🌲 Yet Another Hot Take on “Folders vs Tags”

I like folders due to their cross compatibility, future proofing, interoperability, flexibility, focus, clarity, automation management &etc.

Folders are controversial in personal knowledge management circles. The common refrain I see is “why do you need folders if you have tags? It’s 2021!” For me, the answer is about 4,000 words but pretty simple:

Folders have good backwards compatibility

Most software I use wasn’t created in the 2020s, and interfaces pretty poorly with a tag-based system of organization.

For example, let’s say I want to email a friend my notes about potty training. I could scroll through 1,500 notes to find the one I want (harder if I don’t remember the filename off hand — what if it doesn’t start with p?), or I can just click through “personal > parenting” and and evaluate the ten or fifteen notes there to find the one labeled “oh crap! potty training.”

I suppose that as an alternative I could download a windows search tool (or open Obsidian) and find the file name I need, I could use VS Code or Notepad++ or a similar tool to find all of the files that have a particular tag in them using regex and that … would work. But it’s harder, and less flexible to do things this way. What if I’m on a mobile device and don’t want to run VS Code on a remote server? It’s much easier to interact with files on my file system if they’re structured in a way file pickers are used to.

It’s easier at the command line, too, which is one reason the shift away from folders and filesystems is concerning to me as an educator. When I was writing my concatenation plugin, I could have used the Obsidian API to programmataically copy files that contained a particular tag, but I started out my programming journey with pure javascript, no API. Take my word for it — it’s much easier to tell a computer “do this to every file in this folder” than to tell it “first, find everything with this tag, then put those into a string, then do a thing to every file in this string…”

Just because you can do something the “hard” way, doesn’t mean it’s worth my time.

The flip side, of course, is that if all your files are notes and the notes aren’t easily balkanized between topics, it’s not worth the extra cognitive effort to impose an external structure on them. My “slipbox” folder has notes on everything from raven mating habits to Aegean warfare to the importance of voice in evaluating prose. But it’s very easy for me to separate out my newsletters or character profile sheets from my programming notes and video game plans using folders. I find the endless possibilities offered by tags to be incredibly overwhelming, so folders work really well for me in a way that they might not necessarily for someone else.

A convincing counterargument

I once heard someone explain that one reason they liked using tags in their metadata was because they liked all of the metadata for their note to be self-contained. If you email someone a file, it won’t contain information about what folder the file was housed in, but it will contain the tag. I find this to be an incredibly reasonable argument for including YAML metadata along the lines of folder: folder-name but not for including it as a tag, personally. That said, I can imagine a world in which someone wants interoperability between a program that supports both tags and folders and a program that only supports tags.

The main thing that irritates me is when I see people say that folders are “old-fashioned” and therefore “pointless.” The fact that they’re old-fashioned is one of the big reasons why they’re valuable: more programs, especially at the lower levels of the file system that I prefer interacting with, support and extend their use. I believe in the Lindy effect (robustness is proportional to the length of lifetime) and am used to backwards-compatability efforts. I trust folders more than tags.

I don’t like to clutter tags and indexes

A lot of senior developers don’t like JavaScript and electron (both of which are, uh, really critical to the Obsidian community). There are a variety of reasons for this, but the one I hear most often is that programming boot camps have been churning out people who know JavaScript and not a ton else, who then go on to use JavaScript for every project — whether it’s the best tool for the job, or not. The idiom “When all you have is a hammer, everything starts to look like a nail” comes to mind. Using a tool like Obsidian, I have way more than a hammer (tags, in this metaphor) at my disposal — so why would I limit myself needlessly? Why would I use a hammer when a screwdriver does a better job?

But on the flip side, I wouldn’t use a screwdriver when a hammer is better for the job. Tags do have uses. Tags are a tool. Folders are a tool. Trying to use tags for everything makes tags less useful in my opinion, because then your tags get cluttered by things that could be divided up and made more easily filtered with folders.

I could, in theory, tag all of my files with the same controlled vocabulary that I use for folders. I would have to use a plugin or a program to update them when I change them, instead of just changing the folder name, but ok fine. Plugins and programs for that exist. I could nest my tags so that I can collapse them in the file view, but I can’t hide them — there would still be clutter to sort through if I used tags for organizing and assigning states (which I do with YAML) and flagging things I need to follow up on.

Deprecating folders in my workflow would add to a lot of useless clutter and extra work, which is why I don’t do it.

I use my root as an inbox

I can’t stand a cluttered work surface. If there are papers on my desk, I organize them almost as soon as I sit down. At work, I have inbox trays with different folders for different assignments, on trays for different classes so I know what I need to grade, what I need to hand back, and where to file everything. If a student goes to hand something in late, I tell them to put the paper on top of my keyboard or my chair, because I’ll file it as soon as I sit down.

Not everyone is like this, and they are just as good as I am at getting work back to students on time and in an orderly fashion. Some have firm rules about turn-in bins and binge grade on the weekends, and their desks are messy from a student perspective but the teacher always knows where things are. My husband’s method involves neat orderly stacks of papers on the floor, which makes my shoulders itch but I leave him in charge of all of our truly vital documents because his file cabinet is organized with an absolutely pristine meticulous perfection. Things don’t get put away until he has time to do it right.

I tend to file things in a good enough place right away, because I really hate having stuff laying out in my workspace if it’s not actively being used. My notes are the same way. When I make a temporary note from my phone that’s badly formatted or half-finished or littered with voice to text errors, it lives in my root until I can process it at home from a real computer. There are a variety of methods for quick capture, and when I tried to use an “inbox” or “pending” folder, I discovered that I almost never actually checked it. It became a black hole. So now I quick capture directly to my root folder, because it will bother me until I fix it.

Quick capture

There are a couple of different methods for quickly capturing a fleeting thought into your notes. The most important thing for me was the emphasis on quick, so I use whichever one is easiest at any particular moment. I’ll go into more detail on these later, but for now I want to say that that I still use an “inbox” for notes “pending processing,” but mostly for situations where another program like Readwise is syncing stuff for me.

How I use Folders

There are a lot of organizational systems for folders floating around. Historically — before I found Obsidian.md and the broader personal knowledge management community — my folders were organized roughly by type. They weren’t disorganized by any means, to be clear. My writing folder was pretty haphazard, with old stuff in one folder, completed manuscripts in another, and the rest organized by universe. I had tax stuff and photos in folders of their own, organized broadly by year and type. Old coursework was located in my “school” folder, organized by university and then by class. The rest was sort of shoved into various downloads folders, which was fine because none of that stuff really mattered.

Notes? What notes? Everything was in my head, or posted in an article online, or in an email somewhere. My bookmarks were neatly organized into folders.

The point I’m trying to make here is that my personal knowledge management system didn’t revolutionize my life — it refined it.

When I first came across Bri Watson’s incredible post Cataloging, Classification, Information Science, PKMs and YOU! – Knowledge management I followed all the links. The general principles of Tiago Forte’s PARA system — that a digital system should work cross-platform and be future-proof, flexible, modular, and broad — make sense to me, but the idea of organizing myself into projects, areas, resources and archives doesn’t work for my brain at all. Often, my projects are my resources; everything about my process cascades into another step in a waterfall of ideas that rejoins the “water cycle” of creativity — I make an effort to ensure that I create “public-facing products” for most stages of my process, and even after watching endless discussions of how to implement PARA for the last year, I am still really shaky on “what goes where.” The boundaries don’t really line up with how I live my life.

Nick Milo’s maps of content made a lot more sense to me, and I definitely use them, but like with tags, they’re an additional tool, not a full-on replacement for folders — unless you’re only looking to organize a zettelkasten style slipbox, in which case… before you listen to someone saying folders are antiquated because they have a true zettelkasten and that’s all they need: what does the rest of their computer look like? Do they really only have atomic notes and literature notes and fleeting notes? No… other files relevant to their lives?

I’m sure even Luhmann kept published copies of his publications somewhere other than his slipbox, you know? Maybe on a shelf in his office, next to his slipbox?

Johnny Decimal System

Anyway, I eventually settled on the Johnny Decimal system because it’s the one that made the most sense to me. As someone who has been forced to store important documents on organization-wide intranets and shared drives, wading through idiosyncratic and sometimes duplicate folder systems that are the legacy of a conga line of people who no longer even work with the organization… the use-case it was developed for is something I have context for. It’s designed to let digital and analog systems work side-by-side in a complementary way, which appeals to me. More importantly, it was using principles I had already developed independently, with numerical prepending methodology that aligned to what I was used to from my curriculum.

I was already used to forcing file organization with file names like [02.01] Geography of Egypt - Foldable, [02.02] Cuneiform - Sourcing Activity and [06.01] Geography of the Americas - Paragraph — my curriculum is organized by unit (“Africa”) and lesson (“Medieval States”). Johnny Decimal uses a slightly different numbering schema — categories are pre-assigned but the second set of digits is in order, so for the above to match Johnny Decimal properly it should be [22.01] and [61.01] which would work fine if my curriculum didn’t have more than ten units — something I can’t exactly count on since I’m not the one in charge of it.

The point of the Johnny Decimal system isn’t to prescriptively require you to use “the one true way,” though, any more than Nick Milo is inflexible. This flexibility is one of the reasons I appreciate both Johnny & Nick. My favorite page on the whole Johnny Decimal website is Exceptions to the rules, where Johnny explains example cases (like dated timesheets and tax paperwork) where the JD numerical system might not make sense. Since my curriculum’s numbering system is beyond my control, I’m going to stick to it, and that’s okay, and that’s exactly the kind of validation and useful explanation with examples that I need from a how-to guide on organization.

I also really appreciate that the Johnny Decimal system is a framework, not a template. It’s a philosophy, which clear logic and reasoning. Here are the principles that resonated with me:

Numbers are for ordering

The Johnny Decimal system makes much of much easier it is to remember a number than a category, which made no sense to me at first. I am terrible at remembering numbers — it took me years to remember my husband’s birthday, I’m still shaky on the actual date of my anniversary, and I have no idea what anybody’s phone numbers are. I forget prices basically as soon as I’m done looking at them, and if I don’t write down a measurement immediately I’m going to get it wrong.

Bri explained that it’s not really about the number itself, but rather the muscle memory of knowing “where” a folder “should be.” I might not remember “80” but I remember that stories is almost the last folder in my file cabinet, and I might not remember “40” but my hand knows exactly where to go on the screen to open up the folder with my PDFs.

I’ve also found that it’s helpful for judging the maturity of an idea, too. Bryan Jenks has a great “seed / seedling / sunlight / evergreen” metaphor inspired by Andy Matuschak. I don’t use his tag system (I’m not a fan of emojis and I prefer to use YAML for tracking states — more on that later) but the way my folders are structured, smaller numbers are less likely to be public-facing products than higher numbers.

I’m a lot less likely to share my edit letters, daily gratitude journal and exercise log, taxes and pending files than I am to consider stories, newsletters and articles as final projects. Unlike Bryan’s system, this is less about a file’s status than its nature — my goal is to eventually “ship” everything filed after 70, and all the stuff in the middle is “fodder” that I expect folks might be interested in from a “behind the scenes” perspective but isn’t necessarily destined to be a product. There’s a flow to my numbering system that I also use inside those later folders.

The folder structure might not make sense for other people, but to me this order is intuitive: scraps evolve into finished products, and the resources I need (like the market information) are tucked into the placement where if I’m processing from top to bottom I’m most likely to need them; i.e. after I’ve come up with the idea but before I’ve “shipped” it.

There can be only one

The other nice thing about the Johnny Decimal system is that it reduces decision fatigue. Sometimes people will ask me what I do when I have something that could go into multiple places and the answer has always been pretty simple: if something could conceivably fit in two different folders I need to consolidate my folders. The rule that it should always be obvious which folder a file goes into has made it very easy for me to see whether I really need a particular folder, and really helped ram home why I prefer “type” folders to “topic” folders.

For example: it’s pretty easy to imagine taking a note about friction. If I were still in school, and I had my folders set up according to classes — calculus, physics, engineering — I might not know which folder to file “friction” in. That concept is useful for all three disciplines! If I were using a tag system, I might tag it with all three courses, and that could work. I could also write different atomic notes that reference each other, with each one focusing in on a different aspect like “mathematical equations for friction” “friction when calculating velocity” — but even then, I probably will find those useful for multiple disciplines. So I could take the wikipedia approach and do “friction (math)” and “friction (physics)” or just have my class notes organized per class and interlink where appropriate.

… or I could just have a whole folder called “science.” And eventually maybe I’ll learn that there’s similar overlap between, say, biochemistry and archaeology and that having “science” and “the humanities” in different folders doesn’t work, and I’ll merge them into “school” or even just “real world concepts.”

Personally I chose the third option, but depending on your use-case and background habits, any of the others would probably work fine.

You’re allowed to be flexible

Ideally of course once I came up with my original numerical system I would have stuck with it, but the reality is that since I don’t number (most of) my files, just the folders, swapping things around is pretty easy. Although the “official” Johnny Decimal system calls for numbering each item in the folder in the order you added it to the system, so things get placed one after another next to other things you’re working on around the same time, the reality is that this makes the most sense for a corporate use-case or a freelancer, not me.

So I’ve modified the system to let me go one level deeper if I really need to, and so far I’ve only done so in two cases:

• in the folder I use for teaching. I think it would be completely in line with the original purpose of the Johnny Decimal system to have an entire numerical organizational system just for work. Besides, trying to keep all of my educational materials in one “flat” folder is a bad idea on the face of it since I teach so many different subjects, and each subject my change substantially from year to year. Also, this folder is not a “working” folder, it’s an archive, because I try very hard to only work at work. So it’s basically backups for the end-of-the-year purge crises that happen every year during the summer. By the way, if you’re interested in using Obsidian or personal knowledge management as a teacher, check out the History4Today channel on YouTube.

• In my references & pending folders, because a lot of things happen programmatically there, i.e. Readwise autosyncs to one folder, Zotero syncs to a particular folder, the Kindle highlights plugin syncs to a particular folder… and I don’t want them running afoul of each other.

Automations

Even though I automate a lot of things with ebooks and PDFs and stuff I read online, I don’t want you to think this workflow requires fancy tools. I also am prone to sitting down with a dead tree book, hands taking a pen and annotating either in the margins if I own the book, or on a post, and if I don’t, and when I’m done reading or done reading a section, depending on my mood, I will sit down at my computer and paraphrase things and annotate them and do it all by hand.

There are also a lot of different ways to get highlights and annotations into a vault. And so there there is not a need for fancy tools, if free wise, you know, goes bankrupt tomorrow or my Kindle stops working or Zotero doesn’t work on the next version of Windows or whatever it might be, I will still probably use this system. So I do think that it is fairly tool neutral.

That said, if you’re going to use automations, I highly recommend putting each “raw dump” into its own folder, because these tools tend to do things “en masse” and you might want an easy way to delete and resync everything all at once, or an easy way to see everything that just happened when you pushed a button in a piece of software.

But what about tags?!

Throwing things at the wall to see what sticks

If you carefully maintain your tags with a controlled vocabulary, where you keep a glossary or index to define what everything is for and how your system is set up, tags are probably fine. My habit, though, was developed on websites like Goodreads and Twitter and my newsletter, and as a result I use tags very liberally to maximize my odds of having overlap between tags so I can make connections. This is an emergent system, and after awhile I start to get a feel for which tags I most commonly use. It works great for trying to manipulate the “you may also be interested in” algorithm and is really quite terrible for meaningful organization (which is what categories are for, on my blog).

I personally am way too lazy efficient to do that well.

🤔 Digression: I’m really working on calling myself efficient instead of lazy. Like how I need to stop calling myself “bad at math.” I took two years of high school calculus, I’m not bad at math — I just spend so much time with professional mathematicians and engineers that my sense of normal is skewed.

When I first started bullet journaling I read all about the value of maintaining an index and then could never bring myself to actually do it. I have colleagues at work who require students to maintain perfectly organized indexes of their notebooks, and while I totally understand why, I could never bring myself to do the same. Eventually I tried the highlighter method, where you categorize different kinds of pages with a little highlighter mark or a piece of washi tape, and it corresponds to a little pull out key in the back of your notebook. I didn’t get very far with that either. These days I just write whatever then cross it out when I’m done with it, and type up anything that is still important when the book gets full.

Trying to maintain a controlled vocabulary used to stress me out — what if I misspelled something or forgot the taxonomy or spend so much time agonizing over the best way to organize things that I never actually created anything? So for a long time I just avoided them entirely. If I used a tag at all, it was as a random note-to-self like #followup or #synthesize — something I could see in the tag pane, something that would annoy me precisely because it wasn’t neat and orderly. I wound up using the tag pane as a pkm to do list, instead of following the (extremely sensible, I repeat this advice myself all the time) conventional wisdom that tags are great for helping filter things into or out of the search results.

Then my graph being messy stressed me out, so I tagged every file with the name of the folder it was in. I hated the redundancy, but at least my graph view was navigable. Then the zoottelkeeper plugin, which maintains an automated index of all files in a particular folder, became available. I was able to delete all but one of my “folder tags” (I sort articles into one of two different folders because of how I expanded my newsletter) and now folders do everything for me that folders feel like they “should” be able to do.

A verb-based tag vocabulary

I was left with a mismash of notes-to-self that was starting to emerge into a coherent system. So I consolidated it and now here’s what my tags look like:

🔖 #FollowUp #addmoc #twitterfodder #articleseed #research #crossreference #EditPending 

But you do you!

Anyway I’m not trying to tell anybody to use folders instead of tags and I definitely don’t think my methods would work for everyone, I just wanted to sit down and write a companion to my defense of Obsidian’s graph view as part of my ongoing series about my philosophy of digital notetaking because this dicsussion comes up a lot and I’m kind of tired of copy and pasting the whole spiel into Discord 🙃

If there’s an angle you think I missed, please let me know below, because I’d love this link to be my definitive statement on the matter.