"Polls in 2016, perhaps skewed by those who would not admit they were voting for Trump, showed he would lose to Hillary Clinton." - This is not a very good assessment of what polls showed. https://www.poynter.org/newsletters/2020/should-we-be-paying-attention-to-presidential-polls/?fbclid=IwAR2Vk0TswJdn4w9YEKlQ8TTTcU55gVolB5mEdzsyArV6Bz3NAOebUoTlqvI
So, with just about 1.5 months to go, it looked like a super-close race. More than that, there were still a lot of undecideds.
The problem isn't that polls mislead us. It's that they are so routinely treated as something they're not — indications that someone is "winning" or that strong momentum is building based on relatively small variations in polls with MOEs.
Journalists and analysts should strive to do a much better job understanding and explaining polls, not dismissing them.