Michael Huemer argues that "the benefits of the vast majority of academic research are tiny at best, possibly negative, and much less than the costs." While I don't dispute his claims (in section 3) that universities are prestige-chasing institutions, I think his framing of the issue leads him to overlook the main sources of value that can nonetheless emerge from these highly imperfect institutions.
(1) Extreme cases boost expected value. As I wrote in my defense of impractical philosophy a decade ago, even if the "vast majority" of research is entirely lacking in value, "the few exceptions -- the Turings of the world, whose theoretical passions lead to invaluable insights -- are arguably so momentous as to justify the whole system that enables them." So I think it's a mistake for Huemer to focus on median rather than mean (or expected) value.
Granted, to evaluate whether more or fewer resources should be expended on academic research, we must attend more specifically to the question of marginal expected value. And one might imagine a more efficient system where just the best academic researchers are funded. But such efficiency requires being able to identify the best researchers in advance, and it's far from clear that this is feasible. (Some academics are "late bloomers". Some are one-hit wonders. Much would be lost in a system that was too selective.) So, depending on the details, it may well be worth tolerating plenty of "chaff" in order to get a bit more of the vital "wheat" we need to advance civilization.
(2) Academic knowledge-building is a collective project. Huemer writes, "my writing a new book benefits you only if my new book is good enough to displace one of the books on [your lifetime reading] list." This individualistic focus is, I think, the wrong way to think about academic knowledge-production. Most academic research is not directed at consumers, and as such should not be thought of as consumption-ready "finished products" at all. Rather, our individual publications are building-blocks in the edifice of our discipline's collective knowledge -- or as we call it, "the literature." Textbooks and popularizing works are crucial for translating this collective store into something accessible to individual consumers. But for the most part, academic work should be evaluated for its contributory value, not its consumption value.
For this reason, complaints about inaccessible prose, hyper-specialization, and small readership, strike me as largely misguided. If a work is so little-read or engaged with that it makes no real difference to the development of the literature, that's a problem (but distinct from small readership as such). And if we lack sufficient "big picture" generalist works, or accessible literature surveys, those too are reasonable concerns. We need different kinds of works to play different roles. One of the most important (but easily overlooked) roles is to provide scaffolding for later works to build upon. This means that inaccessible, rarely-read, hyper-specialized research may still contribute to the knowledge edifice in a way that helps subsequent work to be more valuable than it otherwise would be.
(3) Intrinsic value is a big part of it. Huemer remarks that he "won’t comment on intrinsic value." But it's hardly possible to answer this question fairly (especially for humanistic disciplines that obviously don't aspire to practicality) without doing so. We would not be surprised to hear demands to cut funding to the arts & humanities from those who aspire to be a pig satisfied over a Socrates dissatisfied. The case for (most) A&H research ultimately rests on a value judgment about the (non-instrumental) importance of intellectual pursuits in human life and society. A critic might suggest that Plato should have been an engineer. The correct response, I feel, is simply to recoil in horror at such a barren vision of what life is about.
(To clarify: Engineers are great! What's barren is instead the idea that this is the only kind of greatness to which people might aspire.)
Postscript: Opportunity costs? Huemer thinks we'd clearly be better-off with more academics instead "working in other areas — whatever other professions the smartest, most hardworking people would be most likely to join if they weren’t academic researchers." I'm honestly not sure what he's imagining here. Law? Finance? Business consultants? I'd take more academics any day. (Humanities profs might be more likely to become novelists or pursue some other creative/artistic vocation, which is fine and all but not obviously more "practical" than what they're currently doing.)