Harvard professor and New Yorker contributor Louis Menand has a new book out in January, The Marketplace of Ideas, in which he examines the U.S. university system. A very interesting, if unfortunately a little scattershot, series of excerpts appeared in the November Harvard Magazine.
On thing I think Menand totally missed, at least in the excerpt and perhaps not in the book itself, is the question about Ph.D training and professorial jobs on the one hand, and women, childbearing, and family work/life balance on the other. He writes about the rise in competition for academic jobs due to overproduction of Ph.D's, and about the rise in median time-to-degree. But he doesn't discuss the interplay with the life cycle especially as regards family formation.
One could make an argument that a lengthy time-to-degree setup is actually preferable for some scholars who wish to have children during training. The alternatives would be either not getting a degree, not having children, having them before the degree, or having them afterward, possibly during the stresses of a tenure-track position. I don't mean to suggest that this argument must be correct, but it deserves some thought.
Fairly late in the excerpts, Menand acknowledges that graduate students are an important part of the academic labor force. If tenure is important for preserving academic freedom, labor with relatively more flexibility is surely an even more attractive element.
And as a posted comment at the bottom of the excerpt points out, the physical sciences are really very different than the humanities and social sciences in their industrial structure. Postdocs in physical science are extremely common; I think median time-to-degree is thus a bit misleading when compared between physical and social sciences, for example. Lab work and dissertations that flow from a directed lab are also very common. And grad students in physical sciences not only teach, they also produce research, sort of a "double-whammy" incentivizing the professoriate toward keeping them around practically as indentured servants.