On something Frymer mentioned below, Downsian models of party competition assume a "uni-dimensional policy space. Any good campaign manager knows this is bull. First, at any one time there are a number of potential axial issues in a campaign--the trick is to pull attention away from the ones that you are distant from the median voter on, and focus it on the ones where you are closer to the median voter than your competition. Second, voter understanding of the policy content of party positions is mediated by rhetoric--voters make choices on the basis of their impression of party positions, not the positions themselves. Therefore there is substantial leeway in the political process for ambitious politicians who want to make policy closer to their party median, as opposed to the public median. They can use rhetorical packaging and manipulation to conceal the real effects of their policy positions (I commend Jacobs and Shapiro's Politicians Don't Pander on this subject). In a competitive political environment, this sort of manipulation should wash out, as the other side attempts to unveil the real content of the policies underneath the rhetoric--and to manipulate the public's impressions of their alternative at the same time. In practice, the market in public impressions does not necessarily "clear," because parties are not equally organized to influence public impressions of policy packages.
Republicans have been better at Democrats at this sort of rhetorical manipulation for some time now, at least since the Contract with America. I say this without animus--that's grown up politics as it's played in modern America. Instead of complaining about how effective Republicans are at this game, Democrats need to construct a sufficient organizational apparatus to counteract it, and to play offense as well. Until they do they'll keep getting creamed politically, and even when they win elections they will get less out of office than Republicans can manage.