Information Architecture for this Site
Information Architecture for this Site
Core Principle: Flat with Metadata
All content lives at root. Type is frontmatter metadata (tags), not folder location.
Inspired by Wikipedia: every page is equal. Search and backlinks handle discovery. Folders impose a decision cost at creation time that isn't worth paying.
Structure
/ ← everything: notes, essays, concepts, ideas
economics.md
wise-society.md
info-architecture-for-this-site.md ← this file
/logs ← daily log only; date IS the identity
2025-10-20.md
2026-05-10.md
/projects ← discrete scoped work with its own lifecycle
/open-knowledge-foundation
Three buckets. Simple enough to hold in your head. New content → root, no decision required.
Exception: large topics. A topic with many sub-files (e.g. wto/) can have its own folder — but the folder is not a collection type, it's just overflow storage for one topic. The root entry (wto.md) still exists and is the primary page. Folder = size management, not taxonomy.
What Dies
Folders that imposed false taxonomy, now collapsed into root:
ref/— these were just dated notes; they belong at rootblog/,post/,writings/,nonfiction/,works/— same concept, five names
Filenames
- Concept/essay files: topic slug, no date —
economics.md,wise-society.md - Daily logs: date only —
2026-05-10.md - Date goes in frontmatter (
created:), not in the filename
Date in filename is only useful when the date is the primary identity (logs). For concepts, it's noise — it ages URLs and implies staleness.
Fleeting vs Evergreen
Don't decide at creation. Write to root. Things that matter grow through expansion and incoming links. Things that don't — stay stubs. That's fine.
Use #stub tag in frontmatter if you want a signal, but don't let it gate creation or determine location.
Migration
No big-bang migration needed. Gradually move ref/ files to root as they come up. Kill empty legacy folders as they empty out.