The median household income in California is $77.5k [1]. I'll emphasize household. Your 9-month pay was for 20 hours, that means $25k is actually $66.6k annualized which does not include your subsidized housing nor for inflation. The economics just simply is not there. Sure, students might be able to squeak 10-20% in wage increase at the cost of all other university voluntary support such as student group support and things like free gym memberships.
Furthermore, I can very easily see one "professional" TA replacing 5 grad students. Just use HN search to find how many people's advisors told them that teaching does not win Nobel prizes. To be frank, most (not all) grad students do the job poorly. First, the professional TA automatically is 2 grad students because they can devote twice as much time, but then there is economies of scale where they can teach the same stuff multiple times in the larger classes. Again, the number of people would would be signing up to be a professional calculus TA for $100k would be very high.
Sure, you switch out four calculus TAs for calculus with professional TAs. Who is going to TA deep reinforcement learning, or under-actuated robotics, or any of the major-specific courses that takes a lot of relevant knowledge to actually teach properly? Any of the people out there who has the insights to TA these classes thoroughly enough are gonna point and laugh at you if you offer them anything below $200K.
Also, seems like you are completely ignoring the actual "labor" that the graduate students provide for their research labs. The number of professors in Computer Science at MIT who actually writes code at least once a week may shock you, but the truth is, for MIT to be on the top of the food chain someone needs to keep the wheels turning.
> Also, seems like you are completely ignoring the actual "labor"
But I'm not, you are literally not paid for that (unless you are a research assistant). That is "studying" which I keep pointing out is the whole issue. Which feeds back to your point about highly-specific knowledge TAs. The implicit agreement is that the university provides you with the training required to TA these classes in exchange for being a TA. They normally call this waived tuition. That is, they are supplying you with a +50k benefit for free which is how they draw many people to begin with.
Not particularly, I gave that example because I was TAing that class last semester. Incidentally, it seems we have more in common than I thought!
> But I'm not, you are literally not paid for that (unless you are a research assistant).
This may be a difference in experience and expectation, but in my experience I've always seen graduate student salaries paid out to Ph.D. students employed by the university as an RA or a TA. If you are neither of those, then you do have to pay tuition to the university.
> That is, they are supplying you with a +50k benefit...
Which is another problem with not having a union -- the university unilaterally sets these price stickers without anyone challenging them, and then "waives" it and appearing generous. Without a union, MIT is free to inflate the sticker price of a graduate "education" by 20% a year, thus giving students a 20% raise every year while their life situations don't improve at all. The current student body had no mechanism to prevent it or challenge the administration on shady practices, but hopefully this will change with the union.
> The implicit agreement is that the university provides you with the training required to TA these classes in exchange for being a TA.
Finally, most Ph.D. students after their second year won't even take a class but are already TAing, so by your definition the necessary training is already complete; but they are indentured to work for far below market rate for the three years left on their Ph.D., because...?
> Ph.D. students employed by the university as an RA or a TA.
The vast majority of students are paid through TA (which you have admitted that you were last semester). This means all your other "duties" was your "study". I agree with everything you say, but there is literally nothing a union can do about these things because it these things are not your job. I know it feels like a job. Writing code, SSHing into the cluster, analyzing results. All things that I continued to due in my industry job after graduation. But, this is the key part students don't seem to understand, you are not employed by the university to do this. It does not matter if it is part of a course. This is the sole reason international students are able to come on F1/J1. Legally and practically, you are doing this under the guidance of your advisor. Others have commented in this thread as well, RA becomes very difficult to separate what an RA tasks are and what your PhD student tasks are, but this is the crux of the issue. Your union might be able to fight for increase RA pay, but your advisor/department/university can just forego your RA and accept your student work.
> If you are neither of those, then you do have to pay tuition to the university.
Then you are not an employee and cannot unionize. At least, not in a trade union.
> Which is another problem with not having a union -- the university unilaterally sets these price stickers without anyone challenging them, and then "waives" it and appearing generous...
This is the problem with these unions. They make it appear to the students that they can solve these problems, but they legally cannot. A union cannot bargain for these things. Just like a union cannot bargain with Ford on the price of a F150, a grad student union cannot tell the university how to run its programs. At least at my institution, some students thought that, if we unionized, they could negotiate for fewer teaching requirements. The very thing that makes them employees and eligible to be in the union.
Furthermore, I can very easily see one "professional" TA replacing 5 grad students. Just use HN search to find how many people's advisors told them that teaching does not win Nobel prizes. To be frank, most (not all) grad students do the job poorly. First, the professional TA automatically is 2 grad students because they can devote twice as much time, but then there is economies of scale where they can teach the same stuff multiple times in the larger classes. Again, the number of people would would be signing up to be a professional calculus TA for $100k would be very high.
[1] - https://fred.stlouisfed.org/series/MEHOINUSCAA646N