What I think is interesting here is the comparison to CA, in that we had the same issue in CA where pollsters did not understand who was going to show up, and expected a very large turnout edge from Rs.
Now, things reversed course in CA and polls got better for Newsom towards the end. But was that because things truly turned around in a matter of weeks, or pollsters had a better understanding of who actually was showing up? Partisanship took control, but I think what was also prudent for pollsters at the time was how big the early/mail vote was. It was undeniable from the beginning that Democrats were voting in larger numbers than Republicans, and IIRC, pollsters got savvy to this and realized that a LV model where Rs are getting massive turnout wasn't happening.
Now, Virginia does not have registration by partisan so we can't see who is turning out. One wonders if things would be any different if we knew exactly who was turning out in the Early Vote. Could easily show that Rs *are* turning out, but it's an unknown that I think that was something that affected pollsters models in CA towards the end. If we didn't have a party ID for who those mail ballots were, it's possible pollsters might've continued to think Rs had a turnout edge.
Just food for thought and something to think about if T-Mac outperforms the polls
The main difference between this race and CA-Gov seems to be that it only became clear at the end of that race that Republicans had basically nominated Larry Elder, whereas here it's been Youngkin vs. McAuliffe for months.
I imagine McAuliffe would be running away with this if a month before the election the GOP decided to swap out their nominee (which in CA had previously been a literal blank slate, and in VA is as close as you can get to that in human form) for Larry Elder.
Anyway, one poll, throw it in the average, etc.
