Ask HN: How do you keep your files organized on macOS?

I find virtually every taxonomical system insufficient, whether it's a folder hierarchy, tag cloud, etc. Because of English's profusion of synonyms (on top of necessarily murky category definitions for the world, generally), finding a consolidating duplicates is its own war of attrition with myself. So I mostly store things in a handful of very broad topical folders and let mdfind (Spotlight) do all the heavy lifting of finding things I need. The filesystem hierarchy isn't “clean” in this case, but it doesn't matter because I rarely need to rely on it for finding what I'm looking for.

This system does not lend itself to concordance building, browsing, or evincing topical relations, but I'm rarely needing that outside the domain of a particular use case. So if I'm researching for a paper, for example, I use Zotero to map out connections between bibliographic references. It's a more constrained realm where tag management is less arduous.

I mostly use Linux, but I use a terminal 99% of the time, so it's irrelevant. I have the same setup on a Mac.

I have a few top-level directories, and a little structure beneath that — little enough that I remember most of it. My home computer directories are Code, Documents, $EmployerName, Media, Music, Photography, Temp (excluded from the backup), Web.

Yes, those are capital letters. I think they're neater, and it doesn't slow me down since I have several Zsh settings and functions to navigate quickly:

ls ~/d/t/p.j → Completes to ls ~/Documents/Travel/Passport\ scan.jpg

This was configured using the zsh "compinstall" function – case-insensitive and partial completions.

cd ~w → Completes to cd ~/Web

This is because I've done: hash Web=~/Web

This is the least useful, but could be useful for people who like deep directory structures.

cd n-w → Completes to cd ~/Code/new-website

This is because I've done: export cdpath=( ~/Code ) It's important not to put too much in cdpath, as it clashes with the contents of the current directory.

emacs m/m-s/s/m/j/o/e/ → Partially completes to: emacs ~/Code/maven-project/maven-s/s/m/j/o/e/

with the cursor after the s, because there are two possible directories. I can type "" to see the possible completions, or type "u" since I know that will complete to

   emacs ~/Code/maven-project/maven-sub/src/main/java/org/example/

I also have shell function for quickly finding by filename:

  f () {
        echo "find . -iname \"*$1*\""
        find . -iname "*$1*"
  }

I'm also using the UK Dvorak layout. Either UK layout puts ~ in a nicer place, and Dvorak puts / and - in a nicer place, so although it's less effort that it might appear to a US Qwerty user to type these so much.

In a big tree, no tags. It has taken a little while to figure out the shape of my life, but things now fit pretty nicely. I think the work of figuring out the one place where each thing fits is mostly valuable, not overhead. Having a deep tree allows me to incrementally organize files by pushing them partway down toward their final destination when I don't have time.

The biggest current challenge is all the tools that are trying to pull me away from a filesystem. Google Docs has pretty much been a disaster for organization, and I don't use--even for collaboration--unless I absolutely have to. Similarly, Evernote has lost its charm for me. I guess I'm in a minority here, preferring organizing to searching. It gets hard for me to make sense of old documents when they're out of context.

    - Documents
      - 1-Education
      - 2-Teaching
      - 3-Software-Engineer 
        - Unfold-Studio
          - unfold_studio
            - .git
      - 4-PhD
        - Research
          - Active-Projects
            - Text-And-Code
              - Studies
              - Publications
      - 5-Consulting
      - 6-Personal

I tend to use top level folder structure of:

    - Documents
      -- 
      -- 
    - Repos
      -- 

I also make extensive use of a little rust tool I whipped up which tags file paths with user defined tags, which are then searchable: https://gitlab.com/zwick/genie

This allows me to quickly find all file paths that are tagged with for instance regardless of where they are all from within a terminal.

Slightly off topic, but I'd like to argue for keeping your information in (backed up) local-filesystem files in the first place. I work with collaborators at several universities. Google seems to be pushing defaults toward institutional ownership of documents. I can see why institutions would like that, but there's no way I'll make my long-term access to files dependent on someone else's permissions (or Google randomly, unaccountably deleting my account).