NSF first introduced the Postdoc Mentoring Plan as a supplementary document a few years ago. At the time everyone was all:
LOOK AT ME TYPING A "PLAN"! YES A PLAN!
There was basically no information on what we should be writing and panels had no idea what they should be expecting. It was basically a free-for-all and plans ranged from "Trust me, I do this" to two pages that made it sound like the postdoc would be working 6 jobs at once. In the years since, things have stabilized and there's numerous examples out there, providing guidance to people putting their plan together.
But has it DONE anything? Are NSF postdocs mentored better today than 5 years ago? How would we even know?
Ok, so I'll go on record that I am totally behind the idea and philosophy behind the postdoc mentoring plan. I get it, and I honestly want to put my postdocs in the best place to succeed with what they want to do as a career (which may not be a TT position). I think it's valuable for PIs to think about the training environment they are providing and what alternatives there are.
Do I think the PDMP achieves those goals? Probably not.
Why? Because I think the people who take it seriously are those who take postdoc training seriously in the first place. I think it's easy to toss words on a page that sound great without ever doing a damn thing about it. Most of all, NSF funding being what it is, it is RARE for a postdoc to be present when they mentoring plan is put together. Nearly every PDMP plan I see is either "postdoc TBD" or "potential postdoc X". Having an in-house postdoc who is funded and will transition to the new grant is just hard to do, given the grant cycle and budget limitations of NSF. All that is to say that most postdocs are likely to never even see the mentoring plan submitted for the grant they are paid by.
And what does it matter anyway? There is no possible way I can imagine that NSF could enforce any of it. Unless a PI puts specific assessment goals (useless if you don't have a PDF in-house already) or commits money to some sort of external training, there's no way for NSF to evaluate whether you are doing anything you said you would. It's entirely on faith that merely making you think about it was enough to affect change.
And finally, how would we even know whether this is effective? There is no way to assess the difference in postdoc mentoring without infinite variables. The PDMP is like an untestable hypothesis and we're being told to go along because it probably does something. Maybe.
Again, in a vacuum I think it's a good idea. But supp docs in these proposals continue to multiple faster than deanlet positions. I recently submitted a proposal that required 4 supp docs, at two pages each. That's another half a proposal, if you're counting at home. And with the new Nagoya Protocol going into effect, you can bet anyone collecting samples outside the US on NSF money is about to have some new paperwork. The supp docs continue to multiply, so I don't think it's a terrible thing to ask whether or not those documents are achieving their goal.
In the case of the PDMP, there's no way to answer that. And so we just write them so we can hold it up and say we did something. And that, my friends, is the definition of make-work paperwork.