Links vs tags vs folders: knowledge gardening for Obsidian, with Jorge Arango

TL;DR

Hypertext linking with automatic backlinks is the main scalability advantage for connected note-taking systems like Obsidian.

Briefing Cornell Notes

Briefing

A 10,000-note Obsidian vault doesn’t need a rigid taxonomy so much as a design that matches what each tool does best—especially the difference between links (relationships) and tags/folders (classification and state). Jorge Arango frames connected note-taking as an “extended mind” system: notes aren’t just storage, they’re part of how thinking happens, and hypertext-style linking is the core capability that makes large personal knowledge bases workable.

Arango’s starting point is practical: paper is resilient and always-on, but it’s weak at search and random access. Hypertext note-taking apps like Obsidian, by contrast, let users chunk ideas and connect them arbitrarily. The key feature isn’t just that notes can be linked; it’s that the system maintains backlinks automatically, so relationships become navigable. That matters when a vault grows over years into the thousands of notes—something that would be far harder to manage with paper-based indexing schemes.

From there, he draws a line between three organizational primitives: links, tags, and folders. Links are “pointers” to specific objects in the system—addresses you can jump to—while tags describe what a note represents, what set it belongs to, or what state it’s in. Folders, he argues, are for separation: each note can live in only one folder, which keeps the file system tidy but limits cross-cutting relationships. Tags are for uniting because a note can carry multiple tags, enabling many-to-many grouping across “containers.” He also notes that Obsidian’s folder-centric file structure can be managed with automation (he mentions an Auto Note Mover plugin) so tagged notes land in the right place without forcing every idea into a single hierarchy.

Tags are where many people go wrong, he says, because personal knowledge repositories evolve emergently. Top-down taxonomies work for websites and knowledge bases with known categories, but a personal vault often starts with one intention (work notes) and expands into reading, games, and other interests. That bottom-up growth can create tag sprawl and ambiguity—especially when the same term could mean different things in different contexts.

His remedy is disciplined minimalism: use tags for only three taxonomies—(1) type (“this note represents a person”), (2) set (“this note belongs to Duly Noted”), and (3) state (“draft,” “active,” “published”). With those constraints, he says, most organization needs can be met with one level of tags rather than deep nested hierarchies. When tags get crowded, he suggests switching from tags to “index notes” (a page that acts as a hub) and using links inside that hub to represent subtopics.

He also distinguishes capture from organization. During live listening (meetings, lectures), he avoids deciding whether something should be a link or a tag; that meta-work happens afterward. Finally, he warns that tool optimization can become a distraction: meta-work like plugin tinkering and template churn can steal time from actual thinking.

Overall, Arango’s approach is less about finding a universal structure and more about building a system that supports thinking: keep names human-readable, rely on plain-text compatibility when possible, and let links do the heavy lifting while tags and folders handle the limited jobs they’re best at.

Cornell Notes

Jorge Arango argues that scaling a personal knowledge system to thousands of notes depends on using each organizational mechanism for its best job. Hypertext-style linking (with backlinks) is the core advantage of connected note-taking apps like Obsidian, because it makes relationships navigable as the vault grows. He treats folders as “separating” containers (one parent per note) and tags as “uniting” metadata (many-to-many grouping), but warns that tags can become messy when they’re created bottom-up. His practical rule is to keep tags minimal and use them for three taxonomies: type (what the note represents), set (what it belongs to), and state (draft/active/published). Organization should also be separated from capture: decide links/tags later, not while listening live.

Why does Arango treat links as more fundamental than tags for large Obsidian vaults?

He frames links as the primitive that enables hypertext note-taking: notes can be chunked into ideas and then connected arbitrarily, with the system maintaining backlinks automatically. That backlinking turns relationships into something you can navigate, which becomes crucial when a vault grows toward ~10,000 notes over years. Paper systems can’t match this kind of random access and relationship tracking, and many other apps make linking less convenient than Obsidian’s double-bracket workflow.

What’s the practical difference between folders and tags in his model?

Folders are for separating: a note can only have one parent folder, which keeps the file system tidy but restricts cross-cutting relationships. Tags are for uniting: a note can carry multiple tags, enabling many-to-many grouping across those folder “silos.” He also notes that Obsidian’s file-system structure can be handled with automation (e.g., an Auto Note Mover plugin) so tagging can drive placement without forcing every idea into a strict hierarchy.

Why does he caution against “describing the content” with tags?

He says that using tags to summarize what a note is about (e.g., tagging a book with multiple topical tags like “advice,” “acting,” “bodybuilding,” “politics”) tends to create mess over time. In personal vaults, interests shift and tags get reused in conflicting ways. Instead of content-description tags, he recommends minimal tags that represent stable roles: type, set, and state.

How can nested tags become unnecessary in his approach?

He argues that deep hierarchical tags often become baroque. Instead, he recommends limiting tags to one level by using three taxonomies: type (e.g., “hashtag person” for notes representing people), set (e.g., “hashtag Duly Noted” to mark material relevant to a book), and state (e.g., draft/active/inactive). When he previously used hierarchical tags like project/active and project/inactive, he later realized “active/inactive” should be a state tag type that applies across multiple note types, not a child of “project.”

When should organization decisions like linking and tagging happen?

During live capture (meetings, lectures), he avoids deciding whether something should become a link or a tag because that meta-decision distracts from listening and capturing. Organization happens afterward, when the notes are processed. This aligns with a capture-vs-organize workflow: capture first, then reflect and structure.

Review Questions

What capabilities of hypertext note-taking (especially backlinks) make it more feasible to manage a vault approaching 10,000 notes?
How do Arango’s three tag taxonomies (type, set, state) reduce the need for nested tag hierarchies?
In what situations does he recommend postponing link/tag decisions until after live listening, and why?

Key Points

1
Hypertext linking with automatic backlinks is the main scalability advantage for connected note-taking systems like Obsidian.
2
Use folders to separate note types when you want tidy, predictable containers; use tags to unite notes across those containers.
3
Avoid creating tags that describe content in a free-form way; personal vaults evolve and that approach tends to produce ambiguity and tag sprawl.
4
Limit tags to three functions—type, set, and state—to keep tagging minimal and robust over time.
5
Don’t decide link-vs-tag while capturing live information; postpone organization until after the session to protect attention.
6
When tags get crowded, switch from deeper nesting to index notes (hub pages) that organize subtopics via links.
7
Treat meta-work (plugin/template tinkering) as a risk: optimize only enough to keep the system running so thinking time isn’t consumed by system maintenance.

Highlights

Arango’s core distinction: folders separate (one parent), while tags unite (many-to-many).

Tags stay manageable when they’re restricted to three taxonomies: type, set, and state.

He recommends capture-first, organize-later—linking and tagging decisions shouldn’t interrupt live listening.

Hypertext note-taking is viable at large scale because backlinks make relationships navigable without manual bookkeeping.

Topics

Obsidian Structure
Links vs Tags
Folders vs Tags
Tag Taxonomies
Knowledge Gardening