I think the Democrats have been using polls as a cover for voter fraud. In Democrat districts, Democrats use polls to have people expect a certain candidate will win by a certain amount, and voter fraud delivers on what the polls predict. I believe this is what happened in 2018, and is why the polls seemed largely correct.

I believe if the opposing candidate wins in a blowout fashion, it takes the Democrats' voter fraud operations by surprise, and Democrats desperately have to come up with an unanticipated amount of fake votes from a slew of places, to compensate. I believe Democrats knew that President Trump would win by huge margins, so they came up with mass mail-ins, as an important means to provide cover for the huge amount of voter fraud they would have to do. I believe in the case of President Trump, the Left's rules changed a bit, and they used polls to suppress Trump voters, and to provide cover to them having Biden winning by massive voter fraud. President Trump caused such a disruption, Democrats could not use polls in the fashion they normally did, and you had the poll inaccuracies and chaos we have seen.

I think fraud investigators should be able to do a post-mortem exam, and see what their models predict the outcome of what the elections in districts should be, by calling up random samples of people; then compare the actual results for those districts, to get a sense of just how bad voter fraud may have been from district to district, in various states. I believe this could be used as a tool to gauge possible voter fraud, and to provide evidence in court.