Plenty has been written elsewhere about Nate Silver, along with other modellers and poll aggregators, so I have little to add regarding the question of which model was best or how much models added to poll aggregation.
What I do think is remarkable, however, is that the data based approach just got a very public win. This is important, I think that the social sciences took a very big hit to their public reputation with the inability of economists to predict the 2008 little-near-almost-semi-depression. I have no idea how much of a role this has played with the fantasy-land nonsense that has been infecting our politics for the last four years, but I do think it is non-negligible. One of the most important things I learned from studying politics was a healthy respect for how vulnerable the human race is to stories and thus the critical importance of keeping our eyes on the data whenever possible. Stories involving individual action, heroes grasping the true reality through their guts rather than their lying eyes, and academic villains distorting the data for their own gain are basically always false.
With this very public win, hopefully we can move back to focusing on data more and gut feeling a little less.
I have less hope that people will get a good grasp of what data can and can't add to policy, but a step towards grounding arguments in data will be a very big step forward.
Now if only Congress will stop trying to censor inconvenient facts regarding taxation, I might start believing there has been a real change.