It has been my experience that many Ph.D. advisors place little importance on management skill, and that most Ph.D. advisors are poor at management. If the NSF requires that applicants put in a broader interests paragraph on postdoc mentorship then some awardees will be granted permission to use postdocs to get their research accomplished while paying only lip service to educating prospective Ph.D.s instead of "ensur(ing) that research is fully integrated with education so that today's revolutionary work will also be training tomorrow's top scientists and engineers". There are ample examples of laboratories that are quite "top heavy" with more postdocs than students who are funded quite a bit through NSF, even though the emphasis has been on Ph.D. traineeship.