More fodder for the Folgzettel debate

I didn’t see discussion of this in my read of the debate so I’ll throw this out there.

On one hand, hierarchy in your folder structure enables you to think about your information at different levels of abstraction. There is clearly value in this, hence the popularity of header levels and also collapsing bullets in outliners.

But not all organization issues have a clear direction for the nesting. Say for example you track Companies and Tax filings by year. Should you put the tax filings folder in a folder for each company or put each company folder in a folder for taxes by each year?. Or should you have folders for each in which you nest each company in which you nest each tax statetment? It isn’t clear what is the best parent child relationship.

Databases use an idea of a junction table for managing many to many relationships.
Maybe there is value in an idea of a Junction Folders.

Simple Example. Assign a number for each category:
(10) - Companies
(10.1) - IBM
(10.2) - Apple
(20) - Tax Filings

Now, create a new folder:
(10.1)(20) - IBM Tax Filings

In this folder, save each tax filing by year.

The use of periods can capture hierarchy while the use of parenthesis does not if hierarchy is not helpful for organizing the data. So, for example, IBM and Apple are subcategories of Company and that relationship can be captured with a period while Taxes and Company don’t have a single obvious parent, child relationship.

The ID is unique but also gives a sense of the abstracted layers (e.g. companies and taxes). So you can search by ID but also get some knowledge of contextual fit.

These are just some preliminary thoughts, inviting folks to chime in with criticism or extensions…


Isn’t this exactly what tags do?

1 Like

Yes. Tags can handle this sort of relationship structure and categorization deftly and flexibly. Trying to digitally emulate a physical structure (the card box) results in a set of brittle rules.


So do you have a flat file structure with a single file folder where you put all files?

All files are saved somewhere and while tagging and linking is great in Obsidian, it isn’t the default for saving files and file types generated in other programs. So where do you put these other files? How do you tag them? What naming convention do you use? Are file folders ignored all together? If not, what is your naming convention and nesting convention for file folders? If you do ignore the use of file folders, are all files in a giant heap?

I haven’t read into all the posting on it yet so I don’t really have any strong opinions to share with you. Earlier this summer I did collect a bunch of articles on the topic to process when I have the time

1 Like

I have a few folders for specific projects in Obsidian, but my main documents which are not notes (PDF, admin, references…) live in DEVONThink. However, I seldom access them from there. Instead I use the direct internal links DT provides which I use in Obsidian. Links and MOCs are the backbone of everything.

Folgezettel is about a sequence. People seem to forget that Luhmann was also crafting narratives (almost complete sections for books and articles). This is what they were about.

Thanks @anon41901724
Devonthink seems great but Mac OS only. What to do if on Windows? Also, what would happen if you didn’t have DevonThink? Would you have a hard time finding/accessing your main docs?

Do you think Folgezettel was only about sequence or do you think elements of it serve other functions as well? e.g. creating a sense of hierarchy, proximity to other ideas without direct linking, etc

Yes. Thanks @lizardmenfromspace. I think your link first led me down the rabbit hole.
Curious, do you have any system for storing non-note type files? Do you use a file structure? What is your favorite way to find and get stuff that doesn’t have the benefit of Obsidian tools around linking/tagging/search?

I’m not using Windows so I can’t say, and DT stores documents in a non-proprietary database, so I can leave the app whenever I like.

1 Like

Random thought in the room: Instead of using folders, use notes (.md files) as the first hierarchy. From there, you build relationships between time and those companies.

The cool thing about it is you use full advantage of Obsidian flexibility. You can navigate from companies to their taxes (bottom-up) or from taxes (top-down) to each companies taxes. Also, the .md file 2019 behaves as a dimension, which means everything you link to that can be queried with additional filters in the search function.


@Finessest , the idea sounds interesting. Here is the thing that is still confusing for me. Even if you use Obsidian, everything exists on your computer in some sort of file system. You can choose to simply put things in the root directory or can put things in other directories, but they have to be put somewhere. What is a good practice around this?

In your example, I’m sort of imagining a MOC style note with filepath links to other files that becomes the alternative to the Windows explorer app on your computer. That, no doubt would be great, but still wondering what is the best structure to put everything in? Would you have ANY directory folders? Would you create any sort of folder structure?

If I understand his method correctly, I think it was not about hierarchy — not in the beginning, at least — but the hierarchy would start to emerge through the branching of Folgezettel. I think the idea was not to start with a preconceived hierarchy, but rather to establish it through research. I think this is what does make Zettelkasten a research system, rather than a knowledge map. Ultimately, however, I do not think it is necessary to stick to Luhmann’s vision (I personally like it a lot, but I am actively involved in research, so it agrees with me), but rather adapt the system to your own goals. With regard to Luhmann’s vision, I found Daniel Lüdecke’s presentation of the method to be the most revealing (especially, slides on branching); DL is also a developer of ZKN3 (

Thanks @romanov.maxim. The Daniel Ludecke presentation is great. The irony is that the second I clicked on the link, I had to decide WHERE to save the document.
Folgzettel seems to say, just put all files in a giant heap. Don’t create folders or categories. Use the numbering system, linking system and tagging system to organize. Ok. That might make sense for notes but what about other media? Take the example of Luhmann. He had books and papers that he read to generate the notes. He had to physically place those books and papers somewhere. I doubt he simply put all books on a shelf in no order whatsoever (the equivalent of a heap). I doubt he stuffed all the papers he read into a file cabinet with no organizing principle. The same with your electronic files. All electronic files need to be saved somewhere. Should they all be in a single folder with the file naming convention and search doing the work of organizing them or is there value in introducing levels of abstraction by create some sort of directory file folder system? With Obsidian and your notes, you can have everything in a single folder and use the powerful search, linking and tagging to find things but what about with everything else? Devonthink seems to be well received by the Mac crowd. Not sure of a Windows alternative or a mental framework that might work…

Ok, what you are talking about has nothing to do with either Zettelkasten or Folgezettel. Zettelkasten was about research writing and not about organizing all of what he had. He was actually stressing (at least in one of his interviews) that he was spending most of his time maintaining his Zettelkasten — more time than he did writing his books. What he did was ensuring that all the Zettels are connected properly with each other forming an argument. Again, Folgezettel is the mechanism to connect atomic notes into such arguments. This is not a storage system. I am not sure that Obsidian is meant to be such a storage/organization device, although it can be, of course. I still use Evernote for this purpose, but I am trying to be more selective with what I throw into it.

With regard to what you are talking, yes, some kind of hierarchy would be a good thing, the problem is how to create and maintain it. I am still struggling with maintaining mine :slight_smile: (I just use a loose context folder approach). All my PDFs are in Zotero, that helps, but other files keep floating around and I occasionally “lose” them.


@tuckytoo @romanov.maxim | #zettelkasten #lyt

Regarding the folder structure, you can get some inspiration with the para- or lyt-system.

If you’re asking me, what is the best file structure you can build in Obsidian then I’ll probably answer: “none”. It’s an endless fight against order and disorder, and any energy you putting in, the more likely it is to result in chaos in the long run.

But there is something you should do against it, although I will contradict my above answer then.

When we put everything aside and we’re working only with Obsidian given functionality, folders can be useful to reduce the potential room of mess. In our case, folders are useful to automate the creation of a Zettelkasten note (core plugin, Zettelkasten prefixer). Or by default put every image in a specific folder. Quite handy.

You can extend this to your taste by adding folders for different media (audiobooks, YouTube) or by objects like people, workspace, places et cetera. In my opinion, this kind of extension in the functionality level is just a human need because we grew up doing this since the first operating system added folders.

My personal vault in Obsidian should be a replica of my brain, and as far as I know, I don’t think I store my knowledge in my brain in the form of folder storage. An unknown person taking over my vault would have no idea where to start searching. But as soon as I reveal my system, the connections, and relations between the dots, suddenly everything makes sense again and the disorder is no longer a disorder, but pure order.

That’s why, I don’t consider Obsidian only as a note-taking app, but also a browser of knowledge. You need Obsidian or similar to navigate in your .md data, or it looks like a mess in Windows Explorer.

To finish my post, I recommend every newcomer to first write as many notes as possible, make connections, and create a cluster (MOC) after natural collisions between the notes have occurred. The structure arises by itself when you produce content. That’s the beauty of Obsidian.


Thanks @Finessest.
I’ve reviewed Para, have less familiarity with LYT but get that it uses MOCs to help navigation. I suppose you could create MOCs for data outside your vault by either creating dropbox links and to the ‘[’ Name’]’(URL)" syntax or just including the filepath. No doubt there is value to curating these MOCs over time. It just occurred to me that the funny thing that PARA (and maybe the category parts of LYT) is they are systems that in effect try and replicate elements of search algorithms in computer science. For example, PARA would cut the number of records you need to search by 75% (assuming an even distribution) if you know the starting folder. This is because once you pick the right folder, you get to ignore all the files in the other folders.

One might imagine an alternative use of Obsidian as a way to organize and link metadata for finding non-note related data on your computer (with references done through dropbox links or filepaths). Feels brittle though as you would be lost if links broke.

1 Like

Something that Tiago Forte encourages with the PARA method is to use the same structure across however many apps you work across, and name your Projects and Areas exactly the same.

In theory you wouldn’t need to use an actual link to a file in dropbox (or another resource in a different system) from your obsidian note, because you can navigate to files associated with your project/area relatively easily - “just” navigate to the same project or area.

That’s just theory though, and I’m sure practice differs in many cases. Like if there are a large amount of non-note files for a project in dropbox, then I guess linking directly will be preferred over manually navigating each time.

Another organization system I’ve read about (but have no personal experience with) is Johnny Decimal - - in fact your original example of numbering and junction folders is kind of similar. It doesn’t get around having to decide whether taxes are under companies, or companies are under taxes, but it might be interesting to you nevertheless.