REVIEW - The Book of Why - The New Science of Cause and Effect


Title:

The Book of Why

The New Science of Cause and Effect

Author:

Judea Pearl, Dana Mackenzie

Publisher:

Penguin (2019)

Pages:

432

Reviewer:

Alison Chaiken

Reviewed:

January 2025

Rating:

★★★★☆


Recommended.

Bayesian network pioneer and Turing Award winner Pearl’s semi-popular book explains how data analysts can manipulate graphs to reason about causality. Pearl introduced pictorial representations of causal models which make underlying assumptions transparent, and which also provide data analysts a short-hand language with which to compare models. The authors convincingly demonstrate how graphical presentation of hypotheses about relationships simplify reasoning. In that regard, Pearl’s causality graphs play a role not unlike Richard Feynman’s famous diagrams in physics.

The Book of Why is a semi-popularization in that the text has a limited number of equations, but the presented examples have great explanatory power. The ideas are illustrated both by artistic drawings and by entertaining case studies from fields including epidemiology, education, public health, battlefield medicine, law enforcement, polar expeditions and pet training, among others. Pearl clearly has no fear of making enemies, so the text is replete with examples of erroneous published studies, some recent. The authors observe that “causal questions can never be answered by data alone” and criticize the tendency to blindly plug numbers into formulae without first clearly stating a rationale based on a falsifiable hypothesis.

The Book of Why presents capsule biographies of pioneers like geneticist and causality diagram inventor Sewall Wright. The story of John Snow and the London cholera epidemic illustrates the instrumental model method of analysis. Another chapter recounts the bitter controversy over smoking and cancer, which serves to introduce the concept of sensitivity analysis. A chapter on mediated chains of causation drives home the point that no statistical manipulation can derive information from data if it simply is not present. The most poignant story is that of brilliant Stanford graduate student Barbara Burks, whose PhD thesis presented a correct analysis of genetic influence on intelligence. Her conclusions were rejected by her academic advisor, and, unable to find a job, she committed suicide.

The beginning of the text is lighter on formulae than a mathematically inclined reader might prefer. Indeed, the authors note that “a formula is a baked idea. Words are ideas in the oven.” The reader who persists will find that by the time the ‘Monty Hall problem’ pops up in chapter 6, graphical analysis makes the answer obvious. A major takeaway is that causal networks have a limited set of basic patterns which graphs make plain. The analogy to UML diagrams and algorithmic patterns is apparent.

Pearl divides reasoning about data into 3 rungs on the ‘Ladder of Causation’, namely Association (what), Intervention (how) and Counterfactuals (what if). These topics are handled in order with successively more sophisticated techniques. Pearl’s career-long goal has been to give automated systems the power to understand causes of phenomena, and thereby to reason independently about their own errors and possible future corrections. The last chapter explicitly treats ‘Big Data, Artificial Intelligence and the Big Questions’, and argues that automated graphical analysis is the key to progress on General Intelligence.

While the The Book of Why is not a textbook (that would be Pearl’s Causality: Models, Reasoning and Inference), a motivated reader with a knowledge of graphical computations can infer algorithms from the text. The end matter contains copious references to related technical papers with more details.

Website: https://www.penguin.co.uk/books/289825/the-book-of-why-by-judea-pearl-and-dana-mackenzie/9780141982410






Your Privacy

By clicking "Accept Non-Essential Cookies" you agree ACCU can store non-essential cookies on your device and disclose information in accordance with our Privacy Policy and Cookie Policy.

Current Setting: Non-Essential Cookies REJECTED


By clicking "Include Third Party Content" you agree ACCU can forward your IP address to third-party sites (such as YouTube) to enhance the information presented on this site, and that third-party sites may store cookies on your device.

Current Setting: Third Party Content EXCLUDED



Settings can be changed at any time from the Cookie Policy page.