Superforecasting (Book Summary)

Forecasting is not a “you have it or you don’t” talent. It is a skill that can be cultivated

To be a superforecaster:

1.      Focus on questions where are efforts are likely to pay off

2.      Forecast with a time frame, unambiguous key terms, and numerical probability scales. For example: Certain 100%; Almost Certain 93%; Probable 75%; Chances about even 50%; Probably not 30%; Almost certainly not 7%; Impossible 0%

3.      Look no more than 1 year out

4.      Double check by seeking both confirming and especially disconfirming information by asking: “What would convince me I am wrong?”. Stated another way, assume your initial judgement is wrong. Leave no assumptions un-scrutinized. [CAUTIOUS]

5.      Benchmark against (a) no change to current level or trend (b) the consensus of other forecasters

6.      Draw upon many analytical tools such as base rates and regression to the mean. When you have a well-validated statistical algorithm, use it. [ANALYTICAL]

7.      Gather information from as many sources as possible [DRAGONFLY EYED]

8.      Be especially wary of cognitive and emotional biases, esp. in your area of domain expertise [GOOD INTUITIVE PSYCHOLOGISTS]

9.      Engage in group forecasting. However, effective groups (a) do not get along too well (b) speak on any topic (b) have a license to respectfully challenge/question anything = adversarial collaboration (c) bring in outsiders (d) suspend hierarchy (e) have a nucleus of 5 or 6 core members who do most of the work (f) divide the workload to be able to do more work (g) admit ignorance and request help (h) diverse along many dimensions

10.  Systematically track accuracy and believe it is possible to get better no matter how long it takes [GROWTH MINDSET; GRIT]

11.  “Fermi-ize” = Break down complex questions into the knowable and the unknowable

12.  Apply fine-grained probabilistic thinking (more than yes/no or yes/maybe/no) [PROBABILISTIC]

13.  Update frequently in the light of new information. Generally, use small adjustments unless the information is extreme. [THOUGHTFUL UPDATERS]

14.  Pride yourself in changing your opinion and be willing to admit mistakes (which is not natural, esp. when you have made public commitments). [HUMBLE]

15.  Regularly ask “Why?”; don’t believe in fate; Finding meaning in (random) events is positively correlated with well-being but negatively correlated with forecasting ability) [NONDETERMINISTIC]

16.  Conduct pre-mortems to determine what might go wrong and what has to happen for things to go right

17.  Conduct post-mortems of your correct and incorrect forecasts [REFLECTIVE]

18.  Avoid scope insensitivity by exploring how your answer might change with a different timeframe

19.  Appreciate the value of super-questioners

Innate characteristics superforecasters:

1.      Score higher than 80% of the population on knowledge and intelligence tests. Regular forecasters scored 70% higher. [INTELLIGENT & KNOWLEDGEABLE]

2.      Score high on “need for cognition” tests – they engage in and enjoy hard mental slogs like solving crosswords and Sudoku puzzles. [NEED FOR COGNITION]

3.      Are comfortable with numbers [NUMERATE]

4.      Score high on Big 5 personality trait of openness to experience) [ACTIVELY OPEN-MINDED]

On leadership:

1.      Never entirely trust your plan.

2.      Apply mission command (Auftragstaktik) - tell subordinates what their goal is but not how to achieve it.

3.      Preserve and promote the out-of-the-box thinkers

4.      Promote an obligation to dissent but fall-in-line one a final decision has been made

5.      Be (intellectually) humble in the face of the game but not in the face of your opponents

6.      Plan for surprise