I know people have come to respect my deeply-researched, impeccably rational analyses of things, but I’m going to just throw a theory out there, with shaky premises and no supporting data.

Predictions have been a mess this election season:  Trump succeeds and longer than anyone expected; Sanders succeeds and longer than anyone expected; and, most alarming, fivethirtyeight got Michigan wrong.

I submit that there are three things that are blowing up political models and predictions:

1. Social Media – not for dumbing down the debate, or for allowing the kind of targeting and outreach that made 2008 the Facebook election.  Social media is redefining how political communities are forming.  Social media is moving people into “filter bubbles”, helping them find niche political communities they couldn’t before, and simply connecting voices in the wilderness that are saying the same things as they are.  People can reject orthodoxy and conventional wisdom in favor of their own platform/communities.
2. 8 years of a black president being (just) a president – I can’t unpack this, but this is such a big data point for people to incorporate into their realities, that it alters mental models in ways we don’t know yet.
3. Erosion of polling and modeling concepts – concepts such as “likely to vote”, “independent”, “strongly supports Democratic”, “moderate”, “strongly support”, “conservative” have significantly less meaning than they did 8 years ago.  I have a friend who labels himself strongly Democratic and likely to vote but who has long arguments about whether to actively support a pro-choice Republican like HRC.

This all points to continuing to know nothing up to the election and even after as we listen to reporters try to summarize a hundred million votes into a single national mood.

That’s all I got.