Everyone has biases. And everyone knows that everyone has biases, and that these biases affect our judgements. Bias is explainable, and our brains like things they can explain.
One of the leading explainers of our biases is economist Daniel Kahneman, famed for a Nobel win and his book Thinking, Fast and Slow. He’s now teamed with Olivier Sibony and Cass Sunstein to write a book… that’s… not about bias. Entitled Noise: A Flaw in Human Judgement, it deals with—you guessed it—noise, the variability among human judgements that is the result of humans being variable. We have distinct temperaments and personalities; we are different from each other, and we are different from ourselves, certainly from year to year but also even from hour to hour.
All of that noise is totally OK. But it is totally not OK when it means that one petty thief is granted bail while another must await trial in jail, or one asylum seeker gets admitted into the US while another does not, or one child at risk of abuse gets shunted into foster care while another stays put—all because they saw a particular judge on a particular morning.
An end to noise
The goal of the book is to eliminate noise. The first step is recognizing it, which isn’t easy. Unlike bias, noise is not readily acknowledged. But it contributes just as much to errors as bias does. So eliminating it can reduce errors just as much as eliminating bias can.
In Thinking, Fast and Slow, Kahneman described the Nobel Prize-winning work he did with Amos Tversky about the many cognitive biases that often shape individuals’ decisions and cause them to err. Noise expands on those ideas to explore why groups of people make errors and how to deal with it when they do.
The longest section of the book is devoted to defining and identifying noise—a justification for the measures outlined in the rest of the text. “Noise is variability in judgments that should be identical,” the authors explain.
They are not talking about subjective matters of taste, like movie reviews or wine ratings, about which there obviously should be differences of opinion. They are talking about matters like criminal justice and medical diagnosis, in which a lack of consensus should not exist but does. And it exists to a much, much greater degree than you’d imagine: when shown a pair of fingerprints a second time months after first seeing them, forensic analysts make different decisions about if they match about 10 percent of the time.
This noise can be due to some random factors: doctors order more cancer screenings in the morning and prescribe more opioids (but not other pain meds) in the afternoon. People may want to think that their fates about very important matters are being determined by one of a group of seasoned, caring, interchangeable experts who represent a System and not their own individual values at that moment. But their fates are, in fact, being determined by a process much more akin to a lottery (lotteries are cited a lot in Noise). The expert who makes the decision is in essence randomly selected, and their opinion can differ wildly from that of other experts. And that is Not Fair.
One key aspect of noise identification that the authors stress is that “we don’t need to know who is right to measure how much the judgements of the same cases vary.” In many judgments, the true or right answer is unknown or unknowable; for others, it may not even exist. But that doesn’t matter for recognizing problems. Expert judgments should still cluster together. If they don’t, the system is noisy. (If they do, the judges might still all be biased in the same way—but as noted earlier, that is not a noise problem.)
To reduce noise, the authors prescribe a regimen they call “decision hygiene,” which is about as unsexy as it sounds. And they named it that for precisely that reason. They liken it to hand-washing, which is widely known to do wonders to prevent the spread of pathogens—although when you do it, you don’t know exactly which pathogens you are preventing from spreading, and you’ll never know.
The authors posit that decision hygiene will reduce noise-induced errors, but we’ll never know which ones. The analogy doesn’t end there; they note that hand-washing is also known to be kind of an annoying hassle, so even though it is simple and extremely effective, many people who are aware that they should be doing it all too often don’t. Decision hygiene is similar.
One way the authors suggest we make better judgments and predictions is to have better judges and predictors. And who might those be? Turns out they are people who are not only willing to change their minds when faced with new information, but those who go out and seek new information that challenges their closely held views. They are “actively open-minded,” or in a state of “perpetual beta”—they are constantly integrating new ideas and perspectives, constantly analyzing and refining themselves and their views. They do not adhere to foolish consistencies.
They sound like the exact opposite of political pundits everywhere and many of the authorities we currently revere.
Each chapter ends with a list of talking points, like a soundbite-y highlight reel of the contents. These are in quotes, even though they are not verbatim lines lifted from anywhere in the chapter (this must be Kahneman’s shtick, as he does it in Thinking, Fast and Slow as well).
The nonquote quotations are odd, especially since Noise has so many actually quotable sentences. Consider this gem: “The obviousness of this fact [that the future is unpredictable] is matched only by the regularity with which it is ignored.” Or this one, on the problems people may have with noise-mitigation strategies: “Although we will address these objections as sympathetically as we can, we by no means endorse them.”
And Kahneman, Sibony, and Sunstein can expect no shortage of objections. One of those is that people, especially trained experts, like to think that their experience and their hunches and their gut feelings are invaluable. They want to use their discretion; they rebel against the notion that their intuition could be supplanted by an algorithm (although Kahneman and Co. insist that this would certainly eliminate noise). But “the goal of judgment is accuracy, not individual expression,” the authors write. Creativity and personal values certainly have their place. But not if they lead to injustice.