Models can be extraordinarily helpful in effective forecasting and decision-making. They deduce relationships that are increasingly difficult to find amid the exponential rise in the variety, volume, and velocity of data. Conversely, models can be equally dangerous when put in the hands of people who cherry-pick models that conform to their predisposed behavioral biases.
Scott E Page, a Professor of Complex Systems, Political Science, and Economics at the University of Michigan, is giving a free course entitled “Model Thinking.” Below are some notes from the first session on using models to help you become a clearer thinker, more intelligent person, and a better forecaster.
Page began by citing the research of Phillip Tetlock, author of the renowned Expert Political Judgement. Over a 20 year period, Tetlock gathered tens of thousands of predictions from hundreds of experts. They held many traits in common among, details of those traits can be found in this interview, but only those that used formal models outperformed random.
Once properly constructed, use cases of models can be implemented outside the field of original intent. Markov processes, dynamic processes frequently used to model the spreading of diseases, can be useful in deciphering the author of a literary piece without knowing the name – a warning to those who are using anonymity as a cloak for their writing in the age drowning liberties.
The successful implementation of models is exemplified by Bruce Bueno de Mesquita, a Stanford and NYU professor renowned for his accurate predictions. While a drunk uses a light pole for support, intelligtent practitioners use models for illumination, augmenting conclusions with common sense and experience.
Many complementary models are better than models used in isolation. The example given by Page are the linear model used to forecast that home prices would continue higher, while others like Kyle Bass used models showing the large disparity between income growth and house prices, predicting they would fall.
Popularity: 2% [?]