The Book of Why | Judea Pearl

Summary of: The Book of Why: The New Science of Cause and Effect
By: Judea Pearl

Introduction

Embark on a thrilling journey through the world of cause and effect as we explore Judea Pearl’s groundbreaking book, ‘The Book of Why: The New Science of Cause and Effect’. Discover the vital elements behind causality and its significance in various research fields from medicine to climate science. Learn about the Ladder of Causation, a unique concept that sheds light on the various processes behind causal thinking. As we delve deeper into this fascinating world, you’ll see that understanding causation is paramount, especially in the realms of data interpretation and the possible future of artificial intelligence.

The Causal Revolution

The scientific community for the past few decades has downplayed the idea of causation, with “correlation does not imply causation” being repeated ad nauseam. Karl Pearson, a renowned English mathematician, epitomized this view, claiming that science was nothing more than pure data and that causation was scientifically invalid. However, this attempt at ridicule only hid the causative factor. Geneticist Sewall Wright disproved Pearson’s theory by using data to demonstrate that causation could be represented mathematically. His methods were heavily criticized at the time but are now being revived as research fields from medicine to climate science are beginning to embrace causation as a principle. The Causal Revolution has begun.

The Ladder of Causation

Have you ever considered the importance of analyzing data before drawing conclusions? The Ladder of Causation explains why this is crucial. Take the smallpox vaccine for example — early data showed it may have caused more deaths than it prevented, but this conclusion was misinterpreted. By looking beyond the initial observations and asking the right questions, we can uncover important insights that would have otherwise gone unnoticed. The Ladder of Causation offers a process to climb that helps us understand the common causes of seemingly unrelated data points. So the next time you see a strange correlation, remember to climb the ladder.

The Ladder of Causation

The human tendency to look for cause-and-effect relationships is the first rung on the Ladder of Causation. However, neither animals nor AI can get past that step. Data collection also falls short of determining causality. While observing basic probability may suffice for some inquiries, it is not nearly enough for most other occasions.

Progressing up the Ladder of Causation

Humans’ active influence on outcomes using controlled experiments

Humans have the unique ability to actively influence outcomes, progressing up the ladder of causation. To reach the second rung of this ladder, one must ask the question “What if we do…?” and then take action. Unlike the passive first rung, this requires active intervention in influencing outcomes.

For example, the dental hygiene marketing manager may ask, “Will floss sales be affected if we change the price of toothpaste?” Computers cannot currently be programmed to accurately ask these questions, limiting their ability to progress beyond the first rung.

One of the best ways to test the effect of something is through a controlled experiment, where groups as similar as possible are compared, and a test is applied to one but not the other. This method allows variables and their effects to be objectively measured and isolated.

Controlled experiments are not a new concept; they have even been reported in the Bible. In the story of Daniel, Nebuchadnezzar sought out captured nobles to educate in his elite Babylonian diet. Daniel suggested a controlled experiment where he and three friends were given a vegetarian diet while another group had the king’s diet. After ten days, Daniel’s group flourished, and Nebuchadnezzar gave them high court positions.

In modern times, Facebook utilizes controlled experiments by experimenting with different page configurations and comparing groups that see different settings. In conclusion, actively influencing outcomes through controlled experiments is a vital and effective approach in progressing up the ladder of causation.

The Power of Counterfactuals

Humans have a unique ability to imagine how different actions can lead to different outcomes. This ability is put into practice using counterfactual models, which help picture what would have happened if another action had been taken. Counterfactuals are not always an easy concept for machines to understand. In counterfactual questions, humans often skip over the ordinary or expected causal factors and concentrate on the unusual or extra ones. For example, saying a house burnt down because of a lit match rather than oxygen. Machines don’t automatically make these sorts of distinctions and have trouble understanding the concept of necessary and sufficient causes of an event. Understanding the three rungs of the Ladder of Causality is important in helping answer causal questions, and this leads us to the question of which complicating factors should be identified for scientific studies.

Want to read the full book summary?

Leave a Reply

Your email address will not be published. Required fields are marked *

Fill out this field
Fill out this field
Please enter a valid email address.
You need to agree with the terms to proceed