Book review: Noise: A Flaw in Human Judgment

Bad news up front: I found this book hard work and over long.

Having said that, it is — as one would expect from Daniel Kahneman — exquisitely researched, coherently argued and full of loads and loads of insights.

The fundamental insights about the types of noise around predictions, judgement and decision making and, crucially, actions that can be taken to improve judgement in practice are excellent and sometimes counterintuitive.

As I was struggling through the middle of the book, I kept telling myself that it is worth slowing down my learning on this topic and letting it properly sink in because with all the overwhelming weight of evidence (both that presented in the book, but intuitively also the evidence from our everyday business lives), we are clearly individually so bad at prediction and yet continue to believe that we aren’t.. in retrospect, I think I could have done a bit of brisk skimming and be none-the-worse off for it.

Key takeouts for me:
- Humans are not great at consistent coherent judgement and predictions
- Noise (unexpected and unwanted variance in human judgement) is distinct from bias, is overwhelming, but can be broken down into a range of types (occasion noise, stable noise and level noise)
- Decision “hygiene” and material improvements to judgement can be made through processes (and use of algorithms, both simple and complex) and (to an extent) education
- A noise “audit” can be a powerful instrument to kickstart improved decision making in an organisation

I haven’t checked, but if there is a TEDx style presentation on this, I would recommend 20 minutes of that, rather than this entire book!