What is the actual work of a PI? Lots of stuff.

Recently Michael Eisen (lab, blog, twitter) tweeted:

are there any studies of how PIs apportion time between grants, teaching, papers, writing recs, meetings and actual work?

To which I reacted:

.@mbeisen With due respect, I think you are mistaken about what your "real job" is. Teaching major responsability. & planning advising etc.

A small dialogue ensued:

@marc_rr and you know what my real job is because...?
@mbeisen let's say the job of a PI who's a university prof. 140 char lack nuance but excluding teaching from real work seems shocking to me
@marc_rr i am paid to do research, my teaching is voluntary
@marc_rr and that is not me disrespecting teaching - into which i put a shit-ton of time - just a fact of where my money comes from
@mbeisen OK I see. The original tweet said "PI" not "Mike Eisen". I understand your position now. Still not 100% clear about "real work".
@mbeisen and meetings are often an important real work of a PI IMO: share info, make decisions etc

So first a clarification: Michael Eisen is an HHMI investigator. Which as I understand it, means that he doesn't need to teach. So super cool that he does!

Second, I still take issue with the exclusion of "grants, teaching, papers, writing recs, meetings" from "actual work".

The question is, what is the actual work of a PI? Situations can varry widely, so I'll take a common case for now of a Professor who's also a PI, has a lab with grad students (known in Europe as PhD students) and postdocs. Presumably, if you're a Professor you are supposed to teach (but see above), and presumably if you have a lab with students and postdocs you have soft money grants. So what should you be doing?

• Advising and managing your students and postdocs. Depending on their number and degree of autonomy, this can take quite a bit of your time. In my opinion, a very legitimate part of your work.
• Often related to the above, writing and correcting papers, managing the review process (what revisions do you do? which ones do you try to argue away? where do you (re)submit?). Also legitimate.
• Since the students and postdocs are paid, you need to get grants, which means that you need to write them. The legitimacy of this taking up your time is for me a question of degree. I don't think that we would be justified giving grants without asking for a project and reviewing it, and actually thinking about your plans for the next years, how they relate to the competition, how feasible they are, and how they make for not-too-bad PhD projects is important and useful. On the other hand, I can understand the frustration of my US colleagues who need to write an awful number of grants to get a few funded. Moreover, various agencies from the NIH to the EU FP programs ask for all kind of non scientific stuff which is mostly bureaucratic and of dubious use. So I'd say that in principle writing grants is a legitimate part of a PI's work, but that if it takes too much of our time the system is broken (success rate too low and/or too many useless requirements).
• Writing recommendation letters and generally following up on people who worked with you. Seems legit to me.
• Keep up with the literature. Obviously.
• Also obviously (I hope) reviewing other peoples papers and grants.
• Teaching. If you're a Professor at a University, it seems quite legitimate to spend some time on what you're paid for. But further than that, I think that teaching, from undergraduate (Bachelor) to summer schools for PhD students, is a very important responsability, and that we should try to do it well. Which means spending more time preparing classes and interacting with students, and can also mean reading up or going to lectures about pedagogy. Very legitimate.
• Meetings. Ah-ha, no one likes meetings, right? Yet every meeting was called by someone, and others agreed to come. Fact is, meetings can be quite useful and frankly indispensable. You want to coordinate the teaching between Professors to provide a consistent learning experience which meets your goals? You want to decide on the future directions of your department? You want to launch a new interdisciplinary effort? You want to organize a conference? All these things need meetings. So it's true that a mis-managed or mis-called meeting is a waste of time, but so is debugging code. You don't get to do it 100% right. Like grant writing, a question of degree then.
• If you have time, as a Professor you can even do some hands-on research yourself! But in my experience, it would be unrealistic to expect this to be more than 5-20% of your time, unless you've decided on a very small group strategy (i.e., you and one student).
• OK, there's one thing which I have to do and which would not be part of my job in an ideal world: dealing with the ever changing rules of my central administration concerning refunds of expenses and human resources. For this, you need a full time competent admin (disclaimer: I have a competent admin whom I share with 7 or 8 other group leaders).

In summary, I think that there are many aspects to a Professor / PI / group leader's job, and that most of these can be considered "actual work".

We are no longer postdocs (see also this interesting discussion)...