Silver's predictions of the election outcome took much of the shine off the statistician-pollster-guru, and no amount of statistical spin ("we were expressing our confidence that the unpredictable wouldn't happen, but we left open the possibility of the unpredictable!") can restore it to its former glory, but this Fivethirtyeight explainer on the polls that show a huge variance in Trump's approval and disapproval ratings is the kind of detailed analysis that is mostly light, with little heat. Read the rest