Superforecasting

This book is about making good predictions, and argues that this is a skill that can be developed.

Most of it revolves around Tetlock's Good Judgement Project, a contest on predicting geopolitical events, with the surprising outcome that the top forecasters consistently did better than government officials with access to classified information.

I would recommend this book if you want to learn about forecasting and happen to be interested in geopolitics; the useful advice in this book is hidden in the myriad anecdotes around US politics.

Takeaways

1. Forecasting is a skill. You can improve it with practise and good technique.

2. Think in terms of distributions, not absolutes. This allows for nuanced predictions and forces you to think about the uncertainty of your guesses.

3. Update your beliefs frequently. The best performers in the study made small frequent updates to their predictions.

4. Answer the question. You may subconsciously substitute a difficult question with a similar but easier one. Break the given question down into smaller manageable chunks to answer the question that was actually asked.

5. Making granular predictions (eg 81.5% instead of a round 80%) can lead to better accuracy.

6. Be actively open minded. The best performers actively seek information to challenge their assumptions.