Monday morning, March 9th. A startup founder is giving a keynote about how they built a billion-dollar company. The story is excellent: early struggles, a pivotal insight, a near-death crisis, then breakthrough. The audience is riveted. Founders in the room are furiously taking notes, trying to absorb the lesson. Later, outside, one turns to another: "That's exactly what I need to do." She's wrong. She's not learning how to build a company. She's learning how to tell a story about building a company. These are not the same thing.

Stories are the native format of human cognition. They're also the primary mechanism by which we get reality systematically wrong. The problem isn't that you think in stories — you can't help this, you're wired for it. The problem is that you've confused the story with what actually happened, and then made decisions based on the story. The more compelling the narrative, the more damage it does.

The Narrative Fallacy

Nassim Taleb named this the "narrative fallacy" in The Black Swan: we continuously reconstruct the past to make it look more predictable, more ordered, more comprehensible than it was. We do this because stories are cognitively cheap. A good narrative compresses enormous amounts of information into a few memorable causal links.

The cost is distortion. Every good story requires: a protagonist whose choices matter, causal links between events, and a resolution that feels earned. Reality is under no obligation to provide any of these. Most important events are overdetermined — they happened for many reasons simultaneously, most of which no one was tracking at the time. Most "decisions that made the difference" would have been irrelevant if twenty other things had gone differently. Most endings aren't resolutions. They're just where we stopped looking.

When you retell events as a story, you can't help but:

  • Assign causality to correlation: A happened, then B happened — so A caused B
  • Elevate individual agency: The CEO's decision, the founder's insight, the general's move. Story requires characters who drive outcomes. It systematically underweights structure, luck, and the thousand background conditions that made the character's move possible at all
  • Mistake the ending for the verdict: The startup failed, so the strategy was wrong. The startup succeeded, so the strategy was right

That last one is what poker player and decision theorist Annie Duke calls "resulting" — judging the quality of a decision by its outcome rather than by the information and reasoning available at the time of the decision. It's one of the most reliable ways to learn exactly the wrong lesson from experience.

Why Good Storytellers Are Most Vulnerable

The people who are best at constructing compelling narratives are, for this reason, often the most confidently wrong.

A mediocre story-maker fumbles the causality and gets caught. A good story-maker weaves together real facts into a seamless, plausible, internally consistent account — that happens to omit all the disconfirming evidence, all the structural factors, all the luck. Their account is more persuasive than the mediocre one, harder to argue with, and more likely to lead people astray.

This is why post-mortems often fail. You gather the team, walk through what happened, and construct a shared narrative. Everyone agrees on it. The narrative feels right because it's coherent. But coherence and accuracy are different properties. The story you collectively built is one plausible account that selected for memorability and causal tidiness, not for completeness. You've explained the past in a way that will not help you predict the future.

Hindsight bias is the same mechanism. Once you know how things turned out, the story of how they had to turn out looks obvious. You can construct a perfectly coherent narrative from cause to effect. That narrative makes the outcome feel inevitable. "Of course that's what happened." No. The narrative created that feeling. The outcome was not inevitable. The story is lying to you.

The Fundamental Attribution Error Is a Storytelling Problem

The fundamental attribution error — one of social psychology's most replicated findings — is that people consistently overestimate how much behavior reflects character versus situation. If a colleague is rude in a meeting, you think: they're a difficult person. If you're rude in a meeting, you think: I'm under a lot of stress.

This is a storytelling error. Characters need fixed attributes for a narrative to cohere. "He's a difficult person" is narratively satisfying in a way that "a combination of his sleep deprivation, the meeting's ambiguous framing, and two previous interactions that put him on the defensive" is not. So we simplify to character. Then we get the person wrong, interact with them incorrectly, and explain our failures by appealing to their character some more.

The inverse happens with successful institutions. "Apple succeeded because Steve Jobs was a visionary." That's not wrong — but it's incomplete in a way that makes it useless as a predictive model. Remove the structural conditions, the timing, the team, the competitive landscape, the particular cultural moment, and the "vision" becomes noise. The story singles out the protagonist because stories require protagonists. It doesn't follow that protagonists determine outcomes.

The Reference Class You're Not Consulting

Here's the practical test. When you're about to make a decision, you likely reach instinctively for a story: an example of someone who did something similar, a narrative about how it played out, a causal model of why this time it'll work (or not work).

What you rarely reach for is the base rate.

"This company succeeded because the founder did X" is a story. The question is: of all the founders who did X, what percentage succeeded? Stories cannot answer this. The story you have access to has already been filtered by availability — you heard it because it was compelling, which usually means it was unusual. The failures who did X are invisible because failure stories don't get keynotes.

Daniel Kahneman calls this the inside view versus the outside view. The inside view uses everything you know about your specific situation to construct a narrative of how it will unfold. The outside view asks: what usually happens to people in situations like this? The outside view is almost always more accurate. We almost always use the inside view.

The outside view is boring. It doesn't give you a protagonist. It gives you a probability distribution. Human brains are not well-designed to be moved by probability distributions. This is not a design flaw you can reason your way around — it's structural. The workaround is to deliberately build the habit of asking for base rates before you let yourself get absorbed in the story.

What to Do Instead

None of this means you should stop thinking in stories. You can't, and you'd be worse off if you could — stories are efficient, motivating, and necessary for communication. The goal is to notice when you're reaching for a story in a context that requires something else.

In post-mortems, ask what the story omits. After constructing the shared narrative of what happened, explicitly ask: what factors outside this narrative contributed? What would a hostile critic say the causality actually was? What's the version of events where the protagonist's choices were irrelevant? This isn't nihilism — it's completeness.

When evaluating people, ask about the situation first. Before concluding that someone's behavior reflects who they are, ask: why might a reasonable person, under these specific circumstances, have done what they did? You may still conclude it was character. You'll get it wrong far less often.

When making predictions, find the reference class. Don't construct a story about why your plan will work. Find the base rate of plans like yours. How often do comparable projects come in on budget? How often do interventions like this produce the results they're designed to produce? The reference class answer is probably more accurate than your inside view, and it's much harder to manipulate with motivated reasoning.

Judge decisions by process, not outcome. Was the strategy wrong, or was it right and unlucky? These require different responses. A poker player who makes the statistically correct fold and loses made the right decision. A player who makes the statistically wrong call and wins made the wrong decision. If you evaluate past decisions (your own or others') by how they turned out rather than by what information was available at the time, you're letting the story of the ending corrupt your understanding of the beginning.

Takeaways

The narrative trap is not that you tell stories — it's that you mistake them for explanations. A good story selects, compresses, and imposes causality on events that were messy, overdetermined, and contingent. That's what makes it a story. That's also what makes it wrong in ways that are invisible from the inside.

The most reliable sign you're in the trap: the story feels complete. Everything fits. The causality is clean. The protagonist's choices were decisive. The ending makes sense.

Reality rarely looks like that. When your model of the past does, ask what you've left out.

The sophisticated move is not to find a better story. It's to notice when you're constructing one — and ask what probability distribution or reference class should be running alongside it.

Specifically:

  1. Before predicting, ask for the base rate of comparable cases — not a story about why yours is different
  2. After failures and successes, ask what structural factors and luck contributed, not just what the protagonist did
  3. When evaluating others, reconstruct their situation before assigning character explanations
  4. When reviewing past decisions, evaluate them on the information available then, not what you know now
  5. When a post-mortem produces a clean narrative, add a round of deliberately looking for what the narrative excludes

The goal is not to stop using stories. It's to use them at the right scale — for communication and motivation — while reaching for something more rigorous when the question is "what actually happened" or "what will probably happen next."

Stories tell you where to look. They are not a substitute for looking.

Today's Sketch

March 09, 2026