Introduction
Your decisions are far less consistent than you think
Most people know about bias. Kahneman, Sibony, and Sunstein want you to notice a different failure mode: noise. It is the random scatter in judgments that should be the same, like two doctors reading the same scan differently, two judges giving wildly different sentences, or two interviewers walking away with opposite views of the same candidate.
That matters because noise is not just annoying variation. It creates unfairness, wastes money, and quietly wrecks decision quality inside organizations that assume they are being rational. Once you see it, you start noticing a hidden lottery almost everywhere judgment is involved.
Two kinds of error
Bias gets the attention, but noise often does just as much damage
The book opens with a shooting-target metaphor. Bias is when the shots cluster in the wrong place. Noise is when the shots are scattered. A system can be biased, noisy, or both. The crucial point is that you can measure noise even when you do not know the true answer. You only need multiple judgments of the same case.
That is why noise is such a practical concept. If physicians disagree about a diagnosis, or executives disagree about a forecast, you do not need to know who is right to know the system is unreliable. You can look at the back of the target, see the scatter, and know something is wrong.
The authors pile up examples that make the problem hard to ignore: asylum judges in one study admitted anywhere from 5 percent to 88 percent of applicants depending on which judge they got; software developers gave task estimates that differed by 71 percent on average when asked twice; interviewers and performance raters often disagree more because of who is judging than because of what is being judged. Noise is not a side issue. In many settings, it is the hidden source of unfairness and cost.
Takeaway
If similar cases produce different judgments, you have a quality problem even when no one can prove which judgment was correct.