I had an interesting conversation today with one of the senior professors in my department. It was in the context of students who have some difficulty in defining a project. What came in to question was the level of independence of Ph.D. students should display with respect to their research. Certainly, by the end of the degree, the Ph.D. student should be fully capable of defining a research direction and carrying out a research program effectively. But what about at the beginning? Should the student be issued a specific problem that the supervisor feels will lead to publishable research? Or should the student be required to define the direction all on her or his own with minimal guidance from the supervisor - a sink or swim approach? I'm curious where other people stand on this.