(1/n) This is pretty speculative, but I did the following exercise to see what happens if the polls are ~5 percentage points too high for Biden (which is how much MI and WI were off in 2016 according to @fivethirtynate here https://fivethirtyeight.com/features/is-joe-biden-toast-if-he-loses-pennsylvania/).
(2/n) I assigned states with a greater than 5 pp lead to Biden. They add up to 259, so all it takes is AZ, PA, GA, or FL to go blue. These states are in brown, which means they are closer than that ~5pp gap (though PA is 5.1 https://projects.fivethirtyeight.com/polls/president-general/pennsylvania/).
My takeaway is that PA seems much closer than ideal and that I'd like to hear more from folks who study this on (1) how much polling methodology has changed since 2016 (to see if 5pp is too high) and (2) how much bias there may be in AZ GA FL NC where the polls are much closer.