Taleb Seminar

20:11 Sun 10 Feb 2008
[, , , , , ]

Last Monday I went to a Long Now Foundation seminar by Nassim Nicholas Taleb, author of Fooled By Randomness and The Black Swan—both books I would recommend to just about everyone. The title of the talk was “The Future Has Always Been Crazier Than We Thought”, and while Taleb did talk about our historic inability to predict what was going to happen in the future, I didn’t feel that ‘future craziness’ was actually a major theme. (If you change “Crazier” to “More Unpredictable” you get a more accurate title.)

Instead, he mostly reprised the themes he covered in The Black Swan, focusing particularly on the notion of a “black swan event”, on the realms of Extremistan and Mediocristan, and on the uselessness of certain categories of ‘experts’, particularly economists and market commentators.

A ‘black swan event’ is an event that breaks one’s theoretical model for understanding some domain (and thus predicting events in that domain) in some significant way. By definition the event was unpredictable using the theoretical model. Not all black swans are catastrophic (Taleb cites the rise of the Internet, for example, as a black swan), but the catastrophic ones are probably best for getting his point across. An example he uses is that of the turkey, whose theoretical model of existence is that the butcher cares deeply about its welfare. All of the turkey’s available evidence supports this model, until the day when the butcher takes it out and kills it. Taleb says that he’s trying to figure out how not to be the turkey; he also notes that the fateful day is a black swan event for the turkey but not for the butcher.

Taleb proposes a radical empiricism, a severe conservatism when it comes to certain risks, and a highly risky approach in certain areas (he didn’t talk much about those areas, which I was curious about). He suggests that we pay great attention to experience and evidence but do not use it to construct models that we think will be useful in terms of prediction, and to always regard any theoretical models as highly suspect.

To illustrate what he means by Extremistan and Mediocristan, he uses two thought experiments. First, randomly take a thousand people from around the world and add their heights together. Then take the tallest person in the world and express their height as a percentage of the total height of the thousand. At most you’re looking at about 0.3%.

Second, randomly take a thousand people from around the work and add their wealth together. Then take the richest person in the world and express their wealth as a percentage of the total height of the thousand… Here you’re looking at a rather more significant percentage, with the odds of it being several orders of magnitude larger than the total.

The first is Mediocristan, the second Extremistan. In the first, statistical models like the bell curve are valuable. In the second, they’re simply not, because a single variable can have a colossal impact on whatever you’re measuring.

Taleb argues, convincingly, that much of our world is actually in Extremistan, but we fool ourselves by thinking it’s in Mediocristan. He considers this particularly true for economics, and heaps ridicule upon them for their claims to understand (and hence predict) what’s going on in the markets and the world.

He stated that he felt that the economists are just as divorced from reality as the bishops of any major religion, but that the economists are more dangerous because they’re claiming scientific understanding. He also made a passing reference to the ‘new atheism’, saying that if you’re going to criticize people for believing in religion because it’s irrational, you also have to go the whole way and criticize them for paying any attention whatsoever to market analysts.

A rule of thumb he says that he uses is that he never takes predictive advice from anyone in a suit and tie.

It was a highly enjoyable talk, and Taleb is an entertaining speaker. One final note: apparently wine under the ‘Black Swan’ label isn’t very good, and you should never, ever, get some for Taleb—he has quite enough already, because humans aren’t good at predicting that many of them will see the same connection as each other.

« (previous)
(next) »

Leave a Reply