Grading on a curve and The Case Against Education

I recently read The Case Against Education, which argues that education pays off for individuals mostly because it sends signals about intelligence and conscientiousness, not because it equips students with useful skills.  Here's a puzzle which was not addressed in the book, but may be solved by that model.

My university courses were mostly graded on curves. I think Advanced Placement exams are also graded on curves, but I'm less sure of this. At the time this didn't make sense to me. If the point of taking the courses was to gain skills, shouldn't there be an objective scale of how much skill a student gained? And shouldn't students be graded against this objective scale, rather than on how they place in the distribution of students who happen to take the exam on the same day as them? Objective grading seems especially feasible for technical courses like math, but I could imagine it being plausible for humanities courses like history as well.

But if the point of a school transcript is not so much to say "Look at all these skills I acquired!", but rather to say "Here's where I am in the distribution of students who got this degree at this university", then the mystery is dissolved. The latter is obviously more in accordance with the signaling model.

Of course there are other reasonable explanations for grading on a curve, even in technical courses. If imposed by the department or university, it can be used as a defense against grade inflation. I believe it had that effect at my undergraduate university, where some departments had distributions that all professors were supposed to adhere to, and those departments had generally lower grades. But again, wanting to maintain the same grade distribution over many years is inconsistent with that grade being an objective measure of skill gained, unless we assume zero change in pedagogical methods, how much knowledge students have coming in, or just random variation in class achievement.