Do y'all remember that princeton election guy who forecast a win for Clinton with 99% in 2016? I just found a post-mortem interview with him where he says:
Probabilities are not a good way to convey uncertainty. The first reason being that it’s hard to estimate the true amount of uncertainty, and I discovered that.
I... have never before felt such a strong urge to say "skill issue". But people weren't saying "skill issue" yet in 2016, so nobody got the chance. Time is cruel.
like this
Huey
in reply to Ben Weinstein-Raun • •Sam Wang and Nate Silver had an interesting back-and-forth, if I remember correctly. A friend of mine was convinced that Sam was right and that Nate was “putting his thumb on the scale to cover himself.”
In hindsight, it seems to me that Sam’s approach was more of a straightforward averaging of polls, while Nate’s method is more like a gambler’s—integrating his own beliefs into the model.
Ben Weinstein-Raun likes this.