It is two days before the deadline. Your advisor sends paper_final_v7.docx. Your co-author sends paper_final_v7_REAL.docx. Both contain edits the other one lacks, and the results section exists in three different versions.
If that sounds familiar, the problem is not your co-authors. Collaborative research writing breaks because most papers are written with tools built for one person, then passed around like a baton.
Three failures cause most of the pain:
- Competing copies. Email attachments create a new "latest version" every time someone hits save.
- Silent overwrites. Two people edit the same section, and one person's work disappears without anyone noticing.
- Unclear ownership. Nobody knows who is responsible for the related-work section, so it gets written twice or not at all.
The fix is a small set of habits plus one shared workspace. Here are the habits first.
1. Keep one source of truth
Pick a single place where the current draft lives and ban every other copy. If a file is attached to an email, it is a snapshot, not the paper. A cloud editor that every author opens in the browser removes the question of which version is real.
2. Assign section owners before anyone writes
Write the outline together, then put one name next to each section: abstract, introduction, related work, method, experiments, discussion. Owners write the first draft; everyone else comments. This single step prevents most duplicated work.
3. Comment instead of rewriting
When you disagree with a sentence in someone else's section, leave a comment that says what and why. Rewrite only after the owner agrees. Comments keep a record of reasoning, and they stop the quiet edit wars that sour co-author relationships.
4. Use real-time editing for the final stretch
In the last 48 hours, several people often need to touch the same file. Real-time collaboration means you see each other's cursors and changes as they happen, so merges stop being a manual chore. Tools built on CRDTs such as Yjs merge concurrent edits automatically.
5. Freeze the macros and the bibliography early
Agree on notation macros (\newcommand), citation style, and a shared .bib file in week one. Changing \mathbf{x} to \boldsymbol{x} across a 9-page paper the night before submission is how typos get born.
6. Compile often, and compile for everyone
A draft that compiles on one person's laptop but not on another's is a trap. Compile in a shared environment so everyone sees the same PDF, the same fonts, and the same errors. If you hit that wall, my earlier post on 7 advanced LaTeX hacks for AI and CS researchers covers layout and compile-time fixes.
Which setup fits your team?
| Setup | Real-time co-editing | One source of truth | Setup effort |
|---|---|---|---|
| Email + Word/LaTeX files | No | No | None |
| Git + local LaTeX | No (merge by hand) | Yes | High |
| Browser-based LaTeX editor | Yes | Yes | Low |
LetX is the browser-based option I build. It has live PDF preview, SyncTeX navigation, version history and 1000+ templates. The free tier has unlimited projects, unlimited collaborators and 40 compiles per day, and you can import an existing project, including an Overleaf export, as a ZIP. Real-time editing runs on Yjs, so concurrent edits merge instead of overwriting each other, and the built-in AI assistant, Lexi, can propose edits and compile to check them.
Try it on your next paper
You do not need to change everything at once. Start with rules 1 and 2 on your next draft: one shared place for the paper, and a name next to every section.
If you write in LaTeX, import your project at letx.app and invite your co-authors with a link. I read every piece of feedback, so if something feels clumsier than your current setup, tell me in the comments.
What is the worst co-author version mix-up you have lived through, and which rule would have prevented it?
Top comments (0)