I have 3 graduate students working for me right now. I find this to be enough, along with the undergraduate students I advise on projects (usually 1 or 2 at any time), other faculty I collaborate with, and post-docs or research scientists that I support and/or work with around here. There are a few faculty in my department who have 5 or 6 PhD students at a time. I am not sure I want that many people reporting to me and looking to me for direction and mentoring. I know of a few in our field who claim to have 10-20 graduate students in their group. I don’t see how this is humanly possible. In fact, I don’t see how that is even remotely responsible.
To me, having more than 5 PhD students, let alone 10 or 20, is irresponsible on several levels. It means that the faculty member is continuously in meetings with these people (assuming that they have regular interaction with each student). This cheats the faculty out of time to do their own research investigations, and they are probably giving up all home life in order to publish the first-author papers that I see from them. It also means that the students don’t get very much one-on-one time with their advisor. Perhaps this is fine for some students, but I have found that most PhD students like regular interaction with their advisors. In the early years, they like it so that they learn the field and discover a research project that suits them well. In the later years, they like it because they have results and need to show them to someone and get feedback. It is also not good for the field, because such a faculty member is replicating themselves (well, at least producing new PhDs in the field), which increases the pressure on the already over-subscribed traditional funding sources for the field. It is probably only acceptable to have so many PhD students if you know that most of your students will leave the field and not pursue research careers. In this case, under-advising and over-producing them is fine. But, even still, for the few grad students in your group who want to continue as a researcher in the field, life in a huge group might not be the optimal situation.
I will probably have a student graduate this year, and I will probably take a new PhD student next fall. Somewhere between 2 and 4 grad students seems like a good number for me.