I just finished reading “Superforecasting” by Philip Fetlock. He initiated the Good Judgement Project, “a multi-year study of the feasibility of improving the accuracy of probability judgments of high-stakes, real-world events”. The project came first in the forecasting challenge by the IARPA, a US intelligence community observer.
The book, in the first part, covers the results of the Good Judgement Project in an interesting not over-academic way. The conclusion of the project, at least from my perspective, is that you should never think binary and always in the Bayesian way. If you are familiar with Bayesian statistic, you will feel that the whole project will culminate in the Bayesian rule already after having the first ten pages. The participants of the forecasting project are described, what they do and what their results are.
I thought it to be interesting that Fetlock never explicitly states that these participants are just building intuition based Bayesian models of the world which allows them to generate more fine grained forecasts. For me, the book basically summed up an understanding of the world I was familiar with before, but with different words and some empiric evidence. Very useful.
I very much appreciated the remarks in the last chapter about the limits of quantification. I want to cover these in my Bachelor thesis, so this book definitely will get a reference or two.
Overall, this book is a very good read and one more piece of a puzzle called uncertainty I am trying to solve. Recommend. Approx. 18 hours.
The book, in the first part, covers the results of the Good Judgement Project in an interesting not over-academic way. The conclusion of the project, at least from my perspective, is that you should never think binary and always in the Bayesian way. If you are familiar with Bayesian statistic, you will feel that the whole project will culminate in the Bayesian rule already after having the first ten pages. The participants of the forecasting project are described, what they do and what their results are.
I thought it to be interesting that Fetlock never explicitly states that these participants are just building intuition based Bayesian models of the world which allows them to generate more fine grained forecasts. For me, the book basically summed up an understanding of the world I was familiar with before, but with different words and some empiric evidence. Very useful.
I very much appreciated the remarks in the last chapter about the limits of quantification. I want to cover these in my Bachelor thesis, so this book definitely will get a reference or two.
Overall, this book is a very good read and one more piece of a puzzle called uncertainty I am trying to solve. Recommend. Approx. 18 hours.