The Great Undoing
Dear Mr. Lewis:
It is impossible to discuss baseball without mentioning Billy Beane (a former Twin) as described in your book Moneyball (a less memorable movie). Many of us consider The Big Short to be the single best description of the collateralized debt obligation bubble that contributed to the financial crisis of 2007-2008 (a more memorable movie). But it is an act of cosmic coincidence that the undoing of our political order and the publication of your book, The Undoing Project, occurred practically in tandem.
The book starts out on rather familiar terrain, in this case, the application of new statistical analysis to the acquisition of NBA players by the Houston Rockets. The Undoing Project is the story of Amos Tversky and Daniel Kahneman, two Israeli armed forces psychologists who pointed out that even in as competitive a market as professional basketball, with easily quantifiable results, systematic errors continued to be made because player evaluators were not aware of their own mental processes.
The key finding of their work is that humans are, in deeply structural ways, irrational in aspects of decision making, and consequently the favored economic model of the “rational man” fails. Furthermore, in study after study, when it comes to making decisions, humans are predisposed to irrationality. David Leonhardt, op-ed columnist for the New York Times, referred to that research as some of the most influential social science of the past century.
The so-called “gambler’s fallacy” is well known—that if you flip a coin that comes up heads three times in a row, many people believe that there is a greater chance the next flip of the coin will also be heads when in fact, the probability remains 50/50. The same error of judgment in a somewhat more complex situation affects experts in a variety of disciplines. Kahneman and Tversky went on to show doctors misdiagnosing breast cancer because the pool of samples (young women) was too small to be predictive—a basic Bayesian error. This power of a small amount of evidence convinces experts in many fields of the soundness of their judgment, which reinforces preconceived views (confirmation bias).
Further research with clinical psychologists and psychiatrists found that those with the least training (grad students) were just as accurate in predicting whether a psychiatric patient was at risk of committing suicide as more experienced observers. As Dr. Louis Goldberg stated in an interview, “Accuracy on this task was not associated with the amount of professional experience of the judge.”
Kahneman and Tversky are quoted as saying “no one ever made a decision because of a number . . . They need a story.” Remember here, we are talking about some of the most sophisticated experts in statistics, economics and psychology. Or for that matter, general managers of basketball and baseball teams. And maybe political professionals.
In late May, I was at a small dinner with a very highly placed member of the Clinton campaign. In response to a question about Secretary Clinton’s chances of winning, the expert responded that if Clinton won Florida, Trump had no chance of winning the Electoral College. And, we were assured, the Clinton campaign would win in the state of Florida. This assurance was repeated in late July by the same person in a telephone conference, again in mid-October and was again confirmed by many of the polling experts a day before the election. So too was the expert analysis of Secretary Clinton’s electoral victory-to-be in the states of Pennsylvania, Michigan and Wisconsin. The top national polling experts, including Nate Silver, all predicted the same right up to the day of the election.
All of these experts were dramatically wrong—just like the experts discussed by Michael Lewis in The Undoing Project: the Israeli flight instructors, the medical diagnosticians, the NBA general managers, and various groups of psychology and mathematics students. The insight of this book is that experts are often wrong because of the innate workings of their own minds and their lack of self-awareness of those idiosyncrasies.
Secretary Clinton won the popular vote by approximately 3 million votes, but lost the Electoral College. The media is full of reports of field politicos being told by Clinton HQ that the model did not allow last-minute resources to be put into states like Michigan and Wisconsin. Were the experts using the wrong model? Did the experts rely on too small a sample in predicting that future electoral results would look exactly like immediate past electoral results (Bayesian error)? Did early polling data reinforce stereotypes held by Clinton HQ experts (confirmation bias)? Did the Clinton campaign believe that the infrastructure it had constructed was superior to the electoral infrastructure constructed by others in the states of Michigan and Wisconsin (endowment effect)? These are all questions that will come to any politico who reads this book.
What are the answers? Clearly we need a new science of practical behavioral politics, similar to the existing science of behavioral economics.
We should all read Michael Lewis’ book, not least those of us who occasionally claim to be experts. In an urgent plea to Lewis, with a faint echo of Theodore H. White, the next book should be The Undoing of the Presidency 2016.
Vance K. Opperman Occasional Expert
Vance K. Opperman (email@example.com) is owner and CEO of MSP Communications, which publishes Twin Cities Business.