Superforecasting: The Art and Science of Prediction Kindle Edition

There are two kind of pop-science books; one deep and thoughtful based on years of research, one quick and dirty written by a ghost-writer. This book is of the latter kind. Tetlock wrote Expert Political Judgment: How Good Is It? How Can We Know? about a decade ago. That book was deep and thoughtful. I had expected his new book to be an update with ten more years of research and consulting. Sadly, I am greatly disappointed. The book could have been written totally without additional research input. It starts with a couple of chapters of the history of the standard controlled experiment. There is about 50 pages of real content in the 330 pages of the book.There is a lot of content directly lifted from the web (e.g. Fermi-forecasting, Auftragstaktik) - kind of Malcolm Gladwell style, some insight and some misinterpretation.

The style is **extreme pop-science**. What do I mean with that? Far too many pages, plentiful descriptions of minute irrelevant details of individuals (so called human interest points - I guess that is what they teach in creative writing), never any figure or number (e.g. 67% is changed to two thirds), all difficult material removed or put in a footnote. And how come a book with two authors use the pronoun "I" all the time?

The researcher has run a forecasting tournament for several years. He has loads of data, but he does not provide any analysis in the book. He refers to his research in footnotes, but no explanation or description at all. Instead we get statements like 80% of superforecasters are more intelligent than average. What is wrong with running a regression to find out what characteristics are important? Why spend five chapters going through the characteristics of superforecasters? In the end, apparently, two characteristics stand out. (1) Continual updating of forecasts, (2) Being intelligent. That fact is told after around 200 pages of tedious writing. Wtf? I can reluctantly accept dumbing down the book, but it is inexcusable that the footnotes does not include some further help to the reader that wants more depth.

The author likes to give minute details of the superforcasters. Personally, I don't care that Brian likes Facebook updates of cats, that John is retired because he is sick and that he now likes to collect stuff or that Steve is and old colleague of the author that likes opera. Who reads and enjoy this written muzak? It goes on chapter after chapter. We "meet" 15-20 superforecasters.

There is a lot about the superforecasters in the book, but the title of the book is "Superforecasting". This is a seriously misleading title. It makes you believe that you will learn tools to become a great forecaster. You get some, mostly general, points in an eight page appendix. With the researcher's experience, I would have expected a lot of practical advice.

What is good about the book?
(1) The key message that experts are lousy forecasters and do not want accountability is very important, but that was already in the author's earlier book.
(2) Some useful anecdotes that you probably should pick up if you are teaching/presenting on the topic.
(3) Odd bits of information. I liked the discussion of how the German military used what we consider modern management already 100 years ago. As mentioned earlier, there are 50 pages of really good material in the book.

I bought the hard-cover edition. If you make notes with a normal pencil, be careful because it easily pierces the paper.

The book is worth two stars. If you are en educator and want a few anecdotes, read the book. Others should give it a pass. Instead sign up to the author's forecasting tournament. You learn more by trial and error learning. I signed up two years ago and it is a useful experience. You can also check the video features on edge.org. Then spend time reading better books. A few rigorous pop-science books:
* Another forecasting perspective is Steenbarger's Trading Psychology 2.0: From Best Practices to Best Processes (Wiley Trading). It is about trading in the market, but it covers many of the topics from a different perspective. Worth reading his earlier books too.
* And if you haven't read Thinking, Fast and Slow, that is a more important book (but also too fluffy for my linking).
* You should also read Taleb's The Black Swan: Second Edition: The Impact of the Highly Improbable: With a new section: "On Robustness and Fragility" (Incerto), but don't buy his fluffy version of the same topic Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets (Incerto)

As a counter to the highest rated review here, I agree for opposite reasons. Superforcasting is about the forecaster. If a superforecasting algorithmic processology existed, we'd assuredly be accurately forecasting the objective at a higher rate.

Superforcasting is a function of the forecaster. Most anyone can exercise the statistical, objective, subjective methods of forecasting. The secret sauce of the game is the cash on the barrel risk/reward perception of the bookmaker, the pokerstar, 1997 investors in Amazon stock, Vanguard's Bogle or Warren Buffet. It's all about the forecaster.

If you're looking for the methods, check out Predictive Business Analytics: Forward Looking Capabilities to Improve Business Performance (Wiley and SAS Business Series)

If you're interested in aspects of the Art, Superforecasting does not disappoint.

I (Matt) have really enjoyed reading Super Forcasting: The Art and Science of Prediction by Phillip E. Tetlock and Dan Gardner. It's a fun book for taking a dive into forecasting .

I enjoyed reading about how forecasters related to weather, politics, stocks, etc. are often considered professionals even though they may actually be amateurs. I also liked learning about how even though these people are often very bad or unreliable at forecasting, it is in many ways a skill that can be learned. While it definitely leans more towards pop-science than going academically deep into the presented topic, I feel like I learned some new information from this book.

I do wish there was less fluff and I often found myself wishing for deeper analysis of the years of research the authors kept referencing.

If you enjoy books like Moneyball or Freakonomics then I would recommend this book to you.

**I received this book from Blogging for Books for free.**

This book does a brilliant job of pointing out something that has always bothered me. As long as you talk loudly and proclaim confidence, you will get ahead in this world. These people will be the influencers of the general public rather than basic scientific facts and statistics. If we only applied the same statistical thinking that gamblers use to judge a point spread to area of business and politics, we would get a lot better at knowing what to expect in the near future. Gone are the days of gut feel and intuition - let the numbers win out.

I assumed this book would be very analytical in nature, and touch upon subjects like statistics and probability, etc, which made it kind of daunting. Now that I've begun to read it, I realize that it is a very insightful glimpse into how and why people forecast incorrectly and offers analysis that incorporates psychology and the social sciences. This book encourages people to broaden their scope of thinking and to shed their convictions and tackle things with an element of uncertainty.

Hugely recommend this book!

If you like to think, read this book. It'll help you think better. I especially enjoyed the discussion of Taleb's Black Swans in the context of forecasting. It would be very interesting to read a follow-up on the success or lack of same for superforecasters who have tried to achieve success in the stock market. It's a shame that this book is unlikely to be read by the people who could benefit most from it. I'm thinking of a current public figure who has yet to experience his first moment of doubt!

The introduction of how those super forecaster works is very insightful. However, their high performance rely on the fact that most of the issues they forecast are hard, but not black swan events. Their high Bier score tell something but not everything. What I'm interested is when regarding Black Swan events, are their performance better than experts average? I guess not...