The most common reason that makes me delay returning nearly finalized manuscript drafts to students is the missing/inadequate big picture coupled with painful underciting. I edit the text mercilessly and extensively; generally, on the technical parts of the manuscripts, edits do not take me too long. What does take long is to create a proper and compelling introduction with adequate coverage of literature and the exact positioning of our contribution within the context of the state of the art.
Ideally, the big picture -- why we do what we do and what it is that we did that was novel and why anyone should really care -- gets visited repeatedly during the course of the project, well before there are publication-worthy data. The student reads many, many papers, we talk extensively, investigate different approaches, weigh the pros and cons, take detours and go on tangents, revisit the issues many times... The big picture should never be lost in the advising process and I think that faculty, regardless of discipline, will agree with me on this issue. I think we would all like to think that we try to instill in our students the birds view of the research along with the ability to look into the nitty-gritty details. One without the other is not good training.
But then comes the manuscript drafting part. And this is what baffles me -- even my best students, who are technically stellar and I know have read lots of relevant literature -- give me initial drafts with a measly single paragraph "background", in which barely 10 references have been thrown together to supposedly introduce the work, and with generic sweeping statements that beg for a reference but nothing is cited (we're talking comprehensive papers here, so the number of citations is not constrained). As though it is too tedious to position oneself, and the student can't wait to get to the fun stuff, i.e. their own technical contribution. We have a really big problem when the student really does not understand what the big picture is, even though they might have executed a technical task; this is indeed an advising fail. But I see this sloppy underciting and lackluster introductions even from very good students and often postdocs.
That's where I feel like I am picking up dirty socks and generally cleaning up after my kids. I am the maid whose duty is to tidy up, and I go and do a thorough literature survey to make sure we didn't miss anything among the new developments, that we have paid dues to the important contributions of yore, and to actually explain why we bothered with the work.
Now, I have weekly group meetings and we ofen talk about strategies for writing better papers. I am sure I am totally annoying as I constantly repeat my spiel on the importance of citing and proper motivation. We also often talk about efficient strategies for quickly mapping out the state of the art in a field, through identifying key papers and then following their citation branches, while weighing the outcomes with the offspring papers' age, citations count, prestige of the journal, quality of the group, relevance to our own work, etc. Generally, before a student ever sits down to create the first draft, we talk about the outline -- what's the paper about, what is its message, why is it important? Yet, this particular part -- write thy introduction well and with ample citations -- does not seem to stick. Why, I wonder? Is it because they don't care as they know I will pick up the slack anyway?
I don't remember any formal mentoring from my PhD advisor on how a paper is supposed to be structured, or what it means to write a good paper. As I student, I always thought you just read lots and lots of papers and patterns start to emerge -- the good papers give you a feel for the state of the art and open problems, and then tell you which open problem they solved and how. A good introduction is like a vortex: it starts from a broad view of the field, then narrows it down seamlessly to important and open problems, so that by the time the reader is hit on the head with the "In this paper, we..." hammer, there is no doubt in the reader's mind that what is being presented is new and extremely important.
So yes, even after multiple back-and-forths editing a manuscript with a student and major rewrites of the whole text, in order for the process to ever converge I still end up doing a lot of time-consuming clean-up: a full literature check and multiple overhauls of the intro and abstract in the final version. Because, as the lead PI, if the paper sucks,it's my reputation on the line, not that of the student.
Are there any good tips on training one's students to (a) not be too lazy to look up references, (b) not be careless about putting in the references that they are actually aware of and have read and used in research, (c) try to appreciate the importance of a good introduction, (d) try to actually write a good introduction themselves as opposed to fiddle with it pro forma through multiple revisions and essentially wait for the advisor to do it?