The .bib errors that quietly wreck a submission
Nobody fails a review because of a typo in their bibliography. What actually happens is smaller and more annoying: a duplicate entry under two different keys, a \cite{} pointing at something that got deleted from the .bib during a rewrite, or a reference sitting in the .bib that never actually got cited in the text. None of these are dramatic. All of them are the kind of thing a reviewer or, worse, a copy editor flags right when you have zero time left to fix it.
The classic way people catch these is by scrolling through the compiled PDF bibliography looking for anything that seems off, or grepping the .tex file for cite keys by hand. Both work, right up until the bibliography has 60+ entries and you've rewritten half the paper three times since the last full read-through.
A few patterns come up constantly enough that they're worth naming:
Duplicate references under different keys happen almost every time a paper gets co-authored across tools — one person adds smith2021 from Google Scholar's export, another adds smith2021nlp from a Zotero sync of the same paper, and now you have the same source twice with slightly different formatting, and nothing catches it unless someone's specifically looking.
Missing references go the other way: a \cite{jones2019} still sitting in the text from an earlier draft, but jones2019 got removed from the .bib at some point because the argument around it changed. LaTeX will happily compile with a [?] in that spot and it's genuinely easy to not notice, especially if you're not looking at every single page of the PDF output.
And then there's the reverse of that — orphans, entries sitting in the .bib that are never cited anywhere. Harmless on its own, but it's usually a sign the manuscript went through more edits than the reference list did, which is worth knowing before a reviewer notices the same thing.
None of this requires judgment or reading comprehension to catch — it's exactly the kind of thing a script should be checking instead of a human skimming a PDF at 1am. Paste the .bib (and the manuscript text, if you want the missing-reference and orphan checks too) into BibCheck and it flags all three categories in one pass, plus an optional check against live databases if you want to confirm the metadata itself is accurate, not just internally consistent.
And if the bibliography is just one item on a longer pre-submission list — page limits, anonymization, venue formatting rules — there's a separate guide for the whole IEEE/ACM checklist.