One conclusion I've come to over the years is that long-term trends are usually fairly predictable, but the speed and degree to which they actually occur are not.
So I think that a fair number of people on this site are a little too confident that these long term trends will overpower all other aspects of elections in the short term.
This thread from a few months ago in which a majority of posters predicted that David Perdue was more likely to lose than Bevin seems like a good example of what I'm talking about.
