2005 : WHAT DO YOU BELIEVE IS TRUE EVEN THOUGH YOU CANNOT PROVE IT? [1]

nassim_nicholas_taleb's picture [5]
Distinguished Professor of Risk Engineering, New York University School of Engineering ; Author, Incerto (Antifragile, The Black Swan...)
Mathematical trader; Author, Fooled By Randomness

We are good at fitting explanations to the past, all the while living in the illusion of understanding the dynamics of history.

My claim is about the severe overestimation of knowledge in what I call the " ex post" historical disciplines, meaning almost all of social science (economics, sociology, political science) and the humanities, everything that depends on the non-experimental analysis of past data. I am convinced that these disciplines do not provide much understanding of the world or even their own subject matter; they mostly fit a nice sounding narrative that caters to our desire (even need) to have a story. The implications are quite against conventional wisdom. You do not gain much by reading the newspapers, history books, analyses and economic reports; all you get is misplaced confidence about what you know. The difference between a cab driver and a history professor is only cosmetic as the latter can express himself in a better way.

There is convincing but only partial empirical evidence of this effect. The evidence can only be seen in the disciplines that offer both quantitative data and quantitative predictions by the experts, such as economics. Economics and finance are an empiricist's dream as we have a goldmine of data for such testing. In addition there are plenty of "experts", many of whom make more than a million a year, who provide forecasts and publish them for the benefits of their clients. Just check their forecasts against what happens after. Their projections fare hardly better than random, meaning that their "stories" are convincing, beautiful to listen to, but do not seem to help you more than listening to, say, a Chicago cab driver. This extends to inflation, growth, interest rates, balance of payment, etc. (While someone may argue that their forecasts might impact these variables, the mechanism of "self-canceling prophecy" can be taken into account). Now consider that we depend on these people for governmental economic policy!

This implies that whether or not you read the newspapers will not make the slightest difference to your understanding of what can happen in the economy or the markets. Impressive tests on the effect of the news on prices were done by the financial empiricist Victor Niederhoffer in the 60s and repeated throughout with the same results.

If you look closely at the data to check the reasons of this inability to see things coming, you will find that these people tend to guess the regular events (though quite poorly); but they miss on the large deviations, these " unusual" events that carry large impacts. These outliers have a disproportionately large contribution to the total effect.

Now I am convinced, yet cannot prove it quantitatively, that such overestimation can be generalized to anything where people give you a narrative-style story from past information, without experimentation. The difference is that the economists got caught because we have data (and techniques to check the quality of their knowledge) and historians, news analysts, biographers, and "pundits" can hide a little longer. Basically historians might get a small trend here and there, but they did miss on the big events of the past centuries and, I am convinced, will not see much coming in the future. It was said: "the wise see things coming". To me the wise persons are the ones who know that they can't see things coming.