6 Finding ‘Facts’ for Good Thinking

Nate Silver wrote a controversial book with the title The Signal and the Noise: Why many predictions fail but some don’t.3 Silver spends much of the book reflecting on military actions in the theatre of war and also in military intelligence operations.

Among many other insights he made these four observations.

1. Extrapolation is a technique of making predictions based on current facts or knowledge. When we use extrapolation, we assume that a current trend will continue. Extrapolation is a basic form of prediction. Significant failures occur if we continue to extrapolate without considering the consequences, validity and likelihood that such a trend will continue, perhaps indefinitely.

1. A signal is some piece of data in a system that has meaning and many be a predictor of some future outcome. A signal is possibly an indicator of a possible correlation and should not be confused with causation. That is, it may be related to another variable but it unlikely to cause something to happen. Noise is something random and is unlikely to be correlated with other variables or outcomes. However, at times noise can hide or obscure a signal.

1. A signal, in Silver’s view, is an indication of the underlying truth of a predictive problem and thus should be respected. Noise is a consequence of random patterns and can be mistaken for signals. In military intelligence the absence of signals can signify something important and the presence of too many signals can make the difference difficult to distinguish from noise.

1. When something is unfamiliar, we tend to dismiss it as improbable. We confuse the unfamiliar with the improbable. When something looks strange the tendency is to think of it as improbable and therefore unlikely and not taken seriously.