>Historically, in the modern era of polling, polls have done a pretty good job of predicting the vote. In particular, exit polls have been quite accurate. In fact, when we (the US and other western countries) look at elections in other countries, when the results are widely different from the exit polls, we tend to suspect funny business.
>
>Considering how badly the Trump supporters have been demonized and attacked, I do not blame them for wanted to keep their opinions to themselves. Perhaps that might explain why the 2016 polls were so wrong? I know, probably not in your mind - it doesn't fit your narrative.

Saw an article a few days ago that US polling was not as far away from the actual result as Brexit. Did not fact/source-check myself and from memory only:
- Silver put Trump winning at 35%, which is not **that** impossible.
- Popular vote seems to have been predicted rather well.
- Polls in swing states when "aggregated" showed higher margin of error (not only the expected effect from being not at tail of bell)