Frum's point, however, is that
As many political scientists have demonstrated, the parties are becoming more polarized even though the electorate is not. The cause of the “disconnect” (as Morris Fiorina calls it)? Party elites, both Democratic and Republican, have found ways to take command of party institutions and steer their organizations further and further away from the broad preferences of the country.Absolutely correct -- but incomplete. I've talked about some of these issues before, but I think it might be helpful to lay them all out in one post.
The process Drum describes is certainly one party impulse, but it isn't the only one. There's also the Downsian incentive to move to the center (of each district) in order to win. Typically, these impulses are in tension, but the money is in the middle: that is, politicians want to win in order to have jobs, political consultants want to win in order to enhance their reputations, many party-aligned interest groups want to win in order to implement as many of their public policy objectives as possible, party bureaucrats want to win because it tends to generate larger and better financed formal party organizations, and policy wonks want to win so that they can get government jobs (which is not about immediate financial reward, but does tend to help their long-term earnings; they also might actually care about changing the world). Each of these incentives push a political party towards the center, which is the ideal position for winning elections in a two-party system.
Against all that are those interest groups that are willing to risk and all-or-nothing strategy. and support candidates who are less likely to win but would take extreme positions if they do win, and those with no material self-interest in the success of the party who tend to be ideological extremists -- the political science literature calls them "purists" or "amateurs" (although note that the latter is only a tendency; it's certainly logically possible to be a pragmatic amateur). Also pushing parties away from the center are primary electorates. Primary elections -- you might not know this, but very few nations have adopted that particular American innovation -- introduce another possible polarizing incentive, since candidates must appeal to the median voter in a primary electorate in order to win the nomination, and if voters are even somewhat sorted ideologically into parties then that median point will be to the left of center for Democrats, and to the right of center for Republicans.
Most, but not all, of the self-interest incentives push parties to the center. There are also, however, another set of factors that can push parties to the extremes. Parties have their own internal cultures and information flows, and they can affect the behavior of party actors, even if outside observers might find those actions irrational in some objective sense. For example, a politician might erroneously but sincerely believe that there's a hidden vote available to candidates that ignore the middle of the electorate in favor of mobilizing the party's base (it's possible that there might conceivably be rare circumstances in which that's a wise electoral strategy, but normally any candidate who captures the center will win an election; what I'm talking about here, however, is when those rare circumstances do not apply). A partisan policy wonk might mistakenly believe that her issue positions, which are well-received on partisan blogs, are actually far more popular than they in fact are. And some politicians and their staffs may simply hold issue positions because they really believe in them, regardless of electoral incentives (and, in most cases, that won't actually hurt them very much with the electorate, which is far less attentive that pols and other political actors believe).
If you're still reading this long piece, what you probably want is a takeaway paragraph, but unfortunately I can't give you one. Political scientists have reached no consensus at all on which of these incentives, or even which actors, are the important ones. Some of us think that politicians are the crucial actors. Others believe that pols will ultimately follow whatever bargains are reached by party-aligned interest groups. My own (unpublished as yet) claim is that there is no ultimate answer: the results of both which actors will be most important and which incentives they will follow are contingent on the (formal and informal) rules of the game, and on all sorts of other things happening in a society and its politics. I know: not really helpful, is it? Sorry, but that's how I see it. The main thing is that, if I'm correct (which of course I believe I am) there are a wide range of possible stable outcomes when it comes to polarization.
One more thing, which might be fairly important. The core assumption of everyone who has studied party incentives has always been that winning is always a good thing. For purists, it might be a good thing that is trumped by other considerations, but we've always assumed that winning office was at the very least a neutral factor, and in almost all cases, certainly for any professional politician or operatives, incentives would always run toward winning office. That no longer appears to be true, at least (or at least mainly) on the conservative side of the spectrum. There's just no getting around the fact that there is a large conservative marketplace, and that there's more money that can be squeezed out of that market when Democrats take office. I don't know that any conservative operatives actively follow the obvious incentive and consciously try to make their own side lose elections, but the incentive most certainly exists, and may well affect behavior in some cases.
All of which still supports my general point I've made in the past, which is that while polarization is a natural development, there are a lot of different possible degrees of polarization, and the current levels are very high -- and hardly inevitable.