[ptero]

> You can't get students without grants.

That is definitely not true for many non-experimental science departments. Many students are TAs and as such care not a bit about grants. When I was getting my PhD I had to teach 4.5 hours/week (often structured as 6 and 3 at alternating semesters), which gave me my stipend and free tuition (and was a useful skill to polish). Maybe I spent another 4-5 hours preparing for classes and grading (we got student graders but I seldom asked them anything as grading was quick enough).

I was paid by the department and it did not matter to me whether my advisor got grants or not; what mattered were research interests and his guiding of research. This was the picture across the department; changing advisers was quick and based on mutual interest, not financials / grants.

[munin]

That's a very light TA load. It's far more normal to have a 20-30 hour a week commitment to your TA responsibilities (multiple lectures, grading a few hundred assignments every few weeks, etc) so it's a lot harder to fit research around that.

[ptero]

Are you serious? Can you share the school and department? This is an honest question -- I always thought 4-5 lecture or classroom hours was normal. That's what I had and that's what my friends had. Most of us never even used graders assigned as grading duties were light and it would often be more trouble to explain how you wanted it graded anyhow.

I have never heard of a 20 hour TA load. Again, just wondering.

[munin]

Every CS department I have ever heard of has a 20 hour TA load as standard.

[_ptgt]

I was in a stats department for grad school. My assistantship load was nominally 20 hours, but I managed to teach a course as a GI in probably ~10 hours/week (doing my own grading). It took more time commitment if it was my first time teaching the course, and when I had to write/grade exams.

I'd imagine my experience wasn't an uncommon one.

[ptero]

20 hours of work is pretty extreme and I suspect cuts down on your mental ability to do research (you have to take classes, too). I suspect the department usually has to say it is a 20 hr/wk position (probably so student gets free tuition), but the actual work is often less -- my friend who went to U of Maryland for PhD in CS had a similar experience to mine -- 7-10 hours a week of actual work at most.

This is just a couple of data points, though. Maybe we were just lucky.