Where there is human judgement, there is noise.
When and where does human make judgments? All the time and everywhere, right? Job recruitments, insurance claims, criminal sentencing, office and meeting, essay marking and grading, radiology, medical diagnosis, psychiatry (of course!), forensic investigation (noise exists even in finger print).
A lot of originality, scientific data and creativity. The interview with the authors in the end showed how much “self-criticism” they went through in order to produce such a piece of work. Noise has always exist (just like bias), but it’s never the centre of attention…
Very importantly, often people think they cancel each other out, but they don’t! Student A receives a better grade than what she should have deserved while student B receives a worse grade than what she should have deserved, does that cancel each other out? Criminal A receives higher sentencing while criminal B receives a much less sentencing that she should have deserved, does that cancel out? The more interesting thing is, the time of the day, the time of the week, whether the previous weekend the judge’s favourite football team win a game, whether the sentencing is done before or after lunch or when it’s almost time to go home etc, all these affect a judge’s decision! And the difference could be 5 years or 20 years in prison!
On top of discussing noise with many case studies and statistics, they also came up with “decision hygiene”, how we can reduce such noise and biases. Can algorithms (artificial intelligence for example) solve the problem? What other options do we have? What can we do to reduce these noise and biases? (well, read the book for answers!)
Very interesting, yet for me a very hard to read book, especially after the first few chapters. I finished listening to it on the Libby App (somehow they don’t provide the PDF file the comes with the audio book) in about 6 weeks, took me about 12 hours (I listened to it 10 to 15% faster than the normal speed).