|
Taleb is a smart guy. I like the idea of questioning the use of the normal curve, and I'm pretty sympathetic to the view that the US election was a 50/50 event. But...
Criticising Nate Silver's grasp of statistics is like shooting fish in a barrel. The first problem was the guy was trying to apply probabilities to events in the distant future, like an election 5 months away, with no idea what might happen in-between 'now' and 'then'. Second, the guy had probabilities that would change almost daily. Anyone trying to do 1) is dumb. But when they start doing 2) at the same time they are a moron.
He gives the rise of the Internet, the personal computer, World War I, dissolution of the Soviet Union, and the September 2001 attacks as examples of black swan events
I don't see how all of these events were extremely improbable. WW1 was a long time coming, it was unlikely to have started the way it did, but history said it was going to happen eventually. The dissolution of the USSR also doesn't strike me as an extreme outlier. Don't know about the other ones. More importantly, I don't know how he can apply probabilities to these events post hoc, and expect people to accept he understands their causes well enough to attach a probability to each one. These are complex events with numerous potential causes, not all of which are understood.
Generally though, if there are enough events some of them will be outliers. And some of those outliers will be important events. Nothing new there.
|