I agree with Phil that courses should be clearly distinguished when they represent the views of the professor, rather than an attempt to neutrally survey a body of literature. It's probably best to create a separate classification scheme for such classes, especially given that they represent a distinct pedagogical approach. I also think it's best that these not be mandatory classes, either for general or departmental curriculum requirements. In classes that are no so identified, professors could then be legitimately criticized for an absence of neutrality or breadth in their reading list.
On the matter of ideological or religious diversity, I would say that, yes, in most elite schools there is little question that they would be better intellectual institutions with greater substantive breadth of professorial philosophy. In practice, this means more conservatives and libertarians. My main problem comes with how you get there. Phil seems comfortable imposing something close to the existing anti-discrimination regime on departments. I'm uncomfortable with this. I think in a liberal society that the anti-discrimination cure, especially in its stronger, affirmative action mode, is very strong medicine indeed, and should only be imposed where there is a pervasive problem unsuited to more libertarian cures.
On the one hand, I think that the problem of an absence of conservative/liberal professors is a supply problem (less so in law schools, though, I might add, since they hire out of the pool of JDs, some of whom go into academia, and some of whom go into practice--there we have a reason to believe that it is a selection rather than a supply issue). This supply problem is, in part, a consequence of rational expectations of the future labor market they will face, and so in that sense the problem is akin to that of African-Americans, who may, in anticipation of discrimination, alter their behavior (for example, going into professions that are more public-sector-dominated, where they anticipate lower discrimination or network hiring). The solution, in my mind, is for elite graduate institutions to make a greater effort to recruit conservatives into PhD programs in the social sciences and the humanities, and give them as much room to pursue research programs that reflect their beliefs as they allow graduate students on the left. Currently smart conservative college graduates often head into the Washington, DC think tank world instead of going to graduate school, because of their expectation of future discrimination, and in the short term of differential treatment in graduate school. So I think part of the problem is not overt discrimination (although there is some of that), but rational adaptation to the expectation of future discrimination, which may in fact be less pervasive than conservatives think.
The trickier issue is what to do about overt discrimination in hiring, which certainly happens, although, again, it's hard to tell what the magnitudes are. If maximizing substantive outcomes were all we cared about, it might be that imposing something like affirmative action obligations on hiring departments might be the solution, not just to the discrimination problem, but also the supply problem (since this would alter conservatives' rational expectations of future discrimination, thereby inducing more to enter graduate training). The problem is that, as I noted before, this is a fairly illiberal solution (do we really need to add MORE bureaucratic rules on the hiring process than we've already got?), and also that, unlike race or gender, the problem of identification is infinitely trickier. Are we going to ask everyone to take a 25 question test of their political views, and give preference to those who come out on the minimal government/ social traditionalism end of the test? Do we even want to start collecting data like this? What about the "stigma" of affirmative action hiring that conservatives always complain about when it comes to blacks?
In the end, I think the answer to this problem, to the degree that there is a solution (and any good conservative would say that there are some problems that, regardless of their severity, are not worth the treatment, or are not susceptible to treatment), is good old John Stuart Mill-ish: publicity. Large departments that continue to fail to hire or train conservatives are, probably, engaging in either overt or covert ideological bias. They should be shamed into at least becoming sensitized to whether they are importing into their hiring decisions ideological decision-making. To the degree that this is what Horowitz et al are doing, I think there is nothing wrong with it. But I think as good conservatives they should foreswear the use of the anti-discrimination paradigm here, which has its own problems where it is already applied and certainly doesn't need to be extended to every possible human difference.