NOISE: A Flaw in Human Judgment - Daniel Kahneman, Olivier Sibony and Cass Sunstein

1. Noise vs. Bias: Two Components of Error. The book uses a simple, powerful formula to explain total error in judgment: $Total\; Error=Bia{s}^2+Nois{e}^{2.}$This equation demonstrates that both bias (systematic error) and noise (random scatter) contribute to inaccuracy. While biases are often discussed, the authors argue that in many professional settings, noise accounts for a larger portion of the total error. 

2. The Different Types of Noise. The authors break down noise into three main categories:

• Level Noise: The stable, consistent differences between people. For example, some judges are consistently harsher than others, and some loan officers are consistently more lenient.
• Occasion Noise: The unpredictable variations in a single person's judgment over time. A judge's sentencing decision might be influenced by their mood, the weather, or whether they've just had lunch.
• Pattern Noise: The most significant type of noise, this is the unique and idiosyncratic way each person reacts to the specific details of a case. This is a form of "stable pattern," where individuals respond to the same stimuli in a consistent, but different, way from one another.

3. The Pervasiveness and Cost of Noise. Noise is everywhere that human judgment is involved. Its presence leads to serious consequences, including:

• Unfairness: Individuals are treated differently for the same circumstances, leading to a "lottery" effect in systems like the judicial system.
• Financial Costs: Businesses lose money due to inconsistent pricing, hiring, or forecasting.
• Reduced Trust: The public's faith in institutions and professional expertise is eroded when judgments appear arbitrary or inconsistent.

4. How to Combat Noise: 

• Conducting "Noise Audits": The first step is to measure the amount of noise in a system. By having multiple professionals evaluate the same case independently and then comparing their judgments, organizations can reveal the extent of their noise problem.
• Structuring Judgments: Rather than relying on a holistic, intuitive judgment, the book recommends breaking down complex decisions into their constituent parts and evaluating each one separately. This delays "premature intuition" and forces a more disciplined, consistent process.
• Using Relative Scales: Instead of vague, absolute scales (e.g., "excellent," "good," "fair"), the authors suggest using relative judgments that compare a case to others.
• Algorithms and Rules: The book acknowledges that for some decisions, simple rules or algorithms can outperform human judgment by completely eliminating noise. While this can introduce its own biases (if the algorithm is designed with them), it's a powerful tool for achieving consistency.
• Decision Hygiene. The good news from the book is that while biases are difficult, if not impossible, to eliminate completely, noise is something that can be systematically reduced. The authors advocate for "decision hygiene"—a set of preventative strategies to clean up and structure the judgment process.