Taleb corrects Pinker

He observes that Pinker has been fooled by randomness and The “Long Peace” is a Statistical Illusion:

When I finished writing The Black Swan, in 2006, I was confronted with ideas of “great moderation”, by people who did not realize that the process was getting fatter and fatter tails (from operational and financial, leverage, complexity, interdependence, etc.), meaning fewer but deeper departures from the mean. The fact that nuclear bombs explode less often that regular shells does not make them safer. Needless to say that with the arrival of the events of 2008, I did not have to explain myself too much. Nevertheless people in economics are still using the methods that led to the “great moderation” narrative, and Bernanke, the protagonist of the theory, had his mandate renewed.

Now to my horror I saw an identical theory of great moderation produced by Steven Pinker with the same naive statistically derived discussions (>700 pages of them!).

  1. I agree that diabetes is a bigger risk than murder –we are victims of sensationalism. But our suckerdom for overblown narratives of violence does not imply that the risks of large scale violent shocks have declined. (The same as in economics, people’s mapping of risks are out of sync and they underestimate large deviations). We are just bad at evaluating risks. 
  2. Pinker conflates nonscalable Mediocristan (death from encounters with simple weapons) with scalable Extremistan (death from heavy shells and nuclear weapons). The two have markedly distinct statistical properties. Yet he uses statistics of one to make inferences about the other. And the book does not realize the core difference between scalable/nonscalable (although he tried to define powerlaws). He claims that crime has dropped, which does not mean anything concerning casualties from violent conflict.
  3. Another way to see the conflation, Pinker works with a times series process without dealing with the notion of temporal homogeneity. Ancestral man had no nuclear weapons, so it is downright foolish to assume the statistics of conflicts in the 14th century can apply to the 21st. A mean person with a stick is categorically different from a mean person with a nuclear weapon, so the emphasis should be on the weapon and not exclusively on the psychological makup of the person.
  4. The statistical discussions are disturbingly amateurish, which would not be a problem except that the point of his book is statistical. Pinker misdefines fat tails by talking about probability not contribution of rare events to the higher moments; he somehow himself accepts powerlaws, with low exponents, but he does not connect the dots that, if true, statistics can allow no claim about the mean of the process. Further, he assumes that data reveals its properties without inferential errors. He talks about the process switching from 80/20 to 80/02, when the first has a tail exponent of 1.16, and the other 1.06, meaning they are statistically indistinguishable. (Errors in computation of tail exponents are at least .6, so this discussion is noise, and as shown in [1], [2], it is lower than 1. (It is an error to talk 80/20 and derive the statistics of cumulative contributions from samples rather than fit exponents; an 80/20 style statement is interpolative from the existing sample, hence biased to clip the tail, while exponents extrapolate.)
  5. He completely misses the survivorship biases (which I called the Casanova effect) that make an observation by an observer whose survival depends on the observation invalid probabilistically, or to the least, biased favorably. Had a nuclear event taken place Signor Pinker would not have been able to write the book.
  6. He calls John Gray’s critique “anecdotal”, yet it is more powerful statistically (argument of via negativa) than his >700 pages of pseudostats.
  7. Psychologically, he complains about the lurid leading people to make inferences about the state of the system, yet he uses lurid arguments to make his point.
  8. You can look at the data he presents and actually see a rise in war effects, comparing pre-1914 to post 1914.
  9. Recursing a Bit (Point added Nov 8): Had a book proclaiming The Long Peace been published in 1913-1934 it would carry similar arguments to those in Pinker’s book.

Taleb is using a different means to reach much the same conclusions I have. Again. Pinker is essentially applying the same “This Time It’s Different” argument to violence that the mainstream economists applied to the dot com bubble, the housing boom, and the post-2008 “recovery”.

Simplistic thinkers inevitably think in linear terms. They assume tomorrow will be like today because today was pretty much like yesterday. Both those who know history and those who understand probability understand that at some point in time, this will no longer be the case.

History is rife with long periods of peace and tranquility. Those are quite often the sections missing from the history books, because there was nothing much that was noteworthy to record. But human nature being what it is, sooner or later events always becoming more exciting, which usually means more bloody.

It’s not hard to understand why there are fewer wars these days. Nuclear weapons have put an end to the post-French Revolutionary progress towards Ludendorffian total war. But that doesn’t mean they will never be used or that Man will not find other means to fight cataclysmic wars. It’s rather remarkable that anyone would make such abysmally stupid claims about the prospects for the continuation of the “Long Peace” when the USA is moving rapidly towards ethnic civil war, Europe is preparing for extreme ethnic cleansing, and the Dar al-Islam is in the process of uniting under a new and aggressive Caliphate even as the USA attempts to instigate war with Russia.