This article falls well short of providing “the” reason. It presents a hypothesis and then supports it with a single data point. I also don’t see how indirect polling is going to put Nate Silver out of business. You will still need models to aggregate polls and find correlations between voting behavior across communities. That said, indirect polling does sound like an interesting tool.
You mash up the black swan events and the face saving arguments. Black swans have nothing to do with the predictability of face saving. And, yep, the clickbate "simple reason" is a clickbate.
People keep saying that the polls were wrong, but I'm just not seeing anyone supporting that assertion with actual evidence.<p>The results seem like they are very much in line with what 538 predicted based on the polls.