He makes many points, here is one of them:
E&O might be right that cynicism about government perversely increases support for government. But if so, libertarians shouldn't attack the public's justified cynicism. Instead, they should help people see the logical anti-government conclusion of their cynicism. Academics who are cynical about government generally are anti-government; see for yourself at the Public Choice Society meetings. Why not teach laymen to make the same connection?
I worry when I read this. Most of all, it is surprisingly meliorist; I once read a book that suggested voters were doomed to irrationality (albeit to varying degrees). If voters can be taught the correct sophisticated mix of cynicism and pro-liberty sentiment, can they not be taught to support good policies, thus making democracy a well-functioning system of government? The E&O criticism strikes at the heart of an important tension in libertarian thought. Outcomes which might be described as "good libertarian" also require important public goods to be produced at the level of overall public sentiment; there's no getting around that.
Admittedly, being pessimistic about public sentiment under democracy does not a priori mandate being pessimistic about the ability of public sentiment to support and maintain more libertarian settings. (You might for instance think that the public good can be produced under some settings but that democracy per se corrupts public opinion, because of its internal workings, electoral pandering, etc.) Nonetheless, I've yet to see good, well-fleshed out arguments to support the split claim Caplan is proposing, namely that public sentiment can be produced to support good libertarian outcomes but not good democratic outcomes.
Addendum: Caplan responds.