[fabian2k]

The 70:30 prediction against Trump was far better than most. I did see models back then that considered the state polls mostly or entirely uncorrelated, and those produced obviously garbage with 90% or even 99% in favor of Clinton.

But in the end people pick on Nate because he really enjoys being an asshole on the internet. It's far more about when he acts as a pundit, not as an expert on statistics.

[softwaredoug]

People consistently have a hard time understanding that 30% probabilities happen all the time.

[triceratops]

Surely not all the time.

[BobaFloutist]

I think given the number of things that can happen with ~30% probability, there's probably something significant happening with ~30% probability at basically all times.

[triceratops]

I was making a joke but yes that's a good point.

[AnimalMuppet]

Well, we're talking about elections. You have an election where there's a president, 30 or so governors, 33-34 senators, and 438 representatives. Say a total of 64 major offices, or 500 if you count the representatives. You'd expect a 30% chance to happen in 19 major races, or 150 races if you count the representatives.

So in an election, that happens all the time. It just doesn't always happen in the race for president.

[Lerc]

30% of the time it is all of the time.

[Yossarrian22]

Some say 30% of the time.

[krapp]

Where Presidential politics is concerned, I think it's less a case of misunderstanding probabilities and more the success of party propaganda. Every victory is a landslide with a resounding mandate from the populace, every defeat a crushing humiliation and repudiation of your opponent's Unamerican ideals.

[add-sub-mul-div]

Those predictions all became worthless anyway when Comey reopened the "emails" issue right before the election and threw fresh meat to all the stupid people who ate that up.

[bigfishrunning]

I kind of fell off the Nate Silver train toward the end of Trump's first term (so deep in the COVID-19 era...). It feels like around that time 538 shifted heavily away from raw statistics and into punditry, and they seemed less unique among the various political blogs.

[lesuorac]

538 was sold in 2018 with Trump's first term ending early 2021.