Data and Causality

2025-01-25 | By Mariusz Jażdżyk

For many years, I believed that being “data-driven” was the key to solving many business problems. We followed this dogma across various industries, convinced that it was the optimal path.

Today, we have better algorithms and 10 times more experts, but... it’s still hard. :-/

A Revolutionary Approach to Data and Causality

Recently, I came across an incredibly inspiring book – Judea Pearl, in “The Book of Why: The New Science of Cause and Effect,” shows that data is just the first step. 📊

👉 Simply having data is not enough – data without understanding the model or process behind it is useless.

Three Levels of Reasoning:

  1. Data – Analyzing Correlations Example? The increase in the number of firefighters at the scene of a fire correlates with greater material damage. Does a higher number of firefighters cause more destruction? Of course not, but that’s what the correlation shows. 🔥

  2. Causality – Understanding What Causes What It's not the number of firefighters that causes damage, it’s the size of the fire. To establish this, we need knowledge of processes, not just the data itself. 🧠

  3. Simulations and Speculations – What If? For instance, what happens if the fire department responds faster? Will it reduce the damage? This is the level of virtual experiments and predictions that open up new possibilities. 🌍

The Brain Still Outshines in Causal Reasoning

AI? Not quite at that level yet, but we’re progressing quickly.

The Ideal: A Combination of Data, Expertise, and Advanced Linguistic Tools

A recommendation system that utilizes data, expert knowledge, and advanced linguistic tools could be the ideal solution.
Maybe it will be created soon? ;-)


Author: Mariusz Jażdżyk

The author is a lecturer at Kozminski University, specializing in building data-driven organizations in startups. He teaches courses based on his book Chief Data Officer, where he explores the practical aspects of implementing data strategies and AI solutions.