineering culture, at Amy Edmondson's distinction between intelligent and preventable failure, and at a couple of small practices anyone can use.

We will also stay honest.

Not all failure is good failure.

The goal is not to celebrate every crash but to make our failures cheaper, more survivable, and more instructive.

1. Why mistakes feel like a threat

Start with the feeling, because the feeling is where most of the damage happens.

A mistake rarely arrives as neutral information.

It arrives as a small alarm in the body.

Your face gets warm, your attention narrows, and some old story starts up in the background.

The story usually says something about you rather than about the situation.

Not "that approach did not work" but "I am careless," or "I do not belong here."

Psychologists have a name for the fear that clusters around this: we are afraid of looking incompetent, and that fear shapes behavior long before we notice it.

The avoidance move

When a mistake feels like a threat to identity, the natural response is to make it disappear.

We minimize it.

We reframe it as someone else's fault.

We quietly stop doing the thing that exposed us.

Each of these moves feels like relief in the moment.

And each one deletes the very data we would need to improve.

This is the quiet cost of a blame culture.

It is not mainly that people get punished.

It is that people get good at hiding, and a system that cannot see its own errors cannot correct them.

Separating the sting from the signal

There is a small but powerful skill here.

It is learning to feel the sting of a mistake and still turn toward the signal inside it.

The sting is real and you do not have to pretend otherwise.

But the sting is about your comfort, and the signal is about the work.

Those are two different things, and confusing them is where a lot of learning goes to die.

2. Fragile, resilient, antifragile

Nassim Taleb, in his book "Antifragile," draws a distinction that is useful far beyond finance, where he first sharpened it.

He notices that our language has a word for things that break under stress.

We call them fragile.

And we have a rough word for things that resist stress and stay the same.

We call them robust, or resilient.

But he points out that we are missing a word for the third category.

The third category

Some things do not merely survive disorder.

They improve because of it.

A muscle placed under load, within reason, grows back stronger.

An immune system exposed to a small challenge learns to defend against a larger one.

A craftsman who hits a hard problem builds a skill that outlasts the problem.

Taleb calls this quality antifragility.

The fragile thing wants calm and predictability, because any shock is a threat.

The resilient thing is indifferent to shocks; it takes the hit and continues unchanged.

The antifragile thing has a stranger relationship with disorder: a certain amount of stress is not just tolerable but necessary for it to develop.

Why the frame matters here

The point for our purposes is not to romanticize stress.

Too much load tears the muscle.

Too large a shock kills the organism.

The point is that a system, or a person, or a team can be arranged so that ordinary, survivable failures feed growth instead of just causing damage.

That arrangement does not happen by accident.

It is something you design for, and most of this essay is about how.

3. A failure is data, not a verdict

Here is the shift that makes everything else possible.

When something goes wrong, you can read it as a verdict or you can read it as data.

The verdict reading says: this outcome reveals a fixed truth about my worth or my ability.

The data reading says: this outcome is a measurement of what happened when a particular approach met a particular reality.

The second reading is not softer.

It is more accurate.

What "data" actually implies

Treating a failure as data has real consequences for what you do next.

Data is specific, so you get curious about the details instead of drowning in a general feeling of badness.

Data is about a moment in time, not a permanent trait, so it invites another attempt rather than a retreat.

Data can be wrong or incomplete, so you look for more of it instead of over-fitting to a single bad night.

None of this requires pretending the outcome was fine.

It requires separating the measurement from the meaning you rush to assign it.

A quiet reframe

One practical habit is to change the sentence you say to yourself.

Replace "I failed" with "here is what happened, and here is what it might be telling me."

It sounds almost too simple.

But the first sentence closes the case, and the second one opens an investigation.

Only one of them leads anywhere useful.

4. The blameless post-mortem

Software engineering, of all places, produced one of the most humane tools for learning from mistakes.

When a large system breaks, teams hold a review afterward, often called a post-mortem or an incident review.

The word "blameless" in front of it is doing quiet, heavy work.

Where it came from

The modern version of this practice was articulated clearly by John Allspaw and his colleagues during his time in engineering leadership, drawing on ideas from safety science.

One deep influence is the work of Sidney Dekker, whose writing on human error argues that "human error" is usually not an explanation but a starting point.

When someone makes a mistake, Dekker suggests, the useful question is not "who is to blame" but "why did this action make sense to a competent person at the time."

That reframing changes everything downstream.

Separating the person from the error

A blameless review holds a specific stance.

It assumes that the people involved were doing their reasonable best given what they knew and the pressures they faced.

So the investigation goes after the conditions, not the character.

What information was missing.

What signal was easy to misread.

What part of the system made the wrong move easy and the right move hard.

The person is separated from the error so that the error can be examined honestly.

Why this produces better systems

The payoff is counterintuitive but consistent.

When people are not afraid of being scapegoated, they tell the truth about what actually happened.

And truth about what actually happened is the raw material for every real fix.

A team that punishes the messenger loses the message.

A team that protects the messenger gets to see the fault line clearly, and can reinforce it before the next tremor.

5. Intelligent failures and preventable ones

It would be a mistake to read all of this as "failure is wonderful, fail freely."

Amy Edmondson, a professor at Harvard Business School who has spent decades studying this, is careful to separate the kinds.

Not all failures are equal, and treating them as if they were is its own error.

A rough taxonomy

Some failures are preventable.

They happen in well-understood territory, where a known process exists and was simply not followed.

A checklist skipped, a step rushed, a rule ignored.

These are not to be celebrated; they are to be reduced.

Other failures are complex.

They come from the interaction of many small factors in a system too tangled for anyone to have fully predicted.

And a third kind, the most valuable, Edmondson calls intelligent failures.

The intelligent failure

An intelligent failure is what happens at the frontier of knowledge.

You are trying something genuinely new, where the answer could not have been looked up.

The attempt is thoughtful, it is as small as it can be while still being informative, and it is designed so that you learn something specific whether it works or not.

When such an experiment fails, it has still done its job.

It has bought you knowledge you could not have obtained any other way.

The discipline is to spend freely on intelligent failures and to stubbornly reduce the preventable ones.

Psychological safety

Edmondson is also the researcher most associated with the term psychological safety.

It names a simple condition: the shared belief that you can speak up, admit a mistake, or ask a naive question without being humiliated.

Her research links this condition to teams that learn faster and, in some settings, report more errors precisely because they are no longer hiding them.

Safety here does not mean comfort or lowered standards.

It means the fear of looking bad has stopped competing with the work of getting things right.

6. Structured reflection: what, so what, now what

Experience by itself does not teach.

Reflected experience does.

A person can make the same mistake for ten years and call it a decade of experience, when it was really the same year lived ten times.

The difference is a habit of looking back on purpose.

A simple three-part frame

One durable structure, used in nursing education, the military, and elsewhere, moves through three questions.

First, what happened.

You describe the events plainly, without interpretation, as if reporting to someone who was not there.

Just the facts, in order.

Second, so what.

Now you interpret.

Why did it matter, what does it suggest, what pattern connects it to other things you have seen.

Third, now what.

This is the part most people skip, and it is the part that changes the future.

What will you do differently, concretely, the next time you are near this situation.

Why the order helps

The order is not decorative.

If you jump straight to "now what," you tend to fix the wrong thing, because you never really understood what happened.

If you stop at "so what," you have an insight that evaporates by morning, because it was never turned into an action.

Walking all three, in sequence, is what converts a single event into a lasting adjustment.

Writing it down

There is something about writing that reflection alone in your head does not provide.

Written reflection forces vague feelings into specific words.

It also leaves a record, so that in six months you can see whether you actually changed or merely intended to.

The medium can be humble.

A few honest lines in a notebook will do more than an elaborate system you never open.

7. The pre-mortem: failing on paper first

So far we have talked about learning from mistakes after they happen.

There is a technique for learning from them before they happen, and it costs almost nothing.

Gary Klein, a research psychologist who studies how experts make decisions, described it in the "Harvard Business Review," and called it the pre-mortem.

How it works

A post-mortem asks, after a project has failed, why it failed.

A pre-mortem asks the same question in advance, as an act of imagination.

Before you commit, you gather the people involved and pose a specific scenario.

Imagine it is a year from now and this plan has failed completely.

Now, each person writes down the story of how that happened.

Why the framing is clever

The move is subtle and it works on human psychology.

Normally, once a team is committed to a plan, doubts feel disloyal, and people stay quiet to avoid seeming negative.

The pre-mortem gives everyone permission to voice concern, because the exercise assumes failure as its premise.

You are not the pessimist in the room.

You are just playing the game everyone agreed to play.

In practice this surfaces risks that would otherwise have stayed hidden until they became expensive.

The failure happens on paper, where it is cheap, instead of in reality, where it is not.

8. A fair word about growth mindset

No essay on learning from mistakes can avoid the phrase growth mindset, and it deserves a fair, brief hearing.

The idea, associated with the psychologist Carol Dweck, is that people hold beliefs about whether their abilities are fixed or can be developed.

Those who lean toward the developable view tend to treat difficulty as part of growth rather than as proof of a ceiling.

It is a genuinely useful frame, and it rhymes with much of what we have covered.

The nuances

It is worth being precise, because the idea has sometimes been oversold.

A growth mindset is not a belief that effort alone conquers everything, and it is not a reason to praise effort that is going nowhere.

The research itself has been refined over time, with later studies suggesting the effects are real but often modest, and most reliable for specific groups such as struggling students given the right support.

Mindset also is not a substitute for strategy.

Believing you can improve helps most when it is paired with a concrete method for improving.

A separate post takes up growth mindset on its own terms.

Here it is enough to say that it belongs in the toolkit, honestly sized, alongside the sturdier machinery of blameless review and structured reflection.

9. Not all failure is good

It is time to state the caution plainly, because the ideas in this essay are easy to abuse.

The celebration of failure has become something of a cultural slogan, and slogans lose their edges.

"Fail fast" was never advice to be reckless.

The goal is cheap, survivable, instructive failure

The failures worth having share three properties.

They are cheap, meaning the cost of being wrong is small relative to what you might learn.

They are survivable, meaning they do not take you out of the game or cause irreversible harm.

And they are instructive, meaning they are set up so that you actually learn something specific from the outcome.

A failure that is expensive, catastrophic, and confusing is not a growth opportunity.

It is just a loss.

Some things you do not fail your way through

There are domains where the first principle is not to learn from mistakes but to prevent them.

You do not want a surgeon, a pilot, or a bridge engineer treating a fatal error as a nice piece of feedback.

In these high-stakes settings, the learning is supposed to happen in simulation, in training, in near-miss reports, and in the careful study of other people's disasters, precisely so that the real thing never has to teach you.

Antifragility is about arranging your exposures so that the small, recoverable failures feed you and the large, ruinous ones are kept off the table.

Reckless is not brave

There is a difference between courage and carelessness that is easy to blur.

Courage takes an intelligent risk, having thought about the downside and made it survivable.

Carelessness takes a risk without that thought and calls it boldness afterward if it happens to work.

Learning from mistakes is a serious discipline, not a license.

It asks for more attention, not less.

A picture of the three responses

Here is the whole distinction in one small diagram.

A shock arrives  ─────────────────────────────►

  FRAGILE        RESILIENT        ANTIFRAGILE
   [ vase ]       [ stone ]        [ muscle ]

    breaks         unchanged        grows back
                                    stronger

   avoids all      absorbs the      needs some
   disorder;       hit; returns     disorder to
   any shock       to baseline      develop; too
   is a threat     no better,       much still
                   no worse         destroys it

  ── goal: turn identity-threatening errors into ──
  ──  cheap, survivable, instructive stressors  ──

The aim is never to seek the largest possible shock.

It is to sit, deliberately, in the middle column's steadiness and the right column's growth, while designing the fragile failures on the left to be small enough to survive.

Closing — The compound interest of honest mistakes

If there is a single thread through all of this, it is that mistakes are not the opposite of learning.

Handled well, they are its main ingredient.

The fragile response hides them, and learns nothing.

The resilient response endures them, and stays the same.

The antifragile response metabolizes them, and comes out with something it did not have before.

None of this is about enjoying failure.

It is about refusing to waste it.

A mistake you look at squarely, name honestly, and reflect on with structure will pay you back, often for years.

A mistake you flinch away from will cost you twice: once when it happens, and again when it happens a second time because you never learned the first lesson.

So the practice is quiet and unglamorous.

Feel the sting, then turn toward the signal.

Separate the person from the error.

Spend on intelligent failures and reduce the preventable ones.

Reflect on purpose, and, when you can, fail on paper before you fail in the world.

Do that steadily, and your mistakes stop being a tax on your confidence and start becoming a form of compound interest.

Questions to Ponder

  1. Think of a recent mistake that stung. Which part of the sting was about the work, and which part was about how you looked? What changes if you separate them?

  2. In a team you belong to, is it safer to hide a small error or to surface it? What single change would tilt that balance toward honesty?

  3. Where in your life are you paying for expensive, unrepeated failures when a cheap experiment or a pre-mortem could have taught you the same thing for less?

  4. Is there a domain in your work where the right goal is to prevent mistakes rather than learn from them? How do you currently make that learning happen without the real disaster?

References

  • Nassim Nicholas Taleb, "Antifragile: Things That Gain from Disorder" (Random House, 2012). The founding statement of the fragile / robust / antifragile distinction.

  • Amy C. Edmondson, "Strategies for Learning from Failure," Harvard Business Review (April 2011). https://hbr.org/2011/04/strategies-for-learning-from-failure

  • Amy C. Edmondson, "The Fearless Organization: Creating Psychological Safety in the Workplace for Learning, Innovation, and Growth" (Wiley, 2018).

  • Sidney Dekker, "The Field Guide to Understanding 'Human Error'" (3rd ed., CRC Press, 2014). On separating the person from the error in incident analysis.

  • Gary Klein, "Performing a Project Premortem," Harvard Business Review (September 2007). https://hbr.org/2007/09/performing-a-project-premortem

  • John Allspaw, "Blameless PostMortems and a Just Culture," Code as Craft (Etsy Engineering Blog, 2012). https://www.etsy.com/codeascraft/blameless-postmortems/

Discuss on Twitter • View on GitHub