Despite evidence that this election is the most bizarre yet (the first failing of the WIU prediction, for example), we can take solace in the numbers. The polls are not doing anything that they didn't do during the last presidential election. Look at the last 3 months of this election, compared to the 2012 election:
During that time, both experienced a primary Democrat lead, with the Republican taking lead, or getting close, three times. In 2016, it's happened further apart, but it's still happening. The 2012 RCP poll correctly predicted Obama would win by 0.7%, when in reality it ended up to be 3.9% (the reading on September 30th, which indicates that "October Surprises" may be a thing of the past. However, I think there is the likelihood that the final results could further apart than the RCP final numbers (because the voting public may end up different than the "likely voters" due to polarization of politics) -- probably enough for me to not have the confidence in the poll that I did last election.
*The numbers to beat are 80 million, the biggest debate to-date (1980's Reagan vs. Carter), and 114 million, the biggest television airing to date (Superbowl 2015). The tricky thing is that some potential TV viewers are going to move over to the Internet side of things, as this debate will be broadcast on many websites and Social Media apps, and we'll have to decide how to count that.