I’m surprised you use Zotero for that. For me, it would seem the filesystem (JD-conformant lol) would be a better choice. But if it works then that’s what’s important. I use Zotero for all my citations for my MBA and all my ebooks (and all 9.6GB of attachments).
You mention Calibre, which is a great program, but for one reason or another I don’t really like it much, so I only use it for converting ebooks and for deDRMing my Kindle books. Once they’re converted, they go into Zotero in their appropriate folder.
It’s true, I could use a folder-based hierarchical file system instead of Zotero but it gives me two advantages. Firstly, I store the files on a locally-hosted webdav server that I can securely access from outside. This solves the problem of storing more private information in a cloud. Secondly, it gives me metadata. I would think that’s the value for you and your research citations. Like you I only sync the database and not the files.
Calibre is backed by calibre-web, also locally hosted. I treat ebooks the same as you except I leave them in Calibre. The books and calibre database are locally-hosted on my NAS. This gives me access for basic lookup from anywhere in the house. I don’t need to be at the PC where I run Calibre. Again, the benefit here is metadata. Have I read this book yet? What series does it belong to? Who are the authors? A flat file system can’t manage that.
The message I see time and again with JD is, “so I know where to store/find it” and that’s valuable. The JD setup I have worked on so far has greatly clarified things. But, and this is where we’re getting tripped up with “knowledge” storage, is there is only 1 point. We have to choose and that creates a cognitive hiccup when a document can be in multiple places.
In Zotero I have bank statements that show USD conversions. I need these at tax time. The receipt gives me a price in USD, but I pay my taxes in AUD. Because Zotero isn’t a flat file system - every item has it’s own unique folder with the files within - I can have a particular bank statement appear under the bank category, and this year’s tax. JD makes it clear which Zotero collections I need to work with. It can now only be in one of two places, both of which are 100% valid.
That brings us to Obsidian. On the face of it, Obsidian is a note-taking tool. Add backlinks, folders, tags, properties and plugins like Dataview and it is clearly also a system that manages meta-data. As important as the content of a note is its relationship to other notes. The actual storage location of the file on disk means very little. Folders are a super-high categorisation for convenience.
At the weekend I tested building a matching JD structure in Obsidian. It doesn’t work that well. Better is tagging notes with the JD hierarchy (nb: decimal points in tags aren’t allowed so there is a slight conversion needed, but if my index is in Obsidian I can just link back to that).
Moving a file from one folder to another i.e from where it’s written originally, to the Quartz publication notes folder for my website breaks absolutely no semantic meaning. The only thing I may have to change is some code to filter out tags I don’t want to display on the website.
I get more advantage from having notes interconnected within Obsidian (wherever they are) than I gain from the organisation structure of JD as a hierarchy.
JD’s AC.ID flags are simply metadata on a note. They are what is important, not the underlying file structure.
Time to keep plugging away at it and time will tell. PKM is never static.