The Expanding (or Contracting) Grad Universe

If you advise PhD students, do you have an advising philosophy in terms of the scope of projects for students just starting their PhD research? I know there are big labs in which grad students are cogs and do a particular part of a large project involving many people, and I know there are fields in which the PhD research is entirely student-driven, from idea to dissertation. In my field, there is a lot of variability, and this gives some flexibility in advising style. To some extent, the scope of the project can be modified to suit the student's interests and abilities, but, at least for the early phases of the PhD research, the advisor's philosophy about types and scopes of projects is the starting framework.

I ask my question about advising philosophy in that context. I think the question is applicable in cases in which the advisor determines the project (or, at least, most of it) and in cases in which there is more room for the student to take the research in new directions.

Here are two examples of different approaches:

The Contraction Approach: The advisor starts grad students on projects that involve several to many different components, sees how the students do or what particularly grabs their interest, and then the dissertation research proceeds accordingly. In some cases, no contraction of the scope is necessary, but in others, some components of the project are jettisoned. I think this is a reasonable approach as long as what is left still comprises dissertation-level research. Of course, another possible outcome is that the student does not succeed with any component of the research, even when the scope has been reduced, and then everyone has to make a new plan.

Pros: Students are exposed to an array of possibilities and, ideally, one or more will catch their interest and therefore what ends up being the dissertation research is what most interests the student.

Cons: Students can be overwhelmed. Particularly if a student doesn't know how to organize their time to make progress on their research (perhaps while also taking classes and/or working as a teaching assistant) or is easily discouraged by the obstacles that always arise in the course of research, having too many choices can result in no progress on any of the possible research avenues. 

The Expansion Approach: The advisor starts students on one or a few research activities, sees how they do, then the research expands from there depending on student ability and interest. In some cases, the expansion can be rapid and large; in others, not so much, but still represents a widening of the project's scope from the beginning stages. I think this is a reasonable approach as long as the student knows that the research scope must widen from the first small steps, and there is good communication about how much expansion is expected/desirable.

Pros: Students can master some fundamental techniques or concepts before moving on to new or more complex aspects of the research, building a strong foundation for later research.

Cons: Students may be bored by the first research activities, particularly if these mostly involve technical things, and may therefore never become engaged in the overall research. Also, students may feel that the early research activities are "enough", and anything else is "too much". It may not be clear how fast or how much the scope needs to change, so it can be difficult to move the research much beyond the early stages.

I suppose we could also add The Plateau Approach, in which the research activities are pre-determined, and the students just has to start on them, do them, and finish. That isn't necessarily as cog-like as it sounds, but I focused on the other two approaches because those are closer to descriptions of advising methods I have tried.

In theory, the Expansion Approach makes a lot of sense to me, but in practice, I now tend towards the Contraction Approach, even though it has pitfalls as well. I have had too many situations, mostly early in my advising career, in which the Expansion Approach resulted in a student who could never see the "big picture" of the research and who couldn't (or wouldn't) move beyond the early stages of the research. In my field, that's fine for undergrad research or an MS, but not for a PhD.

Of course every student and every project is different, so even if an advisor has an advising philosophy such as one of these (or something else), it's good to have some flexibility. Even so, this brings me back to my original question about whether you, advisors of PhD students, tend to have a particular approach that you favor (because it works? because you think this is how things should work?) when advising PhD students.