Something I've noticed in the lead up to the USA election, which I see in many elections, is that hyper-partisanship tends to really get in the way of people making well thought out predictions (this hyper-partisanship leads to a lot of suboptimal decision making, not limited to just election predictions). It seems this extends into the people who do the polling sometimes as well, which is a factor we saw in the 2016 USA election. I think this is actually part of a broader trend in countries that share some common political cultural elements regarding differences in how open certain groups are to expressing their political preferences. It also seemed to be the case in the 2019 Australian federal election where the ALP was tipped by the pollsters to win comfortably but ended up losing on election day. I think part of why this happens is that there's a substantial bias now emerging in regards to who talks to people doing polling, the models for polls has attempted to change to account for this but how successful they will be remains to be seen. I think part of the root cause for this is that certain big tech platforms have political bias and the people on those platforms whose views are within the "Overton Window" (to borrow a term) of whats "allowed"1 on those platforms are more able to express their opinions and get more accustomed to expressing those opinions publicly on a regular basis provided they are "allowed" on those platforms. People who have opinions that fall outside this window may get less comfortable expressing their opinions publicly and this I suspect filters through into their comfort in expressing their opinions to pollsters. In some senses I think this is a good example of preference falsification in action that in some cases can end up biasing the results of certain polls. Another aspect to this is that some people are simply embarrassed by the candidates they would be voting for, this is a big factor in why people might not share their preferences when they are voting on a "lesser of two evils" choice in their mind.
It's not too late on the 4th of November in Australian time (where I'm posting this from) so due to timezone differences the election count is not complete at this stage. I'm just wanting to be on the record as saying that I think this will be a close election. It was really interesting in the lead up to this election seeing outright censoring of certain news from various groups in "big tech", I think this sort of action then feeds into people making bad predictions of the vote since they just haven't got all the information that they need. Perhaps this will make a big different in places like Philadelphia that were hit by some waves of rioting and looting that didn't really hit the news much but likely influenced local sentiment and voter turnout.
I also think that because postal ballots can in many places be counted as long as they were postmarked before the election and arrive before Friday in combination with the huge number of postal ballots cast there's a really high chance that there's a big delay before the winner is announced. I expect there to be a number of litigation issues around the postal ballots as well, hopefully those get resolved quickly but I expect to see issues there as well. I don't know what will happen but I'd be surprised to see a clear winner emerge on election night or even the day after.
Because of the way in which many platforms are now so heavily driven by algorithmic feeds it's very easy for a platform to control the reach of participants messages and posts. So even if a platform doesn't want to outright ban certain types of messages because of fears of falling afoul of the protections afforded to them by section 230, they can effectively limit certain messages more than others without outright censoring them. ↩