It may be naive today but it wasn't a generation ago, when the numbers were more favorable to a young PhD. Something has changed, which is why people write articles to say, "Hey, stuff's changed!"
A pyramid scheme works well when the pyramid is expanding. It doesn't work so well when the tip of the pyramid is shrinking.
If EACH tenured professor has 1 graduate student per 3-5 years, that's nowhere close to the rate at which new tenure track jobs open up. There are some private sector jobs for Math PhDs, but even those have an oversupply.