Noise

Most writing on judgment, including Kahneman's own Thinking, Fast and Slow, concerns bias: errors that lean in one direction. This book is about the other component of error, unwanted variability. When an insurance company the authors worked with had its underwriters price the same fictional customers independently, executives expected the quotes to differ by about 10 percent. The median difference was 55 percent. Sentencing, radiology readings, asylum decisions, performance reviews and forecasts show the same problem.

The authorship gives it weight. Kahneman won the 2002 Nobel prize in economics for his work on judgment and decision-making. Sibony, a professor at HEC Paris and former McKinsey partner, studies strategic decisions, and Sunstein, a Harvard law professor who ran the White House regulatory office under Obama, co-wrote Nudge.

Critics praised the scholarship and practical remedies but found the book long and dense. The Guardian suggested it could have been half the length. In Prospect, Michael Blastland called the claim of discovery overblown, since noise is a familiar idea in statistics, questioned some of the studies cited, and argued that where errors carry lopsided costs, pulling everyone toward the average can increase risk. Others warned that stripping out noise can make institutions rigid.

Read it for the noise audit and the decision hygiene toolkit, both of which apply directly to credit committees, investment committees and analyst ratings.

1. Run a noise audit. Give several professionals the same realistic cases and measure the spread. Most organizations badly underestimate how much their experts disagree.
2. Noise is invisible case by case. Each decision looks reasonable in isolation, so variability shows up only when you compare many judgments of the same problem.
3. Aggregate independent views. Averaging judgments made separately, before discussion, cancels out noise, while open group debate can amplify the first or loudest opinion.
4. Break big decisions into parts. The mediating assessments protocol has evaluators score separate dimensions independently before forming an overall view, much as a disciplined investment memo separates business quality, valuation and risk.
5. Simple rules often beat experts. Statistical models and checklists frequently outperform professionals because they apply the same weights every time, a case for quantitative screens and scoring systems in credit and hiring.

The book is organized in six parts, moving from discovering noise to measuring it, explaining it and reducing it.

1. Finding noise. Opening chapters present the evidence from criminal sentencing and from the insurance company's noise audit, and define noise as unwanted variability in judgments that should be identical.
2. Measuring error. The authors treat the mind as a measuring instrument and show how error breaks down into bias and noise, and why reducing either one improves accuracy by the same amount. They also show how mood and timing shift a single person's judgments and how group discussion can amplify noise.
3. Noise in predictions. They compare human judgment with simple rules, statistical models and machine learning, and explain why models often beat experts, partly because they are noise-free.
4. How noise happens. Here they examine the psychological sources of noise, including heuristics, matching and the way people use rating scales, and break noise into level, pattern and occasion components.
5. Improving judgments. Case chapters on forensic science, forecasting, medical diagnosis, performance ratings and hiring introduce decision hygiene, which includes aggregating independent judgments, structuring decisions into separate assessments and the mediating assessments protocol for strategic choices.
6. Optimal noise. The authors weigh the costs of noise reduction, including rigidity, loss of dignity and the cost of rules, and suggest where some noise is worth tolerating.

From the world-leaders in strategic thinking and the multi-million copy bestselling authors of Thinking Fast and Slow and Nudge, the next big book to change the way you think. Wherever there is human judgment, there is noise.