I learned about Daniel Kahneman’s new book from a colleague just a few days before its release, I read the synopsis and pre-ordered immediately. “Noise: A Flaw in Human Judgment” came out mid-May 2021. This is not a review, but only my reaction in response to the authors’ central thesis. And it looks like I had a strong reaction.
So the thesis is this: human judgment is flawed because of the variability in decisions among people presented with the same information, and when this “noise” is undesirable, it should be minimised. The book gives some 400 pages explaining why this is a problem and how to fix it. In one small section the authors acknowledge that sometimes eliminating noise means that you won’t have a variety of perspectives, but quickly move on to say if you want diversity of opinions, it’s not noise anymore.
My central problem with the idea of noise is precisely this. Labelling something as noise means valuing it as undesirable, disturbing and unpleasant. As someone who needs to walk around in noise-cancelling headphones at home, I know all too well how irritating and even plain sickening too much noise can be. But I don’t write to local authorities to stop road maintenance works because the machines are on whole day long. What is noise to me, is essential repairs to the whole community.
In the same way, when variability between human assessments of the same information is labelled as noise in decision-making, it’s automatically assumed to be bad. Some people out there would like us to be fully rational decision-making machines, like these mythical ‘unbiased’ algorithms. No own history, no biography, no perspective, no opinion — just a sober assessment of facts. One, this is impossible. Two, it would make the world worse off.
We do have our histories, biographies, and perspectives — and this is what makes us human. We bring them to our lives and yes, decisions too. And it’s very much needed because we make these decisions for other people, who also have their histories, biographies, and perspectives. In many cases it’s not a flaw, it’s a feature.
Labelling this diversity of perspectives as noise is dangerous. It becomes all too easy to dismiss people’s voices as just some noisy rumble. Especially when those who get to decide are privileged and in positions of power. Writing about noise this way in a book aimed at high-powered executives, while giving examples of judges and doctors, only strengthens the message: you get to decide what noise is. If you’re disturbed by the works outside, take the local council to court.
One of the book’s proposals to counter this variability is to use algorithms that always give the same result, given the same information. And this is where I get really upset. For me, digital technology brings the unprecedented possibility to express a variety of perspectives and views that can be now heard — and used to be silenced by the powers that be before. It’s an engine of plurality and a microphone for diversity. In other words, this technology is wonderful because it gives more people more voice. Turning it against these voices makes me want to scream even louder.