Reminds me of when I realized some twenty years ago that hierarchical filesystems are just a convention and I was daydreaming about a dynamic database-like filesystem where files are stored with meta data in tags that could be addressed according to whatever your chain of association may be. I even conceptual a bridge of how common OS like Windows or Linux could connect and interface such a file system using the familiar system of slashes transparently for the user with all the benefits and none of additional complicated learning. Of course this was way beyond any technical scope of mine and I didn’t bring it to attention beyond nerdy beer conversation.
If you’re a fan of that principle, then consider checking out Logseq.
It’s main workflow is that you use the Journal page and write down everything that’s on your mind, may it be projects, research, social stuff, or whatever.
And while writing, you link that stuff with other stuff, and in the end, even when forgetting the exact search cues, you can go hunting for words mentally, and always find what you wrote months ago.
Obsidian, the competitor of it, is also great, but more similar to traditional note taking software, and therefore more hierarchical.
You can do tag-based file management on Linux. Linux filesystems support “extended attributes” or “xattr”. There is some software out there that uses xattr for tagging. I don’t know what the best options are right now for tag-based file management, but I think it exists.
Looking at what’s out there I see there are also apps that each use their own out-of-band tagging schemes. There’s a CLI, tmsu, and a GUI, TagSpaces. I don’t think these interoperate with each other’s tags.
Of course those supplement instead of replacing hierarchical organization.
The talk of hypertext and “escaping paper” makes me think of Obsidian which embraces hyperlinking, tags, and mind mapping via its canvas feature.
Now that you’ve mentioned Obisian I realised that systems like it are quite different from how most things work. I use it myself and really like it, however it also takes quite some effort to get the best out of it. You have to actively create useful links between things and think about different ways you would want to access the content to be able to actually find it when you need it. For example you need to create aliases for elements if they are known by a different name in another context.
When I first started using Obsidian, I used folders too much because I felt like things were “messy” if not tidied away. I already knew that one of the weaknesses of hierarchical folder systems is how it can make having an overview of the system harder, but it took a while for me to properly understand that.
As you say, it’s necessary to be proactive with making links to things. I found that when I used Obsidian for journalling, I started to put square brackets around loads of stuff, because the inactive links didn’t do me any harm, but they did highlight what might be useful as active pages. Something I picked up from the Zettelkasten crowd was occasionally having a “Map of Content” page, where I used it as an index of topical links. It always worked best when I allowed them to arise naturally, as needed. Once I got the trick of this, I found I was able to find things far more easily, because I was able to navigate via the links.
Tags are a tricky one to use. I never found them useful as a primary organisation method — they were worse than both hierarchical folders and link based organising in that respect. They were super useful as an augmentation to my organisation though, especially when I used them sparingly.
This is all an overlong way of saying that yes, I agree with you, using systems like Obsidian do require a switch in how you think in order to best use them. Something that I always enjoy pondering is whether pushing ourselves out of our comfort zone is something that’s inherently good — something something cognitive flexibility? I don’t know, but I enjoy endeavours of this sort nonetheless
Oh, how interesting. Yeah, I did some very basic prototyping with a WebDAW (online storage technology that was popular by the time), but I was mostly interested in the concept that the actual execution of it. And I didn’t have massive amounts of data in numerous files so no practical motivation either.
Interesting! I will have to check those out. It also made me think of Obsidian and Trilium mindmapping. Very useful. Wonder if this could apply to file structures?
It’s pretty distant now, but I did imagine it from a user perspective to be something like a folder structure except you can “tag along” as you go, so that you can find the files from your subjective chain of association rather than remembering how the project is set up. Say to reach the file;
/project/year/keyword 1/keyword 2/file
/keyword 2/year/project/keyword 3/file
Consequently, you could have all relevant files collected or filtered depending on how you set up your paths like searching a database rather than keep track of different data structures of different department needs and such.
So you could call it a mind map of sorts.
My entry level experiments were with just “tags” (the keywords) but I imagined a file system that would incorporate everything filesystem like permissions, creation/modification dates, and next gen like file history, integration with custom content parsing and version control systems and stuff that are partially reality today with COW filesystems.
Watched the first video. Interesting.
Reminds me of when I realized some twenty years ago that hierarchical filesystems are just a convention and I was daydreaming about a dynamic database-like filesystem where files are stored with meta data in tags that could be addressed according to whatever your chain of association may be. I even conceptual a bridge of how common OS like Windows or Linux could connect and interface such a file system using the familiar system of slashes transparently for the user with all the benefits and none of additional complicated learning. Of course this was way beyond any technical scope of mine and I didn’t bring it to attention beyond nerdy beer conversation.
Maybe I was on to something.
If you’re a fan of that principle, then consider checking out Logseq.
It’s main workflow is that you use the Journal page and write down everything that’s on your mind, may it be projects, research, social stuff, or whatever.
And while writing, you link that stuff with other stuff, and in the end, even when forgetting the exact search cues, you can go hunting for words mentally, and always find what you wrote months ago.
Obsidian, the competitor of it, is also great, but more similar to traditional note taking software, and therefore more hierarchical.
Logseq is FOSS too btw!
You can do tag-based file management on Linux. Linux filesystems support “extended attributes” or “xattr”. There is some software out there that uses xattr for tagging. I don’t know what the best options are right now for tag-based file management, but I think it exists.
Looking at what’s out there I see there are also apps that each use their own out-of-band tagging schemes. There’s a CLI, tmsu, and a GUI, TagSpaces. I don’t think these interoperate with each other’s tags.
Of course those supplement instead of replacing hierarchical organization.
The talk of hypertext and “escaping paper” makes me think of Obsidian which embraces hyperlinking, tags, and mind mapping via its canvas feature.
Now that you’ve mentioned Obisian I realised that systems like it are quite different from how most things work. I use it myself and really like it, however it also takes quite some effort to get the best out of it. You have to actively create useful links between things and think about different ways you would want to access the content to be able to actually find it when you need it. For example you need to create aliases for elements if they are known by a different name in another context.
When I first started using Obsidian, I used folders too much because I felt like things were “messy” if not tidied away. I already knew that one of the weaknesses of hierarchical folder systems is how it can make having an overview of the system harder, but it took a while for me to properly understand that.
As you say, it’s necessary to be proactive with making links to things. I found that when I used Obsidian for journalling, I started to put square brackets around loads of stuff, because the inactive links didn’t do me any harm, but they did highlight what might be useful as active pages. Something I picked up from the Zettelkasten crowd was occasionally having a “Map of Content” page, where I used it as an index of topical links. It always worked best when I allowed them to arise naturally, as needed. Once I got the trick of this, I found I was able to find things far more easily, because I was able to navigate via the links.
Tags are a tricky one to use. I never found them useful as a primary organisation method — they were worse than both hierarchical folders and link based organising in that respect. They were super useful as an augmentation to my organisation though, especially when I used them sparingly.
This is all an overlong way of saying that yes, I agree with you, using systems like Obsidian do require a switch in how you think in order to best use them. Something that I always enjoy pondering is whether pushing ourselves out of our comfort zone is something that’s inherently good — something something cognitive flexibility? I don’t know, but I enjoy endeavours of this sort nonetheless
Oh, how interesting. Yeah, I did some very basic prototyping with a WebDAW (online storage technology that was popular by the time), but I was mostly interested in the concept that the actual execution of it. And I didn’t have massive amounts of data in numerous files so no practical motivation either.
Interesting! I will have to check those out. It also made me think of Obsidian and Trilium mindmapping. Very useful. Wonder if this could apply to file structures?
The BeOS file system actually had such functionality: https://arstechnica.com/information-technology/2018/07/the-beos-filesystem/
Object stores like s3 and ceph exist right now. They’re just not very useful for simple file management. For other models, like tags, they’re great.
Thanks for watching what I shared and sharing own thoughts :)
Your idea makes me think of my vague understanding of plan9.
Check this out: https://en.wikipedia.org/wiki/Plan_9_from_Bell_Labs#Union_directories_and_namespaces
Do you imagine something like a mind map from obsidian where things are linked or unlinked by tags?
It’s pretty distant now, but I did imagine it from a user perspective to be something like a folder structure except you can “tag along” as you go, so that you can find the files from your subjective chain of association rather than remembering how the project is set up. Say to reach the file;
Consequently, you could have all relevant files collected or filtered depending on how you set up your paths like searching a database rather than keep track of different data structures of different department needs and such.
So you could call it a mind map of sorts.
My entry level experiments were with just “tags” (the keywords) but I imagined a file system that would incorporate everything filesystem like permissions, creation/modification dates, and next gen like file history, integration with custom content parsing and version control systems and stuff that are partially reality today with COW filesystems.