NIce insights, but maybe we are expecting too much from polls.
The 2016 polls, from my perspective, were showing a coin flip election. Mr. Trump won. We could lay some blame on the email controversy: I would estimate that that event caused 200,000 soft D supporters to stay home on election day. That might have given the edge to Mr.Trump. But the polls won't show a 200,000 difference very well. That's the trouble with polls trying to predict a close race.
The 2020 polls, from my perspective, suggested a not quite meagre Biden victory. The result was a meagre victory. Should we say the polls were totally wrong?
For some reason, I got on pollster's list about 10 years ago. I did my civic duty and answered their silly questions. The next time they called, I quit the poll early after I got too many silly questions. The next time they called, I told them that I wanted $50 an hour for answering their silly questions. That was the answer I gave for the next 4-5 calls. It was a good deal: I didn't get any money, but I wasn't made to feel I belonged in a Grade 3 classroom.
Since I cut the land line, I don't get calls from pollsters anymore. With people like me not responding, the results will always be less than accurate.
Data from polls should be intergreted with a grain of salt. We should be prepared for 5% or more swing betweent the polls and reality.
I don't see anything wrong with current polling. How it is useful is another question.