While I've long been a believer in Alan Lichtman's keys for predicting US presidential elections, I'm actually becoming somewhat dubious about that line of research in that political operatives now appear to be trying to manipulate conditions in the country in ways deemed most favorable to their parties. So if, you're in the opposition, you try to make sure the incumbent President accomplishes absolutely nothing (whether it's in the public interest or not), that there are major unrest and perceived scandals, and that the economy stagnates; while if you're associated with the President's party, you advise him to score easy points instead of focusing on the big picture. And I have to think that the sense of inevitabilty has the effect of discouraging people from voting; or running if they "can't win" (neither one of which makes for a healthy democracy).
None of the above has anything specific to do with Nate Silver's research, but as election prediction is what he is famous for, it applies to his work as well. Isaac Asimov discussed the problem of predictions affecting results many years ago in his "Foundation" series (it is a major theme of those stories). As time goes on, his concerns seem to be increasingly relevant.