By late last night, it had become abundantly clear that poll after poll after poll that had projected a 1992- or 1996-style popular-vote and electoral wave for the Democrats had been wrong, wrong, wrong. Ever since, I’ve been pondering a single thought: In the battle between polarization and polls, polarization won.

In other words, as I wrote in a brief essay in Time, Americans were confronted with two conflicting streams of data. One was the avalanche of polls showing that Biden was clearly in the lead and that if the polling held, he could potentially win 350 or more electoral votes plus an eight or nine point popular vote victory. The other was the decades of data outlining the rise of negative partisanship and increased American tribalization.

We were also often confronted with a conflict between the polling data and the testimony of our own senses. If you live in Trump country you saw the enthusiasm of Trump voters. If you talk about politics at all on social media, you could read the intensity of the division. 

The central reality of negative polarization predicts outcomes just like this week’s results. Remember, the definition of negative partisanship is that you vote for your side not so much because you love your own side, but rather because you dislike your opponents. Therefore, no matter your candidate’s failings, he has one abiding and unassailable virtue—he’s not the other guy.