How do you organize your computer files?

iylock · January 25

How do you organize your files?

I'm thinking of attaching UIDs to all files like Zettelkasten and then creating structure notes.
However, I think that as the number of files increases, management becomes more difficult.

The reasons for considering this are as follows.
1) To write a description of the file.
You cannot write descriptions on sound files or picture files. You can use metadata, but it is sometimes a problem for me as I work while changing OSs.

2) To track relationships between files.
For example, if you want to describe the relationship between one photo and another, you inevitably need a txt file.
Files can also be grouped into different categories as needed, which is difficult to achieve because folders only provide a single structure. That's why I gave up on Johnny-Decimal.

Additionally, music, design projects, or programming work require multiple files, and I'm curious about your opinion on whether it would be useful to assign UIDs to all of them or to assign UIDs for each project.

JasperMcFly · January 25

Because no two files can have the same name, they already have a "unique ID", their name.

So, for me, it is more of a question of how do I refer to them from my notes.

A good desktop search engine (I use Everything from voidtools) can find just about any file with a search against parts of two words in the filename.

So, searching for 'smart not' finds any file on my laptop containing the words smart notes.

Academic papers tend to be easily found when searching against author last name and one or two words in the title. 'Ahrens Smart' is more than unique enough for example.

A structure note can describe relationships between notes and references and describe references.

structure note 15/b

Smart Notes talks about writing short notes as a way of thinking and pre-writing larger projects. See 'Ahrens Smart'

See also 'Doto Sys for Writ'

That said, the subset of files (or books) that I have selected as sources to make source notes and then main notes from, I will assign a Source ID number which I can use to refer from my notes.

'See Ahrens 0048 p.45'

Ydkd · January 26

zk_1000 · January 26

Metadata is just a fancy word for data about data. File systems or digital systems are not a requirement, so you might just want to write your metadata on paper sheets using a pencil.
With that in mind, your description is nothing else than metadata. It can be stored digitally in binary format, such as README.docx or in human readable format.

Also, using one system for all of your files is a bad idea. There is a reason why each file format is specialized for a certain use case. Using UIDs litterally everywhere won't make your life easier in any way.

Start with a single workflow, figure out your requirements and keep improving. Using standardized formats has a huge benefit, but it won't allow you to use exotic features. Exif might not be a requirement for you, but portability certainly is. Speaking of requirements, what do you work on while changing OSs?

ctietze · January 27

@JasperMcFly writes:

Because no two files can have the same name, they already have a "unique ID", their name.

I would like to correct this to "they already have a unique ID, their path".

/pancake/ingredient/banana.txt
/fruit/banana.txt

You can have banana.txt in two different folders without conflict, but still end up with duplicates in the name, where the file name alone would be "banana.txt" in both cases (technically, you may call "file.txt" the basename, and thus "banana" becomes the file name, and "txt" the file extension)

Paths are somewhat stable, at least relative paths, since absolute paths won't survive change in operating systems from Windows to anything else, for example.

I personally believe that PARA is a good starting point to file and find stuff, emphasizing the changing nature of what you need at any given day (i.e. moving files around):
https://zettelkasten.de/posts/building-a-second-brain-and-zettelkasten/#the-system---para

There's also Johnny Decimal if you want to have an ID-like system that grows organically:
https://johnnydecimal.com/

(I personally also use to-the-second precision timestamps like 20250127064710 for invoices because then I can reference PDFs by ID with ease from invoicing applications, spreadsheets, ... -- but that's about it.)

JasperMcFly · January 27

Ah, agree, good point. Path is unique, not filename. My desktop search displays results in primary field as filename, so not an issue for me, but do see possible conflict. Thanks.

micahredding · January 29

@ctietze said:
I personally believe that PARA is a good starting point to file and find stuff, emphasizing the changing nature of what you need at any given day (i.e. moving files around):

How do you do this practically, in terms of the file system? Project and Area folders seem straightforward. "Resources" invites its own subcategorization, along the lines of Areas, perhaps. "Archives" are organized by—what?—year and project?

iylock · January 30

@ctietze said:
I would like to correct this to "they already have a unique ID, their path".
/pancake/ingredient/banana.txt
/fruit/banana.txt

It was difficult to manage duplicate files, so a few years ago I started putting all files into one folder.

@Ydkd Thanks for your comment. The tag idea is good. But this didn't work for me. In one case, I had a situation where I needed to track a person in photos. However, if you have a photo that contains, say, 10 or more people, it is not possible to include all of the individual people in the file name. Tags were not enough to express places, people, and the context and causal relationship with the previous/next file. This is due to limitations in file name length.

@JasperMcFly When I simply managed references, I didn't encounter any problems when I followed your method. However, this did not work when trying to process the files as building blocks. Something like design asset files.

@zk_1000 I don't believe in metadata. This is because while managing files for over a dozen years, I have encountered many cases where metadata was corrupted and information was lost.
While I'm changing the OS, I work on music or design. For music, for example, there are many sound sample files. kick, snare, hi-hat, cymbal...and there are various tones in kick, and even for the same tone, the sound varies depending on the velocity. So there are a lot of files. To make music quickly, you need to create lists of pre-assembled sound sample files. It is disadvantageous to represent various combinations using only categories such as folders or tags. In particular, tags may not work if you switch OS.
This is why I tried to assign UIDs to all sample files and create lists so that one sample file could be included in more than one combination. It is difficult to rely on metadata in these situations. 1) It's difficult to see the whole picture, and 2) metadata can be corrupted when changing OS. It seems to be fine between Windows and Linux, but unstable on Windows and Mac. Especially since I speak Korean. Korean doesn't seem to have good compatibility between Windows and Mac. File names are also sometimes corrupted.

@ctietze For the above reasons I don't use PARA or Johnny Decimal. Also, as @Sascha points out, PARA has no fixed location for the files, which confuses the brain. Asset files are building blocks, so they can be treated as Zettels. Rather than having the file location change over time, I would like to access a stable area like a structure note. Of course, when it comes to projects, I need to reconsider PARA.

@zk_1000 said:
Also, using one system for all of your files is a bad idea. There is a reason why each file format is specialized for a certain use case. Using UIDs litterally everywhere won't make your life easier in any way.

One thing I'm curious about is that you said "Also, using one system for all of your files is a bad idea." Is it really a bad idea to only implement one system? What's the problem? 1) To organize files, you need to introduce at least one system. 2) I think one system is convenient for management. I think you should choose the main system as there may be conflicts between more than one system. Please let me know if I missed anything.

@JasperMcFly And on Windows I use Everything and AstroGrep. (Everything for file search, AstroGrep for full-text search). When I read your comment about how to use Everything, I checked to see if multi-line file list searches were possible in Everything. Search is possible. So I started making a file list like a structured note.

ctietze · January 30

@micahredding said:
How do you do this practically, in terms of the file system? Project and Area folders seem straightforward. "Resources" invites its own subcategorization, along the lines of Areas, perhaps. "Archives" are organized by—what?—year and project?

N.B. I don't have privileged access to the PARA gospel, this is just my thing:

I stick to a linear project archive. It's a folder called Project Archive, and into it I move project folders with a date prefixed, like 2025-01-30 ACME web app. (Over the years, I got so used to navigating around in macOS Finder by just typing ahead to find stuff that I actually don't enjoy finding anything in this archive. I can't get to that folder by typing "ACME", because the type-to-select UI works on the start of the filename, only. That's a date, though. I don't remember that; I remember the client name. I'm hesitant to switch dates to the back though; and I don't try to find stuff in the project archive that often, so I'll survive with the current naming scheme.)

I also have a tax archive. It's not in the same folder. All tax stuff is in a taxes folder, backed up 10 times across the globe

Ok, ok, I also have a paperless office archive. I think I started this in the image of MacSparky's Paperless Field Guide 10+ years ago. It's a "Filed Documents" folder, with 1 subfolder hierarchy for the year, and into it I move every scanned snail mail correspondence or other paper documents. Except taxes. These go into the tax folders. Like every German, probably, I'm really paranoid when it comes to taxes

The actual file system locations:

~/Project Archive/2025-01-30 ACME web app/
~/Documents/Filed Documents/2025/2025-01-30 Insurance info foo bar.pdf
~/Documents/Taxes/2023/2024-05-22 EkSt 2023 something something.pdf

~/Project Archive/ is a network drive on my NAS, actually.

Ydkd · February 4

@iylock said:

Tags were not enough [...] due to limitations in file name length

Had the same problem

The major insight i got from Karl Voit - when storing item, think about how you will (would try to) retrieve it in the future instead of thinking how to store it

You inevitebly will need some sort of DB if filenames are not enough for you and you dont want to rely on OS filesystem
There are some alreay existing soltuins, like:
I know Eagle for visual references management
FESS, tabbles, MS access - need to search for ones you need
Maybe you will need several specialised DBs

Manual reinventing the wheel:
You could assign IDs to files, then track IDs somewhere else:

create text files with same IDs + name and tags inside
have some DB with IDs - even excel
maybe some script to index your root storage folder instead of storing files in DB

Hook for macos can track relationships between files and more, but its macos only and paid

micahredding · February 5

Does anyone use (Mac OS Finder) file aliases?

I've been experimenting with them as a way to alter my file system approach. For example, I might have a comprehensive set of files (PDFs, images, notes, etc), and then create a folder within Projects or Areas, that pulls in a relevant subset of these, as aliases.

This feels helpful, because it's easy to have a canonical location, as well as some more ephemeral, task-centric locations. I've been a bit concerned about the fragility of aliases, but testing a few different things, they actually seem pretty robust.

ctietze · February 7

@micahredding I get uneasy when people use Finder aliases; I shared a bit about it here: https://forum.zettelkasten.de/discussion/3073/closed-making-aliases-of-notes-symlinks

Sounds like you use if for WIP things that change and where Aliases are used as shortcuts. So you may be fine. But I strongly advise to resolve aliases before moving things into any long-term storage/archive.

micahredding · February 7

@ctietze said:
I get uneasy when people use Finder aliases

You could, of course, use hard links or soft links. But it seems like people find aliases easier to see and reason about.

The reason I ask is because the whole issue with organizing files is trying to choose one particular hierarchy to be canonical. Zettelkasten teaches us not to do this with notes, it seems strange to defer to this with files.

And apparently, our file systems were actually built with the tools to do otherwise—to link things from multiple hierarchies, etc.

Yet, strangely, I’ve never seen anyone use these extensively.

Andy · February 8

@micahredding said:

@ctietze said:
I get uneasy when people use Finder aliases

You could, of course, use hard links or soft links. But it seems like people find aliases easier to see and reason about.

The reason I ask is because the whole issue with organizing files is trying to choose one particular hierarchy to be canonical. Zettelkasten teaches us not to do this with notes, it seems strange to defer to this with files.

And apparently, our file systems were actually built with the tools to do otherwise—to link things from multiple hierarchies, etc.

Yet, strangely, I’ve never seen anyone use these extensively.

Some years ago, Dr. Drang wrote a blog post, "Links as filesystem database" (2016), in which he described his use of multiple hard links to solve a file-organizational problem. At the end of the post, he explained further why he used hard links instead of aliases or tags in that case.

In my (non-Zettelkasten) files, I use symlinks extensively (instead of hard links because I need to link across volumes, which hard links can't do). I chose symlinks over aliases a couple of decades ago for this purpose because aliases wasted too much disk space back then (and Finder aliases are not sufficiently POSIX friendly, being an artifact of pre-Unix-based classic Mac OS from the 1990s). I haven't had any reason to regret the choice.

As Dr. Drang's post implies, it's important to think about which tool (aliases, hard links, soft links, tags) is best to solve a particular file-organizational problem.

My overall file organization scheme was influenced by Kerry Gleeson's book The Personal Efficiency Program, early editions of which date back to the 1990s. His recommendations for organizing electronic files are very similar to what Tiago Forte recommended a couple of decades later using the acronym PARA.

As I mentioned in the previous discussion on "Universal File Naming Convention" (November 2023), you may be interested in Seth Brown's 2011 blog posts on naming and searching files: Part 1 and Part 2. Part 1 explained the descriptors that Brown used in filenames (ID string/timestamp, tokens, keys, and file extension), and part 2 showed how he searched files. Brown claimed: "I hope the above search methods and the importance of developing a unified file naming system for your digital files are obvious. Once you have developed a trusted file naming system you will never be unable to locate files again."

There is also Douglas Barone's classic blog post "File System Infobase Manager" (2009) that documents his attempt to develop a kind of file UID system over 15 years ago (although it seems that this system was mostly for his notes, so perhaps not significantly different from a Zettelkasten). Barone mentioned that "the path from a card management system to an infobase is pretty straight and direct"!

Dr_Feelgoof · February 10

I do the UID thing. I use namechanger to prefix all my files with YYYY.MMdd.hhmm. Even google docs. It's like a dewey decimal system. But i can drill down to Year and month. So it works fine unless the file is buried in a deep folder structure. I just haven't retro-actively applied this to older archived files though. But when i need to link something to The Archive, it sure helps without bloating the media folder.

adilhassan · February 11

@iylock said:
How do you organize your files?

I'm thinking of attaching UIDs to all files like Zettelkasten and then creating structure notes.
However, I think that as the number of files increases, management becomes more difficult.

The reasons for considering this are as follows.
1) To write a description of the file.
You cannot write descriptions on sound files or picture files. You can use metadata, but it is sometimes a problem for me as I work while changing OSs.

2) To track relationships between files.
For example, if you want to describe the relationship between one photo and another, you inevitably need a txt file.
Files can also be grouped into different categories as needed, which is difficult to achieve because folders only provide a single structure. That's why I gave up on Johnny-Decimal.

Additionally, music, design projects, or programming work require multiple files, and I'm curious about your opinion on whether it would be useful to assign UIDs to all of them or to assign UIDs for each project.

Using UIDs for file organization, like in Zettelkasten, can be useful for tracking relationships and adding descriptions, especially across OSs. However, it might become complex as the number of files grows. A hybrid approach could work—assign UIDs at the project level rather than individual files and use a structured note or database to link related files. Tagging, metadata tools, or a note-taking app (like Obsidian) might also help manage relationships without making organization overly rigid. What tools are you currently using to manage your files?

Zettelkasten Forum

How do you organize your computer files?

Comments

Howdy, Stranger!

Quick Links

Categories

In this Discussion