In the political season Nate Silver benefitted from far superior internal polling data from the Obama campaign, and was much more accurate than our models built on what we later learned were polling data that did not even survey cell phones. Garbage-in-garbage-out.
In sports we are all working from the same data, and Nate Silver’s projection of a Seattle vs. New England Superbowl was not even close.
Seattle lost a couple of days after his prediction to the Falcons, who in turn lost to the 49ers. The Patriots lost by 15 points.
Now, not many picked the Ravens v. 49ers Superbowl, but we did say before the wild card week that Seattle had the best team in the second half of the year, and they were the only visiting to win that weekend. Then when Nate picked Seattle to go to the Super Bowl after that game, we noted that that was going to be a lot harder than he thought with them having to win at Atlanta, and they fell short.
We did also suggest last week that the Harbaugh family might be the greatest coaching family, and now they are playing in the Super Bowl.
We also did not pick the New York Jets to go into Seattle and beat the Seahawks, which Deadspin suggested discredited every other correct prediction Nate Silver had ever made.
The point here is not to pick on Nate Silver. The point is to note that Republican political models work–we simply need Republican pollsters to do a better job so that we are working from the same knowledge base as the Democrats.
By only polling landlines in 2012–in complete denial of a quarter of the population relying exclusively on cell phones and many predominantly communicating on-line–Republican pollsters did not give Republican campaign managers a chance to formulate winning strategies.
When young engineers and programmers get excited about a GOP stance, and they GOP does not know it, they miss opportunities. When Republican candidates are overconfident because they believe they are protecting a lead, it can cost Republicans elections.