Extremistan

From Tom Nutall's review:

The black swan was the example used by John Stuart Mill to illustrate the problem of induction in the philosophy of science, which states that no amount of empirical evidence for a proposition can ever prove conclusively that it is true.

For hundreds of years, zoologists assumed that all swans were white, because they had never seen any evidence to the contrary. But then with the discovery of Australia came the sighting of a black swan. Nassim Nicholas Taleb, in his book of the same name, has a more specific definition—for him, a “black swan” is an event that meets three conditions: a) it was unpredictable b) it had a big impact and c) after the event, we tried to explain it and make it appear more predictable than it actually was.

I first came upon Taleb via the Long Now Foundation, whose seminars + my reading of Iain Banks is opening up my mind very wide after a long, eventually boring cycle. On that note, when I think about it, I have really only encountered 2 very big ideas in the past decade, the first was eCommerce and the second was Hayekian Conservatism. One of those ideas was new, the other fairly old. And so now I am looking for a couple more big ideas and I have begun to stake out several dozen individuals in whose work I think I can find them. The Long Now is an excellent source of futurist thinking and so is Bank's science fiction. Along the way to these big ideas are a number of intellectual tools, and I expect that I may be able to use these tools in my career in Business Intelligence. The conceptual tool of Mediocristan and Extermistan is one such tool. If you'd have asked me a month ago about the brilliance of markets, I might have pointed you to the <strike>Black Shoals</strike> Black-Scholes (thanks nick) model in general, and the book My Life as a Quant in particular. But Taleb said that Black-Scholes is a joke, and he should know. He worked many years on Wall Street. While I have noticed a significant number of people sniffing at the practices of hedge funds, not least Mr. Ron Paul, in the past several months - often dripping with populist rhetoric, none has convinced me that what they do is fundamentally wrong. But as Taleb describes the phenomenon of the unexpected, I now see the light. Knowing whether or not one is in the domain of Mediocristan or Extremistan will be very helpful in assessing the ultimate utility of Business Intelligence tools. If you apply them to domains in which the extraordinary is probable, you may find them of little use. How one might go about determining the nature of the domain is what I need to chase down.

Taleb describes Mediocristan in terms of human height and the applicability of bell curve-like statistics. If the average human is about 5 foot 9, you can calculate with some assurance that you will see a 6 foot tall human. You might even be able to predict with some accuracy that if one million humans walk through your door, you'll probably see some seven footers. The chance that you'll see an eight footer, maybe one in a billion. But the chance that you'll see someone twice the height of the average Joe? Probably not even human. Human height is in Mediocristan. Human wealth, on the other hand is in Extremistan. If the average wealth of an individual measured in income per year. You might figure that somebody with an order of magnitude more comes in at some fraction and two magnitudes at a very minute fraction. But thinking about wealth, you might have a million people walking through the door and have somebody 1000 times wealthier the average. Extremistan.

Comments

[Charles Follymacher]

You're considering the wrong bell curve. Black swans, income and height are all extremistan if you look at it in terms of frequency of occurance. Best you can say is that income is two-dimensionally extremistan (frequency, distance from mean).

[Cobb]

I forgot to mention 'fidelity to type' in conveying the idea. What lets you know that you are in Extremistan is when the completely unexpected occurrence is of the same type. If something 20 foot tall walks through your door, you are not likely at all to call it human because it would shatter your assumptions of what humans can possibly be. You set the parameters for your bell curve not only on expected frequency but in some ways you consider the long tail's limits by category. Extremistan is an unexpected category breaker.

Now suddenly I am reminded of Gladwell's new book...

[Charles Follymacher]

Right, I was going to say something about expectations vs average, too. And when we feel confident about the range of the bell curve (we've seen a whole lotta swans and all of em are white), we can only bust out of mediocristan if we come across something wholly unexpected, right? As in, f'zample, an order of magnitude richer than Bill Gates or Shell Oil. Or am I still not getting this?

[Cobb]

Well the brilliant thing would be to be able to tell whether or not you are in Extremistan a priori. And perhaps that's what Taleb is telling us - the global financial market is absolutely there.

Two other mnemonics: 1. When Manhattan project scientists worried about detonating the first nuke, they thought that the atmosphere might ignite - perhaps they thought they were in Extremistan, but it turned out that the were in Mediocrestan. Their calcs were an order of magnitude off, and were possibly asymptotic to the disaster line.

2. Douglas Adams' Infinite Improbability Drive renders Extremistan cognizable, even perfectly navigable. But then that's impossible. OK extremely improbable.

http://en.wikipedia.org/wiki/Infinite_Improbability_Drive

[Charles Follymacher]

Heh. Good stuff. It's been an age since I read Hitchhiker's Guide.