Gonna preface this by again reminding you that I'm not a pollster and instead my job is to assess how accurate polls are and build models of what the errors look like. So as someone who spends a lot of time on this, I'd say media coverage on this topic has been crappy so far.

The main issue is "you should wait for all votes to be counted to judge the accuracy of polls" may sounds like lame excuse-making from pollsters, but actually that's pretty damn important as vote counts have changed dramatically and are still changing in many states.

We also know very little about the whys. Why was polling bad in the Midwest (except in Minnesota?) but pretty good in the Southwest? Why did it work in New England but there was a big miss in Florida? This will take a while to unpack, and a lot of initial takes won't age well.

Something like this is plausible, for instance: different patterns of who was at home or working from home because of COVID, which probably correlates with political preferences in all sorts of weird ways. https://twitter.com/kombiz/status/1325059234950770689?s=20

Personally, given a lot of state-by-state and region-by-region variation, I suspect it will turn out to reflect a combination of several different factors. The same goes (polls aside) for vote shifts from 2016 to 2020. This is a WEIRD, CRAZY map! https://www.nytimes.com/interactive/2020/11/03/us/elections/results-president.html