That’s how my friend Will describes books that he hasn’t actually gotten around to reading, but about which he’s consumed so much of the secondary literature that he feels as though he has. (A Theory of Justice seems to be in this category for huge numbers of people—which quickly makes sense if you both understand how hugely influential it’s been and have actually tried to get through it.) And that’s why I think Tyler Cowen’s analysis misfires slightly when he summarizes what he regards as the primary functions of book reviews:

1. They help people learn about good books. If this is true, we should expect a market optimum.

2. No one much uses book reviews, but they make newspapers feel like more prestigious products. In this case book reviews would be an inefficient form of product differentiation by making The New York Times appear more different from The New York Post than readers ideally would like. There would be too many book reviews.

3. People use book reviews as a substitute for reading the books themselves. I call this “book reviews as signaling.” Abolish the reviews and either a) people will have to go read the books (an even more wasteful form of signalling), or b) people will forget about literary matters altogether, which lowers signalling costs.

Now, I use book reviews for that third function all the time—though, of course, I often don’t know until I’ve read it whether I’m engaging the first function or the third. But it seems excessively reductive to reduce this to mere signaling. For one thing, most (nonfiction) books are just too damn long: Disturbingly often, you’ve got the substance of (say) two perfectly good long-form 6-8,000 word feature articles smushed together and then padded out to about 300 pages, because you generally get a lot more cred for having written a book than a handful of excellent articles. (This is the inefficiency of Tyler’s function two, but higher up the production chain.) That makes a review suboptimally short if you want to absorb the central ideas, but since they don’t typically make handy monograph versions of books, you’ve got to err in one direction or another, and given the opportunity cost of slogging through 300 pages in search of the valuable 50, the review-substitute will often be the best option.

It’s also worth noting—and maybe this is function four—that longer reviews, at least in thinky mags or publications like the New York Review of Books, are often really freestanding essays that take their subjects as a jumping-off point for a synthetic argument that draws together the ideas from several books. In these cases, “review” is somewhat misleading, unless it’s seen as intended in something like the academic sense, “a review of the literature.”