In 2016, the polls were reflecting a coin flip election--even if many people were using the same numbers to predict a D victory. In 2020, the polls were predicting a narrow D victory, which it was. I'm not sure how the pollsters called these two elections so correctly. For various reasons, I don't believe in their accuracy:
https://medium.com/tiered-democratic-governance/are-polls-believable-423e12092d74
What the traditional non-voters do in such an election is indeed a mystery. They have the outcome in their hands--and it's hard to read their preference and motivation to vote. The pollsters are not that good.
You might be interested in my plausible outcomes for November 5. The math is there for Republicans to take a big beating on that day:
Project 2025 is finally getting some attention. That could entice nonvoters to vote. But that issue will likely be over by September.
