There is a lot of work to be done, but there are many people who work 80 hrs/week to be paid < $100k / year.
I'd love to see the numbers on what the all in dollar value of benefits provided to the students is, as I think that would shift the opinion of quite a few people here.
Work experience is education. That's why you get paid more for past experience - it saves the company on a cost they would otherwise have invested in you. Doing a PhD is just another way of getting work experience, except you get paid a lot less.
I think it's reasonable to claim that PhD's are compensated a lot by virtue of the "cost" of educating them and other benefits. But this education is something you receive in a normal job too. And just like in a normal job, the employer gains a lot more out of the exchange. Grad students are typically worth a lot more to universities than they cost, otherwise there would be no sense in admitting so many. I'd wager the ratio of value to cost is perhaps greater in academia than it is in a lot of other industries.
Even allowing for some slack, I'd suggest that a grad student is seeing a pre-tax equivalent of $25k - $100k, and probably skewed to the lower end of that, based on your numbers.
Given that the educational (and academic publishing) systems exist as gateways into a highly-limited cabal, the arrangement cannot reasonably be called free-market.
I'm in favour of unionisation.