So we now have two models that are more pessimistic than 538 and I think the pessimists are a bit more correct.
Still, Biden is in a better place than Trump was in 2016 and there are some reasons for optimism.
I hadn't noticed that NV and AZ are pretty close and I think in AZ there are good reasons to believe he'll gain ground.
GA I think you still need polls to tighten, otherwise you're relying on a freakish big error.
Right, but those states are reasonably tightly correlated and at the moment he's behind in PA even in 538.
You need Biden to develop 1-2 more "out cards".
I will say that I think one of the weaknesses in Silver's modeling is that it weights tail outcomes (Biden winning MI/NV/WI but losing VA or MN, etc) as too likely.
The counter argument is that no one tail outcome is likely, but unlike House races we just don't see tail outcomes in the presidential
Honestly given how unprecedented this election is the entire modeling exercise is pretty intellectually dishonest. The signal just isn't there, the most honest thing you can do is point at the close polls and the different indicators going different ways and say it's about 50-50
Yeap. Some signal might emerge nearer to, and if Trump shits his pants at the debate he could lose.
I'll be fair to Trump: He probably won't shit in his drawers.
Probably.
I think the closer we get to the election the more you can just go to the polls, and those might still be wrong but it's basically straightforward. But right now they're using all kinds of historical stuff that e.g. doesn't know the challenger is a former incumbent, doesn't know about Roe, etc
If you want to trade on your pundit reputation, you are going to say "Trump's probably winning" no matter the the data says. If he loses, fine, can you blame (Pundit) after 2016? If he wins, you were the contrarian who spoke truth to power.
Except of course, if everyone's a contrarian...