Scrivner Reference Manager debacle

Hey folks,

This is going here as a warning. To start with I am a developer with 20 years of experience. I say that not to boast, but to warn. I’m hacky. My tolerance for half working freeware is higher than most. I’m also an academic in CS.

Here’s the deal. It’s a lie you can get Zotero and Scrivner to generate a bibliography reliably on Windows. If you just need inline citations, the rtf/odt scan plugin method will work ok. I figured out a super convoluted hack a few months ago, and now it doesn’t work at all.

Here be dragons. If you’re publishing for IEEE, just don’t. If you manage to prove me wrong and make a video that works on windows consistently by going through something like Libra Office, I am sure others will be thankful.

I have wasted about 40 hours of my life on this because I love this application. L&L seem unwilling to support a reference manager internally. Clearly, folks like us - aka STEM academics, are not their focus.

What can you do? I say this without any irony and know my brand loyalty here is Linux and PC. Use a Mac. On a Mac, Bookeneds will generate a bibliography in IEEE format. You will not even need the reference documentation to work out how if you’ve tried the above process before. It’s obvious and painless.

I could cry. So much time wasted. This is fine. I get a Mac from work so I’ll just use it to finish up PhD and for papers if I use Scrivener again.

Let this be a warning and a request in one. Literature & Latte, know we exist. Just like humanities nerds, we love your tool. Please spend some free minutes thinking about how you could make this work or formally recommend a way it could work for Windows and Linux so we can sanely publish in IEEE and ACM format without losing our minds or giving in and switching platforms.

Windows folks in STEM, run screaming back to Word or Overleaf. This is a nightmare.

Thanks in advance,

With all respect, I don’t quite understand how this is a problem with Scrivener? Temporary citations are plain text abstract tags and are thus citation-style agnostic; it is the responsibility of the formatting plugin (i.e. Zotero in your case) to handle the formatting. I don’t think this is platform specific (unless Zotero has a bug only on Windows?) Are you sure you downloaded the correct CSL file:

I think L&L has been clear about the process: use temporary citations and a post-processor (Endnote, Bookends, Zotero etc.) — that works just fine for many academic users in STEM fields (your assumption only people in humanities use Scrivener is misplaced). CSL should handle most STEM formats (APA,MLA,IEEE) as these are not so complex (unlike some legal formats which are a nightmare for any bibliographic software). If Zotero is buggy, then consider using Endnote (which your University may have a subscription for already):

I personally don’t like having to do a manual post-scan, and so use Pandoc with citeproc to generate my bibliographies automatically from a Scrivener compile via a script (it uses the same CSL files as Zotero); if you are technically competent it is a really nice way to generate well formatted documents[1] automagically from Scrivener.

[1] Pandoc processing works great on Windows especially now there is WSL/scoop and other command-line tools, the only current caveat is Scrivener 3 beta post-processing pane is not yet functional AFAIK, so you run pandoc manually in this case.

I double dog dare you to try and get this to work with Zotero on Windows. Go ahead… I’m here all day. It will not happen. As I said above, you have to use Bookeneds (or Endnote, but it costs more). The problem here isn’t that L&L isn’t clear, it’s that they just don’t have some inbuilt citation system like Word. The post processing process is broken at present for the latest version of Windows w/Zotero to generate a bibliography. You can get it to handle inline citations, but from there, it’s a nightmare. You might be able to put them all inline and the generate one in Word? Not sure. I didn’t try that but my gut says no from the last time I put citations into Word’s system (it was 2012).

The issue here is just this entire workflow is convoluted. It’s not at all straight forward. What I want is this - to be extremely obvious - I want them to possibly think about bulding plugins to make zotero and mendely work INSIDE of Scrivener. That would be amazing and it would avoid needing to go to github, download an open source plugin, rely on said open source plugin to actually work properlly, and then run it through Libra Office. It’s just a massive pain.

Does that make my above clear? If you are still lost, maybe go try it yourself. Pandoc is something folks have mentioned to me. I’d be more likely to just you latex and give up and hand cite.

If Scrivener actually does what I suggest, my guess is that they’d see a rise in sales because it’s convoluted and a bad user experience full stop. If I hadn’t paid for it, I would have gone back to word or done it all on overleaf.

That’s just my own experience. Like I said, give it a go and see if you can make it function, because {biblography} should do it but it’s borked.

Well perhaps other Zotero users can chime in, I am a macOS user content with Bookends+Scrivener+Pandoc for writing scientific papers. My point stands, this is a Zotero bug and Scrivener works exactly as advertised (it provides a mechanism to talk to a reference manager to insert temporary citations, and formatting relies on the quality of the reference manager tool itself, which in Zotero’s case appears insufficient for a basic workflow). A “debacle” infers some real disaster of Scrivener’s own making, whereas your underlying intent is basically a feature request (i.e. this is a wish list item, not technical support).

From what I’ve gathered from the several threads on this before, to rearchitect Scrivener’s text engine to be “extensible” is a huge undertaking. So what you are advocating for is not feasible without a major version that totally strips out the RTF engine for something completely new. In an ideal world, I agree that Scrivener would have some sort of API and extension mechanism, and we could write plugins, no one has argued against this in theory, just that in practice it is a major undertaking and its development priorities must compete against all the others. KB is currently cogitating on the next major steps, and I’m sure some of this will make up part of his decision.

I have made a wishlist request to include a cite-proc engine as part of the compile process, this would enable Scrivener to generate full bibliographies without the difficulties of having to rearchitect the whole text engine (or build in a reference manager system, you would still use your preferred ref manager). But this wish depends on being technically able to embed such an engine, and the best engine (rust-based) is currently in a state of developmental flux… An easy solution would be to just embed Pandoc itself, but KB thinks this dependency is too large (or he did the last time this was discussed)…