How much difference did Bernie Sanders’s historic upset in Michigan make?

At FiveThirtyEight — whose proprietary “polls-plus” model gave Sanders less than a 1 percent chance to win in Michigan — Harry Enten notes that the almighty “delegate math” still points strongly in Clinton’s favor. In the New York Times — whose even-more-secretive demographic model apparently showed Sanders losing Michigan by twelve points — Nate Cohn argues that the “probabilities” of Sanders actually winning the nomination are unchanged by the Michigan upset.

Oddly, Cohn’s crowning piece of evidence seems to be the fact that the PredictWise electoral market barely budged after Sanders’s win, giving him a 7 percent chance to win the nomination, as opposed to 5 percent the day before. Of course, this is the same online odds aggregator that gave Sanders a 9 percent chance to win Michigan on Election Day — literally nine times more accurate than FiveThirtyEight, but hardly a beacon of insight into the Democratic race.