Noise
A Flaw in Human Judgment
Kahneman, Sibony and Sunstein explain 'noise', the unwanted variability in judgments that should be identical. They show how organisations can reduce it.
Bias is systematic error, and noise is random scatter. Two judges can give very different sentences for the same crime, and two doctors can give different diagnoses for the same patient. The same person can even judge differently depending on mood or time of day. Noise is everywhere, often larger than bias, and largely invisible until you measure it.
Written with AI; may contain mistakes.
- Noise vs. bias. Both add to error. Reducing either one improves accuracy.
- Noise audits. Have several people judge the same cases to reveal how much they vary.
- Decision hygiene. Structure judgments, break them into independent parts and aggregate independent views.
- Rules and algorithms. Simple models are often less noisy than human experts.
Written with AI; may contain mistakes.
- For important decisions, gather independent opinions before discussing them.
- Use checklists or scoring rubrics to structure assessments such as hiring.
- Delay forming an overall intuition until you've evaluated each factor separately.
Written with AI; may contain mistakes.