Folders vs. linking vs. tags—the definitive guide (extremely short—READ THIS!)

Use a lot of folders if you’re using Obsidian for managing writing projects, such as articles, blog posts, video scripts, etc.

Obsidian is great for managing writing projects. One folder = one project, containing all the notes used in that project. When the project is done, keep the folder around until you’re sure you no longer need it, then archive it somewhere or delete it.

You can even use your Obsidian folder to store Microsoft Office documents—Obsidian can’t read those, but you an open them easily enough.

Use minimal folders if you’re using Obsidian for building a personal knowledge management system.

Say you’re a historian who studies the Roman Empire. You’re building a personal knowledge management (PKM) system based on notes you take on books and academic papers about the Roman Empire. That’s just one project, so use minimal folders for that.

This is where you use links heavily, in a manner extensively described elsewhere. If you’ve used Wikipedia, you know how to link between documents in a PKM.

Older people tend to be more comfortable using folders.

I’m not being ageist here. I’m an older person myself! We came up in a world where folders and document names were the only way of organizing documents, and we’ve gotten really good at managing those. There’s no reason to switch if folders are what you’re used to.

Also, folders have been used LITERALLY 3,500 YEARS, since the invention of writing. Books and bookshelves are a kind of foldering system–the book is a container in which you hold documents (like a folder) and the bookshelf is a bigger container (a bigger folder). Systems that have been refined for thousands of years often work very well.

Don’t use tags. They’re useless.

Seriously, some folks like a lot of tags and if that works for you, go for it. I think tags are confusing, for reasons that have been adequately discussed elsewhere (Do I tag this document “dog”? “dogs”? “pets?” “mammals?” with the dog’s name?) and I use tags sparingly or not at all.

Tags can be used in a fashion very similar to folders, except a document can only be in one folder, but it can have multiple tags.

DevonThink, a powerful document organization app for the Mac, lets you put the same document in multiple folders (which DT calls “groups”). In DevonThink, groups and tags are nearly interchangeable.

Related: Some years ago, I spent a whole weekend researching what the heck the difference is between “tags” and “categories” in WordPress. I determined that it comes down to “some people prefer tags, and some people prefer categories.” I have found this insight useful in the years since, when evaluating software that seems to offer multiple ways of doing the same thing.

But what about linking?

I use links to describe the contents of a folder. I have a folder note for each project, and links to each document in the folder. The text of the link supplements to document name to tel me more information about the document being linked to.

But what about properties?

I’m using properties quite a bit, where I might previously have used tags. I might talk about that another day.

Mitch, you said this was going to be a definitive document—the only document you need—but you left us hanging on Properties there, and you left a lot of other stuff out.


Also, this document isn’t all that short.


Generaly I agree with you. Link and properties can do almost everything.
I have 2 usages of tags:

  1. #todo ; easy to write, easy to delete
  2. Moving note to folder
    Plugin Auto Note Mover requires tag (or specific name)
  3. Nested tags are workaround for nested properties (mulit level YAML)
1 Like

Your comments on folders make sense, but I am not convinced about the uselessness of tags (a comment I hear frequently). The reason I am not convinced is that I find them very useful. I have about 1,000 academic papers that I manage with tags. Example tags might be #Finite_size_effects #Spin_Glass #TAP_equations. So if I quickly want to know what papers I have on finite size effects in spin glasses its easy - I just put the tags in the search box (the consistency question of #dog vs #dogs is not a problem when you have auto-complete).

My question is, how could I achieve the same ease of retrieval without tags? I have heard I should use links (which I use extensively for other purposes), but I am not sure how or what the advangtage might be?

1 Like

I use folders and tags. They are both dimensions of organizing documents. Or, in the case of tags, bits of documents. Each method is optional. And each method has a variety of techniques for using them, and different use cases in different circumstances. It’s humorous to declare tags are “useless” then go on and explain various ways to use tags.

Linking is not in the same organization category as folders.

I assume the academic papers are PDFs. How do you put tags on those? Do you create a container Markdown document for each PDF? That seems like a lot of work. Or do you use some other repository for your academic papers, like DevonThink for Mac?

How do you use tags differently from folders?

At least for me, folders do have a few advantages:

  1. Bulk operations on collections. This is useful for when I want to archive a project or copy a slice of a vault for some other purpose. I can do that with queries, but folders under “unix” (including MacOS) are just queries with a very robust set of pre-built CRUD (Create, Read, Update, Delete) operations.

  2. Working with a vault outside of Obsidian. Last weekend, my primary computer was inaccessible. But I was still able to make changes via github’s editing interface. If I’m using fzf or other fuzzy-finding file-management software, file paths become zero-effort tags.

  3. Providing limited classification options in advance. Johnny Decimal works better with my brain than tag clouds. (Not a purist, I break multiple Johnny Decimal rules.)

  4. Future-proofing: Frankly, almost everything that’s not plain text in a folder has become obsolete in about a decade. There are some outliers like PDF, but PDF is notoriously hard to work with as an information format. The vault is where the knowledge lives. Obsidian is just the current editor/ide that makes some queries easy and pretty.

I see tags and folders as sort of orthogonal to each other, which is how libraries have been using them for over a century. The classification system ensures that the book has one, and only one place in the library while in circulation, shelved with items linked by most prominent topic. The tags (in the catalog, or helpfully printed on the title page) permits examining items on a qualitative basis similar to a k-nearest-neighbor search.


How do tags work in libraries? I’ve been ebook-only for so long that I forget.

I do everything in Obsidian. I create a markdown note and link to the pdf. The markdown note is very useful because there I can say why I was interested in the paper and what other papers are relevant to it.

The note title format is “year - authors - title” so with the tags I can easily narrow my search to papers on a particular topic, in a given year or with particular authors.

1 Like

My memory is that card catalogs were cross-referenced to multiple categories. So you could find Macbeth under English Dramatic Works, Scotland in Literature, and Tudor Literature. This evolved into tag and keyword searches with the transition to OCLC (online catalog of library collections.)


I agree. I find tags quite useful, thanks, as you say, to autocomplete.

It’s also important that with a few handy plugins you can create and manage subtags (you can have Dog/VeterinarianBills and Dog/Alsatian/Traits or whatever, so the overall tag “Dog” doesn’t become a mess). And you can manage tags from an overview list. (Ie, you aren’t trapped by earlier decisions – you can eliminate a useless tag or combine two when you decide they are really one concept.)

I can’t imagine using Obsidian without tags.


What @KaeS said:

Folders are one dimension of organization, tags are an alternative dimension of organization.

So I might have a folder with individual notes, each of which recapitulates the chapters in a book on the history of the newspaper business in the U.S. in the 20th century. Then, within the notes I might have tags that reference #NYTImes, or #BostonGlobe, or #EmporiaGazette, etc.

I could use search in Obsidian (or the tag browser) to look for #EmporiaGazette, which would return the notes in the folder I mentioned, plus other notes mentioning that publication. Or, I could merely click on #EmporiaGazette in any note and a search opens in the sidebar that lists all nodes containing that tag. It’s as if my node hierarchy were rotated 90 degrees on the fly, and now I see a “virtual folder” of nodes relevant to the Emporia Gazette.

I won’t get into nested tags, which could yield yet another 90 degree rotation of the note space.

1 Like

Thank you for the thoughtful response.

If you’re looking for notes about the Emporia Gazette, why not just search on that string? Why bother making a tag?

Because I value definition over uncertainty :sunny:

In other words: Tag auto-complete protects against typos?