Types of noise

In my blog about “noise” last week, I did as if there is only one type of noise. Also Kahneman et al. do so in the beginning of their book Noise. In fact, there are several types of noise. It is important to know them, if you want to reduce noise, for each type requires a different approach. This blog explains these types of noise. If you want to talk about these types of noise in general terms, you can use the term system noise: variability in judgments of the same case, anyhow.
Kahneman et al. often use judicial cases, so let me start doing so as well. In court there are no objective judgments for comparable cases, for a judgment does not only depend on the objective facts of the case (like that a man with a white collar job who steals € 10,000 always is sentenced to one month in prison, or to a fine of € 5,000 euros), but a judge is supposed to consider the circumstances of the crime (did the man steal for the first time or is he a recidivist; does he really need the money, for instance, because he cannot pay the rising energy prices or does he have a big bank account; etc.). Nevertheless, some judges have the reputation for being harsher than average and others for being more lenient. Such a deviation from the average is called level noise: The judgments of several judges are scattered around the (virtual) average judgment of comparable cases (the virtual “right judgment”, as we could say). Of course, we can wonder whether there really exists a right judgment or, generally, a right decision in a case where we must decide, but there is no denying that on average some people are harsher in their judgments than others; some are always more optimistic when they must decide and others are more pessimistic; and so on. It is often unwished-for; it is something we want to avoid, since it is noise.
Even though some judges are harsher on average and other judges more lenient; even though some deciders are more optimistic on average and others more pessimistic; even then there is variation in this “harshness” or “optimism” depending on the case at hand. A certain judge may be generally harsher than average but relatively more lenient towards, for instance, white-collar criminals. A lenient judge may on average give harsher sentences to recidivists. A generally optimist sales manager may be pessimist when she must decide about future investments. In other words, whether judgments, decisions and what more are harsher, more optimistic, etc. than average is not once and for all so for a certain person: it depends also on the types of cases at hand. A harsh judge may judge some types of cases harsh, given his reputation, and other types of cases relatively lenient. There is often a kind of pattern in his harshness. Therefore, this type of noise is called pattern noise. Note that we are still talking about unwished-for deviations from virtually right judgments (decisions, or what the case may be), if many judges (deciders, etc.) would judge the case. This type of noise makes that your judgments or decisions are scattered around your decision style (harshness or leniency, for instance).
Then there is yet a quite banal kind of noise: occasion noise. All kinds of accidental and circumstantial factors may have an influence on what you judge and decide. To mention a few: fatigue; mental pressure; that the sun is shining or that it is a rainy day; that your daughter has birthday; that your favourite team has won. All such occasional factors and circumstances may influence your mood and can influence the way you judge and decide; in general the way you act. They make that your judgments or decisions are scattered around what you personally ideally (free from this kind of noise) would decide.
As we have seen, noise is not a unidimensional phenomenon. There are at least three kinds of noise. This has practical consequences, namely that there is not one method to reduce noise. So, although level noise and pattern noise may be reduced by a training that makes you aware of the problem, it will not reduce fatigue. On the other hand, sleeping well will not have an impact on the other kinds of noise. Although noise is a general problem, it requires a specific approach, not only according to type but also in agreement with the problem at hand, for deciding as a judge which judgement to pass is very different from deciding as a manager how much to invest, for instance.

Source
Daniel Kahneman, Oliver Sibony, Cass R. Sunstein, Noise. A Flaw in Human Judgment. London: William Collins, 2021; pp. 69-93.