The Signal and the Noise

My prior belief was that as someone who loves Bayesian probability, this book would be enjoyable. After reading it, was happy that I was correct, even if I was disappointed I did not attach a probability to my prior belief.

Because that’s what the book is about: avoid thinking of predictions in terms of binaries, attach probabilities to predictions. If there is one thing you will learn from the book, this would be it.

If there are two things you will learn from the book, is that while practice might not make perfect, it does make you better.

The third, and possibly most important thing you can learn, is that not all predictions are equally hard. Predicting the cents value of a stock tomorrow is harder than predicting when the sun will rise.

These three things, taken together, answer the question the book raises: why so many predictions fail – but some don’t. It even answers it somewhat actionably.

Think probabilistically, use trial and error, and be aware that there might be an upper limit, or at least a sharply deciling effort/qualty curve.