Ooh. I recently read The Black Swan, by Nassim Nicholas Taleb (originally published in 2007); I read the 2nd ed. (2010).
Thesis: the world tends to be defined by high-impact events which are very hard to predict. They happen a lot more than people give them credit for.
A few examples are trends in finance, bestsellers in publishing, and the impact of the internet. Taleb observes that people tend to have ridiculous hindsight (and other) biases with respect to these sorts of events.
The only real recommendation in the book is to categorize situations you’re trying to predict into four “quadrants”: binary outcomes vs. outcomes with magnitude, and normally-distributed probabilities vs. non-normally-distributed probabilities. Taleb terms the normally-distributed region “Mediocristan” and the non-normally-distributed region “Extremistan”. When you have non-normal distribution, and outcomes are based on magnitude, that is the “fourth quadrant”—where you have to beware of black swans, and know that you’re going to have trouble making good predictions.
He talks about the probabilistic model in Mediocristan, where the Gaussian / normal distribution applies—it tails off exponentially, rapidly reducing the probability of outliers to miniscule, so it’s much easier to rely on the absence of outliers. And he gives dire warnings against applying the normal distribution to the Fourth Quadrant, though many do anyway—for example, the Black-Scholes option valuation equation models the price of stocks as varying using a Gaussian distribution, and (in the 2nd ed.) makes the case that this was the reason for many derivatives-based financial firms collapsing in 2008.
It’s a fun read for students of rationality because he talks a lot about empiricism and biases. One idea he emphasized is the “ludic fallacy”—“ludic” meaning “game-like”, and referring to people relying on excessively simplified models to make predictions, or thinking too much inside the box. Wikipedia has a better summary of this idea. It’s something I don’t see talked about much here, though I think it applies to rationality.
Ooh. I recently read The Black Swan, by Nassim Nicholas Taleb (originally published in 2007); I read the 2nd ed. (2010).
Thesis: the world tends to be defined by high-impact events which are very hard to predict. They happen a lot more than people give them credit for.
A few examples are trends in finance, bestsellers in publishing, and the impact of the internet. Taleb observes that people tend to have ridiculous hindsight (and other) biases with respect to these sorts of events.
The only real recommendation in the book is to categorize situations you’re trying to predict into four “quadrants”: binary outcomes vs. outcomes with magnitude, and normally-distributed probabilities vs. non-normally-distributed probabilities. Taleb terms the normally-distributed region “Mediocristan” and the non-normally-distributed region “Extremistan”. When you have non-normal distribution, and outcomes are based on magnitude, that is the “fourth quadrant”—where you have to beware of black swans, and know that you’re going to have trouble making good predictions.
He talks about the probabilistic model in Mediocristan, where the Gaussian / normal distribution applies—it tails off exponentially, rapidly reducing the probability of outliers to miniscule, so it’s much easier to rely on the absence of outliers. And he gives dire warnings against applying the normal distribution to the Fourth Quadrant, though many do anyway—for example, the Black-Scholes option valuation equation models the price of stocks as varying using a Gaussian distribution, and (in the 2nd ed.) makes the case that this was the reason for many derivatives-based financial firms collapsing in 2008.
It’s a fun read for students of rationality because he talks a lot about empiricism and biases. One idea he emphasized is the “ludic fallacy”—“ludic” meaning “game-like”, and referring to people relying on excessively simplified models to make predictions, or thinking too much inside the box. Wikipedia has a better summary of this idea. It’s something I don’t see talked about much here, though I think it applies to rationality.