I have, as most people, I infer, from the useful 2015 post How to not mess up your bibliographies with Bibtex by C. O. Wilke, had a lot of headaches because of Bibtex.

Has there been notable progress since 2015, and, whether or not so, is there useful accessible literature which can help those like me who do not even consistently succeed with compiling Bibtex-files under Natbib?

I am aware of Biblatex, but I have not become convinced that it is a plausible replacement, or pedagogical improvement of Bibtex.

  • 1
    If you don't like biblatex, there is github.com/nzhagen/bibulous. For a while there has been word about MlBibTeX, but I don't think it has gained any traction yet (tex.stackexchange.com/q/334969/35864). I guess the main problem for any "successor" is that academia can be fairly conservative and that publishers generally do not like to experiment with new things if the old workflow still seems to work.
    – moewe
    May 18, 2021 at 6:30
  • 3
    Quite frankly, when I read the Wilke article you refer to I get the idea it is directed at users and how to be consistent in entering your data so Bibtex won't misunderstand you. Not that Bibtex is in any way majorly bad at anything, except at guessing what you want. Which is probably true of all software. The article is biased as well, with things like "Yet we generally never want first names spelled out in a bibliography, they should always be abbreviated to initials only." Which is entirely untrue in my neck of the woods.
    – Plergux
    May 18, 2021 at 6:42
  • 2
    Biblatex is not a replacement for bibtex, but for e.g. natbib. The replacement for bibtex is biber and that is a huge improvement as it can handle unicode and so other languages and scripts than English. Biblatex is a huge improvement too as it clearly separates data and representation and gives you better access to the data. May 18, 2021 at 7:09

1 Answer 1


Most of the linked post, and indeed the question here, is more toward library science than anything technical. That doesn't mean we can't address some parts, but does mean that it's not really fully amendable to programmatic solutions.

The BibTeX database format, whilst simple, is flexible and more importantly it is a database: just because it's plain text, we shouldn't forget that. In particular, that means you need to think about using special tools to managed it, and that managing data quality is part of usage.

There are a number of tools dedicated to working with BibTeX databases: I use JabRef but there are of course others. The key things a dedicated tool can offer include

  • (Semi)automatic import of data using DOI, ISBN, etc.; the quality of such sources varies, but they can save a lot of time
  • Per-field editing/searching (e.g. to normalise journal names)
  • Integrity checking

On the library science side, just because some bibliography style doesn't use say first names or journal issue number doesn't mean that these are not information that can (should) be logged. How much you choose to make a 'full' database will depend on your use cases, mindset, etc. In my own subject area, it's titles that are 'fun': how much formatting do you include in these, as it can make reading them harder in plain text!

On BibTeX keys, the wide use of DOIs means that for journal articles, there is today a (reasonably) good choice for a unique identifier in many cases. Of course, that will depend on how easy-to-read the DOIs you get given are: 10.1002/anie.202100443 is pretty good, 10.1016/j.ica.2021.120283 is OK, 10.1016/s0020-1693(00)80784-7 is awkward, 10.1002/(SICI)1099-0518(19990715)37:14<2413::AID-POLA15>3.0.CO;2-# is totally horrible. Personally, I tend toward keys I can use - as chemists have the last author as the lead, that tends to be 'last-author-surname-plus-year'.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .