The big ideas from the world's best books, free in 5 minutes
Like a movie trailer, but for books. Watch, listen, or read along, then go as deep as you want.
Great lessons shouldn't depend on spare time, money, or how you learn. Free in 8 languages, for everyone.
1M+ books on demandListen and read alongStart without an account
The Book of Why by Judea Pearl & Dana Mackenzie
Audiobook Summary and Review by StoryShots
Your data can't tell you why.
You can.
Introduction
For a century, science treated "why" as a forbidden question.
That is the problem behind The Book of Why: The New Science of Cause and Effect, by Judea Pearl, and the revolution that broke the taboo.
Why Correlation Is Not Enough
By the mid 1950s, researchers had tied smoking to lung cancer.
Yet the US government waited until 1984 to mandate the phrase "smoking causes lung cancer."
The holdup was a ghost: perhaps a hidden gene drove both the craving and the disease.
Statistics had no formal way to rule it out.
That silence was no accident.
Early statisticians argued that correlation was the only legitimate basis for scientific inquiry, and the field built itself around "when" questions, never "why" ones.
Then geneticist Sewall Wright drew arrows between causes and outcomes for guinea pig coat patterns, and calculated that heredity explained 42 percent of a coat pattern.
Those arrows were the first causal diagrams.
Every headline claiming that people who do X are healthier hides a question nobody answered.
Seeing a pattern is easy.
Knowing what to do about it is harder.
Climb the Ladder of Causation
All reasoning sits on three rungs.
The first is seeing: noticing which symptoms point to a disease.
Most animals, and most machine learning systems, live here.
The second is doing: what happens when you take the drug or change the policy.
The third is imagining: whether your headache would have vanished without the aspirin.
The middle rung hinges on one idea.
Seeing that X occurred is not the same as setting X yourself.
The gap between those two probabilities is confounding, and randomized trials exist to close it.
Causal diagrams close it when a trial is too expensive or impossible.
Here is the catch.
Two different causal structures can generate identical data, so no pile of data settles which one is true.
Only a model of the world can.
Every other creature sees what is.
Humans alone ask what might have been.
That third rung is where blame, regret, and credit come from.
It is also where machines still fall short.
You Are Smarter Than Your Data
If one line sums up the book, it is this: you are smarter than your data.
Data do not understand causes and effects.
Humans do.
Deep learning has built machines with impressive abilities and no intelligence, because they lack a model of reality.
A system that only fits curves cannot explain why turning a barometer dial will not cause rain.
The same rule applies to your decisions.
Before trusting a statistic, sketch the arrows.
Ask what else could cause both the thing you changed and the thing you measured.
If you cannot answer, more data will not save you.
Models do not replace thinking.
They give it words, and you cannot answer a question you have no words for.
A 2016 climate study used exactly this logic and put the odds at 90 percent that Europe's 2003 heat wave would not have happened without human-driven warming.
Data show you what moves together.
Only a model tells you what moves what.
If this changed how you read a headline, someone in your life probably needs to hear it too.
Final Summary
This summary of The Book of Why by Judea Pearl ties three ideas into one thread.
Correlation was never enough, three rungs separate seeing from understanding, and your causal intuition beats raw data.
But the full book goes deeper, into confounders, mediators, and the traps behind Simpson's paradox.
For the full summary of The Book of Why, head to the StoryShots app.