One of my most talented students is about to graduate this summer. He has done some quite difficult work and will have, if all goes as planned, his three first-author papers and a couple of second-author ones. So, even from the bean-counting standpoint, he's done a respectable job (albeit not as earth-shattering as I think his talent would potentially allow... But I have written before about him not listening to me, so I won't repeat myself here.)
He joined my group 4.5 years ago. In our field, 4-5 years for a PhD is expected, with anything over 6 years being uncommon. I generally talk early and often with each student about what needs to be done for them to graduate, and their eventual graduation date usually reflects a balance between when they would like to defend and when I think they have done enough. I am not one to hold on to students indefinitely, and as long as they have published at least three papers from the dissertation, I will not object to them defending. So far I haven't had to bend the three paper rule, but I imagine I probably might have sometime in the future. (Although I don't imagine ever being comfortable with a student having zero publications from a dissertation.)
Anyway, what I have been thinking about is how the relatively short duration of the PhD means that one student never does get to fully explore a rich, interesting research topic. Think about it: for the first 2 years in graduate school (give or take a semester, I suppose the load depends on the university and the major), students are mostly focused on coursework and make progress in research largely during the summer. Progress is initially slow, but then in years 3 and 4 it becomes more rapid. By the time the student is fully mature and capable of appreciating the bigger picture as well as the technical nuances, it is time for them to graduate. From a purely research standpoint, this is really a shame, as they abandon their project when they finally have the expertise to tackle the most interesting and difficult open questions. If the student were to stay for another year or two, they could really cash in on their expertise and produce some quality, high-impact publications. But, this is where the students invariably can't wait to get out.
There is always the option of staying as a postdoc for a year or two after graduation, but this is not the best thing for most people. In my experience, people may stay for a short postdoc (~ 6 months) to wrap things up, and this stint doesn't adversely affect one's CV. But I have to say that I haven't seen many people who stay at the same institution past a few months, and the few cases I have seen have only done so because they have two-body issues (waiting for a spouse to graduate, spouse has a good job and won't leave area, etc.) Most people still expect to see a postdoc from an institution different from one's PhD alma mater.
I was a student not that long ago, and I understand how it feels to want to graduate and just get out of there. You are sick of your project(s) and just want to do something else. Live somewhere else. Make a little or a lot more money. Also, being able to say that one graduated in 4-5 years as opposed to 6-7 is certainly a worthy consideration, perhaps one that surpasses a few extra publications of the CV/resume for most people, especially those with nonacademic aspirations. There is no doubt that there are many good reasons for people to want to get out as fast as they can and take up a position (a postdoc, a job) elsewhere, so I have no intention of holding back people who have done enough for a PhD.
What I am interested in here is what happens to the research project they leave behind. I am not talking about projects that fail -- rather, I am talking about a successful research project that has perhaps after a 3-year grant now opened new and exciting vistas for your group to explore. We are teachers, we train people en route to advanced degrees, and have obligations to them, but we also have obligations to the federal agencies and the public to advance the science and technology through research. Often, I feel these two facets of a professor's job are actually in collision with one another...
Most of us already run pretty lean operations -- it would be nice if I could fund three students to work on the same subset of problems at any given point in time, with their seniority staggered, but it's very hard to have so much redundancy in funding, at least for a theorist like me, working in an applied physical science field. In my group, everyone has their own main project, as well as a couple of satellite ones. There may be some shared expertise, e.g. a postdoc may offer some expertise to a graduate student and be a second author on the student's papers, but each group member knows on which projects they are the lead and on which ones they have a secondary role. (I have never had squabbles over authorship order.)
In my field, it is not the norm to have multiple postdocs; you may have one postdoc, maybe two. Only the very large experimental groups have permanent research staff; I don't, and most of the colleagues in my department and closely related ones don't. Students are the generally the owners of their own projects. When a student graduates, usually there is a new student continuing the work, but such a student is a novice and needs a fair bit of time to ramp up. Even if you hire a postdoc to continue the work, that person will need ramp-up time, maybe not as much as a young student, but still... From a purely research point of view, the loss of momentum that accompanies these transitions is the norm rather than an exception.
The situation becomes pretty difficult when there is a special type of expertise needed and the person who has it leaves. It can take a very long time to find someone with the needed aptitude, train them, and have them take over. I am not sure how often this happens in experimental sciences, but I certainly have projects that a number of students could do, and then there are those that only select students with a certain kind of talent can do. Ensuring continuation on the latter kind of project can be a real challenge.
What are your experiences in terms of maximizing momentum in a specific research direction in the light of invariable student graduations or other personnel turnover? How much staffing redundancy is common on a project in your field? How much do funding issues affect continuity in your group? Do you feel like there are sufficient options for renewal funding beyond the standard 3-year grants (don't know what NIH R01 duration is, but NSF and DOE and many DoD grants are 3 years)? How much redundancy is fair to students and postdocs (i.e. each one having own project as opposed to there being uncomfortably great overlap, where your group members are competing among themselves) provided that funding is not an issue?