I try to shy away from direct politics, as it both makes me angry and also is an area where my opinion is little more than spitting in the wind of bullshit, regardless of which side you root for. That being said, this is some smart reflection on data-driven journalism of the kind Nate Silver helped to pioneer.
The Republicans proved last election cycle that trying to say the polls are wrong is a losing strategy. The Democrats need to pick up on this too, and quick if they want a chance to move the line vis a vis Silver’s Senate predictions. As I have no reason to believe Silver is anything other than a forthright quant and data nerd, he should be quite aware that his predictions are subject to a macro version of Heisenberg’s uncertainty principle: by simply publishing his results, he’s changing public sentiment.
I do believe Silver has an agenda here, but it’s a straightforward and understandable one: he wants people to want his data. His senate predictions prove he’s succeeding. Now, the onus is on those being bet on (to borrow and extend Young’s Vega metaphor). Good statistics is science: not believing in it in no way affects its accuracy.