Dear LL Team,
Several times I started to try using Scrivener for my documents. I am honest, I really give it a try. And every time I stopped after short due to the lack of integration e.g. of Zotero.
I follow the forum quite a time.
What I definitely don’t understand ist the denial of the need? Your user base could be so much larger. Is it just a kind of philosophy?
And, as a Bookends user, I don’t understand why Zotero users have trouble using temporary citation links and scanning after compile.
Mark
And I agree with Mark. I really don’t understand what the problem is. I wrote an 85,000 word doctoral thesis using Scrivener and Bookends, and never had a single problem with citations and referencing. It’s perfectly easy. Nothing more is needed.
People making this request are often not very clear about exactly what they are requesting. What does “Zotero integration” actually mean?
Also, Scrivener’s current approach, based on temporary citations from the tool of the user’s choice, has the very significant advantage that it is tool-agnostic. It doesn’t require Scrivener to choose the “best” bibliography management software, but leaves that decision to our users. Since software development resources are finite, closer integration with Zotero (or any other tool) necessarily reduces the resources available for improvements to Scrivener-specific functionality.
Katherine
Have you asked Zotero to make integration with Scrivener instead?
This is what I mean, instead of saying “yes, could be an idea”, it is tried to get the responsibility on another side.
The question is: which is the leading system? Is it Zotero? No. Zotero is a supporting System. The main goal is to write a document. Therefore Scrivener is the leading system. Easy logic.
No, writing is just one step in the process. Searching, collecting, organizing, reading and annotating articles is just as important a step. And you seem to forget that Zotero and Mendeley is only two reference handlers. There are others, and some I find much better, like Papers 3 and Bookends. Not to mention Endnote. Do you seriously expect L&L to choose one of all the available reference softwares and dump the others?
And as the others have written in their answers, you can already use temporary citations in Scrivener and transform them into correct ones when you are done writing.
Have you heard about the software Manuscripts? It was an attempt to write a Scrivener clone that would be integrated with a reference software. It never got past the first super buggy phase and was finally abandoned.
In most cases, the supporting system is the one that is expected to do the work. If you want to write a plug-in for Microsoft Word, say, you’re responsible for making your code work with Microsoft’s APIs. Same with third-party tools for Photoshop, browser extensions, etc.
Katherine
This, as I understand it, is one of the key reasons why many feel Scrivener isn’t the right place to have deep reference manager integration – it’s not, for large and complex documents, where you are doing your final layout. Scrivener is where you do your research and writing. The compile step is fairly unique to Scrivener’s approach and it’s representative of a mindset that’s different from pretty much all other software out there on the market.
For simple projects, you may be able to compile and forget, but for larger projects, you probably need to pour the compiled document into some other program where you do your layout. That’s the program where you want your reference managers to integrate and do their final work – because that’s where those details become critically important to your workflow and final product.
This, I think, is the nub of the issue. There have been whole threads in which I’ve asked users what “integration” would look like, and yet no one has been able to tell me. It has hardly been that we are “denying the need”! I have asked users who want this what this would look like to them, but beyond vague words like “integration” they cannot say or disagree (we had the same problem with requests for Evernote integration aeons ago).
Other posts here also make good points: as a single developer with limited resources, it is not sustainable for me to write thousands of lines of code, and then maintain them, for each different reference manager. (It’s also not possible to build our own built-in references manager, I hasten to add. We’re not Microsoft, and outside of MS there’s a good reason why reference managers tend to be an app in themselves.)
So, I ask again, what would integration look like? (That’s the first question. After that, there’s the question of whether the reference manages have APIs available that support that integration, and after that the question of how feasible it is to do such development on our end.) Where, for instance, would the bibliography be maintained live in Scrivener, and how would the citations manager know about it? How would this work with Scrivener’s footnote streams? How would this work across the many different files within a Scrivener project, and then how would all of this be maintained and work throughout the Compile process? And how would any of this work better than the temporary citations system, which has always worked well for a great many users? These are genuine questions.
All the best,
Keith
I think it’s our old friend the False Consensus Effect appearing for the nth time. “I would like this, so everybody else must need it, too.” One day someone will do a serious psychological study of the incidence of the phenomenon on the internet. But no criticism implied. I had no idea that such a cognitive bias existed a few years ago.
Interesting answers, thanks a lot for all your comments!
My understanding of the leading and supporting system is as following:
The leading system - in this case Scrivener (which btw is a real good system from my point of view - needs to “open” doors" for plugins. The better this works, the more plugins you will have, supporting additional - non core - tasks.
If you look on other software for other purposes, those systems with an easy expandability by plugins have been the most beloved ones and by that the most successful.
Hi manni27,
It seems to me that you have ignored KB’s post above. In case you don’t realize it, he’s the founder of Scrivener: the lead developer, the main man, the head honcho. The big cheese. El jefe. The big Kahauna. Numero uno.
Keith’s the decision maker. He’s is the one you need to convince. Other posters have made a similar request to yours here and there. He’s asked them questions about what it is they’re really looking for, to which they never supplied answers. As a consequence, he’s never provided the solution they requested.
Now he’s asked you the same specific questions he’s asked the others, and it seems you are likewise declining to answer. So don’t be surprised when he likewise declines to provide the solution you’re requesting.
Best,
Jim
Hey Jim,
Thanks a lot! 
Your comment made my day today (I was not in the mood so far). I see I have to invest all my persuasion abilities to convince Keith.
Manni
Hello Keith,
First of all thanks a lot for your reply. I definitely understand the issue of having limited resources for developing. And honestly as a person also coding (if I can find the time to do so) I have to say that you really really do a good job! Thanks for this great tool.
What I would really be happy about is a functionality within Scrivener of being able to include a bibliography like the Zotero plugin does for MS Word or like BibTeX and referencing the single content of the bibliography in the rest of the document in an easy way.
As a background, I am actually writing my dissertation. Therefore I would like to use Scrivener. MS Word is just getting slower page by page. And I don’t really want to use LaTeX, I want to invest the time I would have to for LaTeX into my own document content.
Maybe I don’t know Scrivener so well right now, but so far I did not find out how to fulfill this. Expecting from Scrivener a functionality like Zotero would go too far, therefore my assumption was maybe an integration with a plugin.
Best regards,
Manni
The main question is still unanswered. You still haven’t described exactly how you expect this integration to work. Give examples, step by step.
And why can’t you use temporary citations, i.e. insert cite keys, which are then converted to full-scale references after you have compiled to final output?
I made a wishlist post with a detailed and potentially actionable solution here: Academic Bibliographies: support citeproc as an option during compile
This would use temporary citations and a BibTeX source generated by Bookends/Zotero etc. — During Compile, citeproc does the heavy lifting of turning temporary citations into formatted ones + a bibliography using one of thousands of pre-existing CSL styles. The job of curating and maintaining references would remain squarely with the dedicated software that is best at dealing with that.
KB would need to integrate a citeproc engine, just as he does for e.g. MMD or Java convertors already (i.e. the underlying idea is to use the compiler workflow that takes advantage of already existing and widely used tools). The problems as you will see from that thread (apart from the philosophical issues), are that almost all citeproc engines are scripting-language based, so KB would need to do something like embed a Node server or rely on the OS-provided runtime. The one engine that is being designed for flexible binary embedding is rust-based, but that is still in development, and I doubt KB wants to deal with a potentially moving target…
If you use Pandoc, this is handled as a filter for you automagically. But if it was in Scrivener’s compile, it would allow all users (perhaps many for whom Pandoc is too geeky and complicated to deal with) to generate any output that benefits from academic bibliography generation, without having to manually post-process RTF files.
In the meanwhile Manni, why exactly can’t you get temporary citations + manual scanning to work? It works just fine for other users.
nontropo, thanks a lot for your comment. I will have to check the temporary citations and manual scanning in detail.
@manni27 There isn’t really much detail to study. Typically (there is not much variation among the programs I have used for this) you type (or copy and paste) a temporary citation in your text or footnote. This will look something like {Smith, 2020, #123@10} – in other words, the author, the date, an ID number, and after the ampersand, the page number (page number is usually optional). BibTex is a little different, but not that much. Some programs, like Bookends, allow text to be prepended or appended to the temporary citation before insertion. If you specify the bibliographic manager in Scrivener’s preferences, all you need to do while writing is to hit Command-Y and you will swap to the bibliographic manager, where you can find your reference, and another keystroke will take you back to your text and insert the temporary citation.
If I know I am going to be inserting references to the same work repeatedly (not uncommon in the work I do) I will copy the temporary citation, and just paste it in when I need it, without switching to Bookends. This can be helped by using a clipboard manager like Copied, which stores previously copied items. If I know a particular reference is going to crop up repeatedly in my work over an extended period of time (e.g. a standard text in my field) I set up a text substitution. So if I want to cite Freud’s Mourning and Melancholia I just type “jjmourning” and the temporary citation {Freud, 2005, #12682} is automatically inserted into the text for me (I use the prefix “jj” because, as far as I know, there are no words in English that begin with “jj”, and “jj” is in the middle of my keyboard). But of course you can set this up how you like. Among other things, a possible advantage of using temporary citations is that they are ordinary text, so you can search for them using normal find. This might be useful if you want to find how many times you have cited a work, or where you have done so.
When you have a final draft, you compile to rtf, and (in my case in Bookends) you scan the rtf file, which automatically replaces the temporary citations and creates the bibliography – usually in a matter of seconds (it did with my 85,000-word doctoral thesis). I find it difficult to imagine anything much simpler.
In days gone by I did use Word’s “Cite While You Write” feature. But only for about thirty minutes. It was dreadful. Every time you inserted a citation it had to create a field, update the whole document, reformat the bibliography (including sorting all the entries), and autosave. For a single-page document it might have been bearable, but I cannot imagine what it would have been like to do a longer piece of work with hundreds of pages and hundreds of references. I believe experiences of this kind are why some of us are highly sceptical of “plug-ins” that are supposed to do various things “magically and without effort” for the user. Often, they do not work very well. Like so many things they hold out great promise and seem very attractive before you use them seriously. Then when you use them, you find out what the disadvantages are. Simple solutions like temporary citations are fairly bullet-proof, and they are also portable and durable. I was able to take texts I had written in 1995 using temporary citations created with Endnote, and scan them with Bookends twenty-five years later. That matters when you are building up a body of research material and you are not sure what the future may bring.
Thanks a lot for this “quick user guide”! On Friday this week I have to deliver my intermediate results, so this weekend I will definitely try out what you described. I hope it works, though MS Word is getting a nightmare… I will return on you.