538 has a cool interactive tool similar to 270towin.com, where you can award certain states to Trump or Biden and watch the model recompute the election odds in real time.
The advantage of the 538 version is how it takes into account interstate correlations (based on past and recent data). So, if you (say) set it to assume that North Carolina will go for Biden, it will revise the map so that (say) South Carolina will be slightly more likely to do so than it would have otherwise.
Yeah. If you flip Montana or New Hampshire, the election results change a lot, not because MT or NH has a lot of electoral votes (they don’t), but because they indicate a big swing in their correlated states.
Cool.
I plan to use the tool for an exercise in the Political Geography course I happen to be teaching right now.
The electoral college story which hasn’t really changed for a while is that Biden has around a 5 point margin in the states that he absolutely needs like PA and WI but only 2-3 points in bonus states like Florida and Arizona.
That is comfortable but 5 points was around the polling error in states like PA last time so high confidence in a Biden win assumes that state-level polling has improved since 2016 which I think is plausible.
We will have to see what the debate brings but my guess is it’s going to be another ineffective Trump performance which together with an ad barrage could cost him a point or so in the final two weeks which would give Biden that extra margin of comfort.
From Nate Silver:
Wow. That’s huge.
Politico has a run-down of the swing state polling picture, with 2016 final polling and results for comparison. Biden leads in 11 of the 13, and is ahead by at least six points in MI, MN, NV, NH, PA and WI. Those alone are enough to get him to 290 EVs.
Actually, those alone would only get him to 66 EVs. 
Ah, this addresses the question I asked in another thread: whether early voting and mail in numbers right now show actual increased engagement or front-loading of the same voters as 2016, and how we could tell.
The “slight tightening” Nate Silver predicted as likely is now happening. For the first time, Biden’s chances have definitively declined in the 538 model (just a percent or two, but it’s real).
This seems mainly due to Ohio, which was polling at dead even but is now leaning a point or two toward Trump, but it’s also influenced by North Carolina-Arizona-Florida (which had been polling around 2-4 points for Biden, and are now more often in the 1-3 range) —- and most worryingly, Pennsylvania and Wisconsin, which had been polling around 8 points toward Biden (sometimes a little less), but now typically are at just 5 points (sometimes a little more).
The New York Times is (cautiously) bullish on Pennsylvania based on their recent poll, though, so that’s good.
I am not seeing any real tightening on 538. Trump’s win percentage has been around 13% since Oct 14. And Biden’s percentage saw small dips many times during June to August.
Anyway we should be getting the first of the final state polls from each pollster in a few days which will give us a clear picture of the final stretch. I don’t forsee any big changes though it’s possible that Biden’s money advantage and the Bloomberg Florida push might boost him by a point or so in the key states. A string of endorsements probably won’t hurt either.
I see very slight tightening. PA is at +5.7 (down from 7.2) on average, and WI is at +6.6 (peak was +7.9)
FL is + 2.4 (recent peak of +4.5)
Overall 538 is still at 87%
Brian
Right, down a tad from a brief flirt with 88%, six days ago. A flattening, if you will. A plateau.
The saving grace at the moment is Iowa, where the latest high-quality NY Times-Siena poll has Biden ahead 3%.
Correction: the bullish-for-Biden Pennsylvania poll cited by the NY Times’ Nate Cohn was actually conducted by Muhlenberg College, but apparently it’s high-quality.
It’s really not tightening at all, it’s showing a return to Biden’s rock-steady lead that he’s had all along. We’re only seeing Trump’s post-debate and post-Covid-diagnosis dip disappear.
I’ve been tracking the swing states’ polls on 538 since late August.
September 6, BIden’s lead:
AZ: +4.7
FL: +2.8
MI: +6.7
NC: +1.8
PA: +4.5
WI: +7.1
Avg: +4.75
Then there were 4 weeks of similar polls for each swing state. Biden’s swing state average for those 4 weeks was +4.63.
Next, we had the shitshow debate and Trump’s COVID announcement. The next two weeks saw a spike in Biden’s lead. His swing state average for the middle two weeks of October went up to +5.9.
And now today, Biden’s lead:
AZ: +2.6
FL: +2.6
MI: +7.0
NC: +2.6
PA: +5.7
WI: +6.6
Avg: +4.76
Biden’s actually doing better in PA now than he was two months ago. Granted, it’s still a little too close for comfort, but as of now, Biden has not lost support, and it looks like Trump hasn’t gained any. ETA: Biden’s also higher than 50% in PA right now, and in late-August/early-September, he was below. So that’s good too. PA is still “safe Biden.”
Today, Mark Meadows officially threw in the flag and surrendered the United States in the fight in World War COVID:
Of course, even if you have mitigations, if the hospitals are overloaded they won’t be able to give them to people. I hope this swings FL to Biden, maybe the rest of the elderly who were voting for Trump will realize how he’s all for letting them die.