Seth Roberts has an interesting take on the real purpose of college.  He starts unpromisingly:

Almost all college students want to figure out what job to choose. The
answer will depend on what they do well, what they enjoy, and will have a
big effect on the rest of their life. The better the answer, the more
successful and happy they will be. For them, that is above all what
college is for.

The obvious problems with Seth’s claim are that (a) most college classes have few corresponding careers, and (b) most careers have few corresponding college classes.  But he swiftly acknowledges these points:

There is nothing terrible about college classes. I don’t say that this
or that humanities course is “useless”. The trouble is lack of balance:
too many normal classes, too few “classes” that explicitly help students
to learn about the world of work and how they might fit into it. Only a
few colleges — often low-prestige “trade schools” — do much to help
students learn about possible jobs, what they enjoy, and what they are
good at.

Then Seth brings a big smile to my face by reminding me of my years at UC Berkeley:

Judging by how Berkeley courses are taught — they do little to help
students decide what job to do, unless they are seriously considering
being a professor — most professors have little or no interest in
helping students this way. I suspect, however, they don’t know what they
might gain from doing so.

In short, Seth’s theory doesn’t fit the typical college, even at the elite level.  Unless, of course, Seth happens to be your instructor!

At Berkeley I taught a class called Psychology and the Real World whose
goal was exactly that: help students find their way (a particular
problem for psychology majors, few of whom go to graduate school in
psychology). They could do almost anything, so long as it was
off-campus. It was little work for me and the students learned a lot.

Punchline: Seth still has no explanation for why actually-existing higher education pays so well in the real world.  He should just embrace the signaling model as an accurate description of the status quo, then add, “Professors can do a lot better; I already do!”

HT: Tyler