Critique a potential workflow for writing a scientific book (DT3 + Overleaf/Latex vs Scrivener)

I’d like some honest feedback. Early this year I did a small (~90 pps) open source book (“Doing Engineering Calculations” if anyone is interested). I did the bulk of the writing in Scrivener then took the Latex output and did the final polishing in Latex (with Zotero as my reference manager). I was less than pleased with the amount of final polishing I needed to do.

I’m about to embark on a much larger book project, and what I’m thinking is storing all the research notes, snippets, and fragments in DTP then doing all the writing directly in Overleaf/Latex. Probably whatever drafting I do in DTP would be in Markdown to allow export (via Pandoc or whatever(*)). I would mirror the chapter structure in DT3 to correspond to chapters in the book. This would take Scrivener out of the picture and allow better customization of Latex (including whatever templates the ultimate publisher I agree to go with may use).

I wonder if there are serious technical book authors in this community who have used such a workflow, and whether anyone has any thoughts I might have neglected to consider.

Thank you all.

(*) I also just discovered a Latex package that will allow “native” markdown to be embedded into a document and converted by the Latex engine itself - though I haven’t tested its robustness.


What is LaTeX doing in all that? Or rather: why do you need it, and if you do, why don’t you use it directly? I think that more intermediate steps create more friction.

And MD as a precursor to LaTeX sounds like a very sophisticated way to point a gun at your foot.

Good questions. WIth a book with lots of figures, tables and equations, Latex is a better platform IMHO than e.g. Word. Also, early in my career I wrote a book in (the late lamented) WordPerfect - and when I came back to revise it about a decade later most of the files were unusable or only partially useful. So I have resolved to do all further major work in formats that cannot be deprecated by the whims of a particular vendor.

I have been doing technical presentations and papers, as well as preparing lectures, using Latex for probably a decade now.

For some simpler writing (without the equations, tables and figures noted above), IMO MD drafting can be done fairly quickly on multiple platforms (including DTTG as well as Obsidian, for example).

1 Like

I apologize if I left the impression that I have anything against LaTeX. Though I never got around to liking it, I’d probably have benefitted a lot from it, had it been around at my university time.

My question was rather: If you’re using LaTeX in the end anyway (and I understand that you prefer that over Word or Pages or whatever), why not use it from the beginning? Of course, I have no idea of the state of affairs on MacOS nowadays, eg if there are simple to use and comfortable LaTeX environments. But in my mind, MD relates to LaTeX like bulldozer to pincers. If you need the latter, it might be better not to run over your target with the former (to strain the analogy).

Overleaf is a very nice cloud based Latex environment (which supports backing up to dropbox and github). Basically my idea was to use DT to replace the Research portion of the scrivener binder, to collect resources, ideas etc. before incorporation into the draft per se.

What I do when doing major writing projects with Scrivener and LaTeX.

I have used LaTeX for twenty years, written several books with it, and at some point also thought I could work with Markdown. My efforts included one book, for which I accumulated all material in DT. The phase of collecting material and notes is obviously a fantastic experience in DT. Once you have more maths, tables, references, there is a point where working with markdown just did not make sense (for me). Pandoc workflows were not working well for me. I therefore now separate the collection of material and the write up. On Mac and iPad I use Texifier (with there own compiler failing with Biber currently), or TexStudio. Grammarly works in Texifier but not in Texstudio, Texstudio allows for search across files, … lots of subtle differences, pros and cons. Zotero is my reference manager of choices (after various others being used for some years) I love the coarkboard view of notes in Scrivener but for texts with lots of math, tables, references, I gave up on Scrivener.


DT is excellent for this, gathering and organizing material.

If you want something new you can use Quarto

This is markdown flavour that work nicely with Python/R, basicially an superset of pandoc (you can use VSCode to write code or even write text directly as markdown in indexed folder in Obsidian/Devonthink ); You could use there nice latex templates (however sometimes it needs a bit of tweaking), I courrently adapted a template and write my PhD thesis using this workflow

A nice tool for picking citations here

And this thread contains interesting ideas

1 Like

I use mix of markdown and latex templates - when I more complex tables or page structure I just include raw latex code;

When I need to work with just LaTeX (for example when creating poster) I use LaTeXStudio in VSCode (I used Texmaker for my master thesis but the LaTeX plugin in Vscode is more convenient for me now)

I have been using TeXStudio on macOS for many years. A good, complete, environment. In most areas of physics it is de rigueur to write your dissertation in LaTeX. The ones written in Word never look exactly right. A typical dissertation is not fundamentally different in scope than a typical book, so this should work equally well. For equation-heavy contents, there just seems to be no alternative that comes close. I’d be wary of switching formats somewhere on the way, and relying on tools like Pandoc to convert reliably. I can sympathize with the idea of avoiding some of the verbose begin/end syntax by using shortcuts such as a markdown package. Depending on the details, this could be useful, or badly backfire, as @chrillek points out.

In terms of LaTeX environments, I would argue against Overleaf unless you heavily collaborate on the project. Its tools (palettes, forward/backward jumping between source and render, etc.) are primitive compared to any of the local environments such as TeXStudio, TeXMaker, TeXShop. I find the lack of tools so annoying and limiting that even on collaborative projects, I tend to issue a section-freeze to everyone, edit that section in TeXStudio, and then upload it again. To me, Overleaf is a very fine example of why web apps still suck big time. But as a collaborative project container and quick editor, it’s of course great.

So my suggestion is to write the book document from scratch and all the way in LaTeX. Use DT for the research and information collection. Build a good reference BibTeX database, and use BibDesk to manage it (just make a single one for all references and reuse it for all your publications, unless you work in multiple, largely disjoint fields).

Thanks. I have a lot of experience with the Mac Latex clients as well (I prefer TeXStudio). I’ve pretty much decided on a DT+Latex workflow and avoid Scrivener. Yes about BibDesk. It is a piece of cake to copy refs from Zotero into BibDesk as needed. My interests are quite heterogenous, so I generally do a BibDesk for each Latex project (my Zotero database has over 7000 entries, most with pdfs).

Since I have an Overleaf subscription and I save the files in Dropbox, it is pretty easy to go back and forth between the Overleaf and TeXStudio environments.

Thank you all.