Tagging Audit

Purpose

This document captures the state of the Torah Garden tag system as of March 2026, before any remediation. It serves as a baseline for measuring progress and as a diagnostic reference for understanding how the vocabulary grew organically.

Summary Statistics

MetricValue
Total documents381
Tagged documents248 (65%)
Untagged documents133 (35%)
Unique tags in use801
Singleton tags (used once)342 (43% of vocabulary)
Canonical tag definitions in About/Tags/17
Atlas/ files with tags0 of 57
Research/ files with tags~0 of 42

Coverage Gaps by Section

Atlas/ - 57 Files, 0 Tagged

The Atlas section is completely untagged. It uses a parallel metadata system (type:, category:, hebrew:, frequency:) that captures structured data well but is invisible to Quartz tag queries and the Obsidian tag browser. Three subsections are affected:

  • Divine Names/ (~20 files) - No tags: field despite having rich dh_source:, category:, and type: frontmatter
  • People/ (~25 files) - No tags: field; category: patriarch exists but doesn’t surface in tag views
  • Places/ (~12 files) - No tags: field; geography type is buried in prose

Why This Matters

A reader exploring the #patriarch tag finds nothing in the Atlas. The Abraham, Isaac, and Jacob files - the most-linked People entries in the vault - are invisible to tag-based navigation.

Research/ - ~42 Files, 0 Tagged

Research files (theonomastics studies, source analysis, chapter analyses) have no tags. This makes them undiscoverable by topic. A reader interested in #j-source finds the canonical definition but not the 42 research documents that apply that concept.

About/Tags/ - 17 Files, 2 Now Tagged

The tag definition files themselves were untagged at the time of audit. This is ironic: the canonical vocabulary lived outside its own system. The Tagging-Audit.md and Tagging-Guidelines.md files now carry tags: [methodology, overview] as a corrective example.

Books/ - 13 Chapters Untagged

Mostly index and overview files (e.g., book-level introduction pages). Chapter analysis files are generally tagged; the gap is in navigation/entry-point files.


Tag Vocabulary Problems

1. Singleton Tags (342 of 801)

Nearly half the vocabulary appears only once. These tags provide no discovery value - a tag only helps readers find related content when multiple files share it.

Examples of singleton tags from the vault:

  • personal-sovereignty
  • integrated-authority
  • covenant-election
  • divine-mediation-through-human-agency
  • covenantal-framework-for-legal-systems

These phrases describe a single document’s argument rather than a topic that recurs across the vault. They function as labels, not navigational nodes.

The Root Cause

Without a controlled vocabulary reference, each author (or each writing session) coins a new tag for a nuanced idea rather than selecting from existing terms. The result is a vocabulary that grows with the document count but never gains density.

2. Mixed Register: Source-Critic vs. Reader Tags

Two fundamentally different kinds of tags coexist with no distinction:

Source-critical tags (scholarly apparatus):

  • j-source, p-source, e-source, d-source
  • documentary-hypothesis, deuteronomistic-reforms

Reader-facing thematic tags (topical navigation):

  • covenant, sabbath, creation, priesthood

Both are valid and both should exist. The problem is that no file documents the distinction, so new tags can fall into either category without coordination. This leads to hybrid coinages like priestly-covenant-theology that belong to neither register cleanly.

3. Atlas Metadata Parallel System

Atlas files use rich structured frontmatter (type: person, category: patriarch, hebrew: אַבְרָהָם) that is more precise than any tag could be - but this metadata is only queryable via Obsidian Dataview or Bases, not via the standard tag system. The result:

  • Tag-based navigation misses all Atlas content
  • Atlas metadata is effectively siloed from the tag graph
  • The two systems don’t reinforce each other

The fix is not to remove the structured metadata but to add a minimal tags: field that bridges the two systems (e.g., tags: [patriarch, divine-name]).

4. Inconsistent Naming Conventions

The existing 801 tags include variations that fragment related content:

ProblemExamples
Capitalization mixedcovenant vs Covenant
Multi-word strategies inconsistentholy-war vs holywar vs holy war
Noun vs adjectivepriest vs priestly vs priesthood
Singular vs pluralgenealogy vs genealogies

The canonical definitions use lowercase, hyphen-separated slugs consistently. The content files do not always follow suit.


Failure Mode Categories

Type A: Absent Tags (Coverage Gap)

Files that should have tags but have none. Primary offenders: all of Atlas/, all of Research/, most of About/Tags/.

Impact: Content is invisible to tag-based navigation. The vault has 133 untagged files (35%) - more than a third of all content.

Type B: Over-Specific Tags (Singleton Explosion)

Tags that describe a single file’s argument rather than a recurring topic. These appear once and never help any reader discover related content.

Impact: The 342 singleton tags inflate the apparent vocabulary without adding navigational value. They also make the tag browser noisy.

Type C: Missing Canonical Coverage (~80% of Content)

Only 17 canonical tags are defined. The vault uses 801 unique tags. Even if 342 singletons were eliminated, ~440 tags would remain with no definition or rationale. Readers cannot know which tags are authoritative.

Impact: New files cannot be tagged consistently without a reference. Every author reinvents the vocabulary.

Type D: System Fragmentation (Atlas Metadata Silo)

The Atlas parallel metadata system captures excellent structured data but doesn’t connect to the tag graph.

Impact: Tag-based navigation and structured-metadata navigation are two separate islands. A reader can explore one or the other but not both simultaneously.


Remediation Scope

The three documents in this series address these failure modes:

DocumentAddresses
This auditBaseline measurement and failure mode taxonomy
Tagging-GuidelinesType A (rules for what to tag), Type D (Atlas bridging), naming conventions
Tag-VocabularyType B (singleton discipline), Type C (canonical vocabulary reference)

Tag remediation (actually adding tags to the 133 untagged files) is a separate effort tracked outside this document.


Methodology

Statistics in this document are derived from:

  • .dev/quartz/public/static/contentIndex.json - machine-generated index of all vault content
  • Direct file scan of Atlas/, Research/, and About/Tags/ directories
  • Frontmatter analysis of Books/ chapter files

The contentIndex.json is regenerated on each Quartz build and is the authoritative source for tag statistics.


Part of the Tags collection | About Portal