The Narrative Fallacy

"He dropped out of college, wrote code in a rented room, was turned down by dozens of investors, but never gave up, and his company eventually went public." Stories like this are inspiring, and they make success seem like it follows a recipe: just persist and you'll make it.

But the people who dropped out, persisted just as hard, and failed never appear in the story.

This is the narrative fallacy: we naturally love coherent stories, so we string scattered facts into tidy cause-and-effect plots, and as a result overestimate how well we understand and can predict events.

The more complete the story, the more likely it's wrong

Compare two statements:

• A: This company's revenue will fall next year.
• B: Next year this company's key executives will leave and its new product will be delayed, so revenue will fall.

Many people find B more believable because it has causes and a plot. But B requires more things to happen at once, so it can only be less likely than A. Suppose executives leave with 30% probability, the product is delayed with 50%, and the two together lead to a revenue drop with 80%. Then B comes to just 0.3 × 0.5 × 0.8 = 12% (assuming the first two are independent), while A might be 40%. This is the conjunction fallacy from lesson 1.3.

Stories crowd out luck and base rates

In a good story, everything happens for a reason and there's no room for luck. So we forget to ask: how likely is this kind of thing in general? How many identical stories ended differently?

How to guard against it when forecasting

• Do the math on your story: count how many links in your reasoning must all hold, and multiply their probabilities.
• Write a second story: using the same facts, can you tell an equally plausible story that points to the opposite outcome?
• Check the base rate first: however compelling the story, first ask how often this kind of thing happens.

How to spot it in a review

Look back at your reasoning: was it a complete, tightly linked story with no numbers in it? Did you give a high probability because the story "made so much sense"? If so, the narrative fallacy may have played a part.