Black Swan

This is a quite controversial book. Overall I like the book, and most likely I will read it for a second time with a highlighter. I guess most the criticism comes from the author's relentless attack on mainstream economists and statisticians.

The author makes a lot of (good) points about how we deal with uncertainties, synthesizing them into one big theme -- Black Swan, the rare and highly consequential events that seem more likely ex post. 9/11, stock market crash, and best sellers are all examples of Black Swan.

I will try to summarize his points from different perspectives. Most of them are not that original, as the matter of fact I sort of knew a lot them before reading the book, scattered around in different fields. But this in no means diminishes the author's achievement in this book by putting them together in a bold, provocative, sometime quite emotional way.

Philosophically the author makes a few points in epistemology. We tend to place more weight on what we know than we don't know. It's often the part that we don't know we don't know that will come back to haunt us. Also we to tend theorize more than necessary, we overstate the causal links that might not be there. The author goes a great length to attack "platonified" knowledge, the overcatogerized, ossified, dogmatized theories.

Psychologically we fall into a lot fallacies: the survival bias, the conformation bias, the narrative fallacy. As the result the world appears more orderly to us than it is.

For the technical point of view, we are trapped(fooled) by the dominant notion of normal distribution, the bell-curve, or broadly speaking the Gaussian family of distribution. But in reality a lot of random variables in so-called extremistan don't not follow Gaussian at all, instead power law distribution is more appropriate. So by blindly sticking to the bell-curve, we vastly underestimate the odds of rare event since it's probability drops so quickly once it's beyond a few standard deviations.