Black Swan is a sort of sequel Taleb’s earlier book, Fooled by Randomness, and it Taleb’s more thorough explanation of his thoughts on the nature of randomness and the implications of how to deal with that. He first discusses two fictional nations, Mediocristan and Extremistan. In Mediocristan events are Gaussian, which is to say that most things cluster about the mean, with rare outliers. Weight is from Mediocristan; the heaviest people are not going to be substantially heavier than the mean. Another way of saying this would be that removing one sample from the pool does not meaningfully change the result; even if you remove the heaviest person from a very large group of people, the mean weight does not change.

However, in Extremistan, events are not Gaussian. One rare event may dwarf all other events. Book authors are from Extremistan: the best-selling author sells way more books than the second best-selling author, and a handful of authors account for most of the sales. In Extremistan, removing one sample can potentially change mean very drastically. In an extreme example, he considers a turkey. Every day of the year the family brings him food and he is happy and well fed. But on Thanksgiving Day morning, there is no food and he is killed and eaten for dinner. Whether the turkey includes Thanksgiving Day in his sample drastically alters his outcome.

It turns out that many events that we think are from Mediocristan are really from Extremistan. Events are Mediocristan are defined by a certain “gravity:" a force that acts to limit deviations from the norm; there are simply serious practial factors that limit how heavy a person can realistically get (access to food, money for food, the weight-bearing capacity of our muscular-skeleto system, etc.). Events in Extremistan do not have this gravity. Individual book buyers, for instance, are free to all buy the same book if they wish; in fact, many do. As it turns out, instead of a Gaussian distribution clustered about a mean, the distribution in Extremistan is often more like a power law: the most frequent events drop off quickly and sharply, leaving a long tail of fairly unlikely events. Effectively, the popular events are wildly popular, and all the rest are pretty much unknown.

There are several practical consequences to this. First, it seems that we as humans tend to assume that randomness takes a Gaussian shape, that things revert to their mean. For events in Extremistan we must not assume this; we must assume that there may be an unlikely event that, if it happens, dominates everything else. If you are a book author, that unlikely event is becoming a best-seller, so do not count on being the rare event. Assume that you will be the normal, low-selling long tail. If you are a book publisher and you have a best-seller, assume that the next will not be.

The second practical consequence is to limit your exposure to unlikely events that have a serious negative downside and expose yourself to unlikely events that have are large upside. In the late 1990s a hedge fund called Long-Term Capital Management assumed that price differences were Gaussian (because the math is tractable that way; there are no formulae that work in Extremistan). One day the price difference were not Gaussian for long enough that LTCM lost more money than it had made in it’s history. It went bankrupt and nearly took the U.S. financial system with it. They were exposed to a negative event that was unlikely, but catastropic. Some unlikely events are good, though, and it may be worth having a steady, but small, drain if every so often one of those exposures gives you a huge gain.

The third consequence is that we cannot necessarily take credit for our successes (or give credit to others). We may have simply been riding the coat-tails of a rare event that catapaulted us to greatness. The successful may not be successful because of what they did (although their diligence, discipline, hard work, etc. enabled them to capitalize on the opportunity); much of success is simply luck.

Taleb does not give any good method of determining if a particular process is in Extremistan. However, it is known that a power law distribution occurs whenever individual actors can choose what to do. The sizes of cities, for example, follow a power law. People can choose which cities to live in, and people often choose to live in a city because of what other people have already chosen. So if choice is involved, the process is likely from Extremistan. On the other hand, if natural processes are involved, there is some “gravity” and it is likely to be from Mediocristan.

Black Swan is an interesting a readable discussion of probability and the mistakes we often make in dealing with randomness. However, it has a number of serious flaws. The foremost is that the book is steeped with Taleb’s intellectual arrogance. Part of that makes for fun reading, as he skewers pretty much every well-known school of thought. Equally strong, however, is the belief that skeptical rationalism is the one true religion, that anyone who looks at the world otherwise is full of themselves. Taleb makes sure the reader understands that he only hangs out with the enlightened intellectuals, not the deluded wannabes. Ironically, this leads to a sense that it is Taleb who is full of himself and prone to be ignoring something big himself. The ideas are useful to incorporate, but perhaps choose someone else as a role model: it seems awful lonely up at the pinnacle of intellectual thought.

Ultimately, Black Swan comes down to several fairly simple ideas (even though Taleb cautions us to not attempt to pull out a cohesive, sure-to-be-oversimplified idea). First, the unlikely, unpredictable events can dominate our fortunes. Second, Extremistan is more frequent than we think. When distilled down to this, the book seems rather obvious, yet while in some sense we intuitively know this, we still live our lives assuming things will revert to normalcy. Taleb’s work is a very readable discussion showing that we need to get over that idea.