How to ask the superviser to not mangle the manuscript?

Question

I gave a LaTeX manuscript file to the supervisor for revision. The paper is fairly long and complex, with lots of cross-referencing and citations. I use BibTeX for managing the otherwise unmanageable task of generating the correct list of references, which is recommended by the journal. Most of the BibTeX entries were generated by querying the ADS database so there wasn't even much manual typing. The journal provides a high-quality .bst file that correctly sorts and formats the BibTeX database entries.

Before I gave it to the supervisor, I told him that he's free to come to me if there should be any technical difficulties such as BibTeX. He didn't, so I assumed that everything would be OK.

It was agreed that I'd have the final control of the master copy, and I assumed that meant that he would try to "patch" the manuscript in a way similar to updating and merging a codebase, i.e. you follow the convention of the mainline source code, transforming the project by incremental updates, and only introduce drastic changes in the face of absolute necessity.

It turned out that he's mangled the manuscript file in a combination of creative ways:

1. He seems to have manually converted the BibTeX-generated reference list to his manually typed (based on his own understanding of the style), embedded \bibitem entries, which contain errors.
2. The ADS bibcode keys, as unique identifiers for reference items in the BibTeX file and citation keys in the manuscript (using natbib's \citep and \citet mechanism) were converted to his own mnemonic keys based on no apparent rules.
3. He also added his own references, again as manually typed text, that are incomplete and not necessarily correctly-formatted.
4. He has mangled the \ref and \label mechanism by stripping quite a lot of the labels arbitrarily, so that much the cross-referencing no longer makes sense.

I have to write several text-transformation scripts, at one point interfacing a simple search AI of the ADS web API, attempting to revert the damage he has done to the manuscript source files, but this task is not fully automate-able. It is also rather likely to introduce additional human errors in this process. All these take time.

The point is that, it shouldn't have been necessary had he not mangled my manuscript file -- which was arguably written with maintainability and automate-ability in mind -- in the first place.

Now I have a second manuscript given to him, and I want to ask him for a favor, namely considering not mangling the source file in his way this time.

However, it is complicated by the following considerations:

1. Training him on the use of BibTeX etc is definitely a waste of his time and resources (as well as mine).
2. Imposing source-level conventions and tool-compatibility is a kind of power over others, and I shouldn't be in that power, even if I'm the ultimate controller of the manuscript.
3. It may incur the wrath of The Powers That Be.

So what can I do to reach a compromise?

Do you think it's a good idea to say/write to him like the following?

Hi XXX, thank you for your time revising the manuscript. I am really really grateful for the improvements in the quality of content.

However, for the second manuscript, may I suggest trying BibTeX this time? It saves us the time of manually typing, checking and correcting the bib list. The journal we're targeting already provides high-quality bibliography style file that does the magic. In addition, if we're to change the target journal, it would be simply a matter of changing the .bst file.

You don't even have to type anything to fill the entry fields in the .bib file. There's the ADS service, and each entry is a few clicks away, with some minimal copying/pasting. If you like I can show you this in a 2-minute demo.

If we don't have many references to add, this is a non-issue anyway. If we do, I'm willing to help with the technicalities any time. Of course, you can even forget about BibTeX and directly add your additional references this way, and I'll take care of the merger. It saves your time of manually re-formatting the references that are already in the BibTeX database, and you can better focus on the important science rather than the style.

Thanks!

Answer 1

If your advisor isn't fluent in LaTeX, send a printed copy (or a PDF), to scribble over, and incorporate changes yourself.

Use a version control system (I like git, you might want to try mercurial (hg)), and keep a branch for your work. Create a branch for "external" inputs, and use e.g. git's tools to break up the differences into patches you will apply and stuff you just omit.

With my students the policy is that the work is theirs, I'll write into a branch of the shared repository with suggested changes, which they cherry-pick. Only in rare circumstances will I override their writing.

Answer 2

My PhD supervisor got to the point where he could edit LaTeX files just fine after several gentle reminders not to mess with the backslashified stuff. He'd just work on the large text blocks and drop any equation edits by the lab handwritten on paper. I'd suggest giving your supervisor the both PDF and the source file you want them to look at so that they can either make notes on the PDF or small edits to the .tex files (and hopefully not both). If they happen to find you working diligently to undo their edits to the LaTeX and understand that you shouldn't be wasting time on that, hopefully they get the point.

Answer 3

I myself have very heavily automated systems which I know, but nobody else does. Therefore, all I suggest is to edit the plain LaTeX with simple annotations for citations or other modifications (which I later polish) or annotating the pdf are both viable options.