I'm sincerely curious. Do you know where I can find Silver discussing polls that kind of jigger with their results?
https://www.reddit.com/r/fivethirtyeight/comments/1ggpqqi
Silver's become a bit of a crank, but his overall point that we appear to be seeing herding seems sound.
There just isn't enough variance in the polls we are seeing released.
There are subtle ways to skew a poll ever so slightly. You simply change the composition of the voter pool. You want to move the needle a bit for Harris you start your polling selection with a few percentage more women. The pollster then states that they think the female turnout will be higher. Inversely, you do the opposite to skew towards Trump.
Or you fuck with the weighting in ways that aren't justified. (You don't adjust the selection pool itself.)
Amounts to the same thing.
Or you just don't release polls you don't want to release.
As I posted in one thread or another, Trump is quite convinced the polls he pays for just tell him what he wants to hear (according to his interview with Rogan) and that polls showing that he isn't winning are illegal and the people who make them should be arrested. (According to his stump speeches.)
The consistently tight poll results one after another from a large body of pollsters has been called out as very unlikely by some. One wonders if many of the pollsters are deliberately trying to stay with the herd.
This was Silver's argument (and not his alone).
If your polls aren't outliers, people won't be mad if you miss things by the same amount as everyone else.