OK, so Bryan's latest book is on stupid voters. One solution is to educate them. But his next to-be-written book, on education, says that education is merely a signal of ability. The data on GRE scores arguably validates that.
In any case, how can you believe on the one hand that education is merely a signal and on the other hand believe that education can produce more rational voters? Won't people's rationality as voters (or, more accurately, their willingness to vote in ways that Bryan or I would consider rational) depend on ability alone?
1. Steve Miller and I have a paper where we examine the extent to which the tendency of education to "make people think like economists" is actually a disguised effect of IQ. We find that at least one-third of the apparent effect of education should actually be attributed to IQ. So while there is something to Arnold's concerns, there is still plenty of room for education to matter.
2. You might object that it will be extremely costly to significantly raise the average education levels of people with average or lower IQ. I'm sympathetic, but this ignores a cheaper, more realistic alternative: Revising the curriculum to emphasize subjects, like economics, with large political externalities.
3. Just because education is largely signaling, it does not follow that students are not learning anything! The point, rather, is that students are not learning job skills. I don't deny that students learn history in school; I just deny that knowledge of history makes people (historians aside) into measurably better workers.
Furthermore, strange as it seems, the signaling model should make us more optimistic about the prospects for curriculum reform! After all, if schools are teaching job skills, asking them to teach more economics to improve the quality of public policy is pretty quixotic: "Please, couldn't you focus more on society and less on what actually benefits your students?" But if schools are just making students jump through hoops to prove themselves, what's the harm in switching to hoops with positive externalities? Asking students to signal their ability by learning subjects with positive externalities could well be a free lunch.