Maggie Appleton's Digital Garden: Six Patterns for Building a Personal Knowledge System

Table of content
Maggie Appleton's Digital Garden: Six Patterns for Building a Personal Knowledge System

Maggie Appleton is a designer, anthropologist, and developer who works as a Staff Research Engineer at GitHub Next. She trained as a cultural anthropologist, which shapes how she looks at technology through the lens of culture and human behavior. Her site maggieappleton.com is one of the most-cited examples of a digital garden, featuring hand-drawn illustrations, visual essays, and a taxonomy of seedlings, budding notes, and evergreen content.

Appleton developed her digital garden approach through work at GitHub Next, Elicit, and egghead.io. Her framework solves the core problem of knowledge work: how to organize, grow, and share what you learn without publishing only finished work.

The Six Digital Garden Patterns

Appleton documented these patterns in her essay A Brief History & Ethos of the Digital Garden. See Digital Gardens for a detailed breakdown of this approach.

1. Topography Over Timelines

Problem: Chronological feeds bury your best work.

Solution: Organize by relationships and context, not publish dates.

Implementation:

Example: Instead of “Posts from 2025,” create “Systems Thinking” that links related ideas across time.

2. Continuous Growth

Problem: “Published” implies finished. Nothing is ever finished.

Solution: Treat every note as a living document.

Implementation:

3. Imperfection & Learning in Public

Problem: Perfectionism kills publishing velocity.

Solution: Use growth indicators to show maturity level.

The Three Stages:

StageDefinitionPublish Threshold
SeedlingsRough notes, early ideas, open questions“Thought this once”
BuddingDeveloping thoughts, partial connections“Thought this 2-3 times”
EvergreenMature, well-tested, densely linked“Thought this 10+ times”

Implementation:

4. Playful, Personal, Experimental

Problem: Corporate blogs sound like corporate blogs.

Solution: Your garden, your rules.

Characteristics:

Maggie’s approach: Hand-drawn diagrams, illustrated concepts, anthropological lens on technology.

5. Intercropping & Content Diversity

Problem: Text-only limits expression.

Solution: Mix formats like a real garden mixes crops.

Content Types:

Tools Maggie Uses:

6. Independent Ownership

Problem: Platforms change rules, die, or lock you in.

Solution: Self-host on infrastructure you control.

Requirements:

Tradeoff: Higher setup cost, complete control.

The Evergreen Notes System

Appleton builds on Andy Matuschak’s evergreen notes framework. Four essential qualities:

1. Atomic Notes

Rule: One idea per note.

Bad: “My thoughts on AI, education, and digital gardens”

Good: “Language models work better as reasoning engines than answer machines”

2. Densely Linked

Rule: Connect by meaning, not category.

Implementation:

Tools: Obsidian graph view, Tana supertags, custom scripts.

3. Clearly Titled

Rule: Use imperative or declarative titles that state the claim.

Bad: “Thoughts on AI”

Good: “Treat language models as tiny reasoning engines”

4. Concept-Oriented

Rule: Focus on ideas, not events or people.

Bad: “What I learned at the conference”

Good: “Observable I/O builds trust in AI systems”

Maggie’s Current Tool Stack

ToolPurposeWhy She Uses It
TanaPrimary outliner and PKMFlexible structure, supertags, AI integration
ZoteroReference managementAcademic-grade citation tracking
ObsidianPrevious PKM (migrated from)Markdown files, local-first, extensible
ExcalidrawVisual diagrammingHand-drawn aesthetic, exportable SVG
MiroCollaborative mappingTeam workshops and concept mapping

Migration path: Obsidian (2019-2023) → Tana (2023-present)

AI Design Principles from “Squish Meets Structure”

Appleton’s talk Squish Meets Structure covers how to design products with language models. Four principles for AI tool design:

1. Treat Models as Tiny Reasoning Engines

Use as reasoning partners that show their work, not answer machines. Instead of “Write me a blog post,” use “Generate 5 different angles on this topic, explain your reasoning for each.”

2. Embrace Compositionality

Chain multiple small prompts instead of one massive prompt. Pattern: Input → Analysis → Options → Selection → Refinement. This makes reasoning transparent, components reusable, and errors isolated.

3. Avoid Outsourcing Complexity

Use AI to enhance reasoning, not replace it. “Language models can help us think more, not less.” Build tools that scaffold thinking instead of automating it away.

4. Prioritize Observable I/O

Show inputs, intermediate steps, and reasoning. “If you can’t see how it reasons, why would you trust its reasoning?” Expose prompt chains, confidence scores, alternative outputs, and result versions.

Key Insights

On publishing thresholds:

On organization:

On AI tools:

On ownership:

Start Your Own Digital Garden

Week 1: Foundation

  1. Choose a platform (Obsidian, Tana, or static site)
  2. Create your first 3 seedlings
  3. Link them together

Week 2: Systems

  1. Add growth indicators to your notes
  2. Write one evergreen note (atomic, linked, concept-oriented)
  3. Publish imperfect work

Week 3: Habits

  1. Tend existing notes instead of only creating new ones
  2. Link new notes to at least 2 existing notes
  3. Promote one seedling to budding

Month 2: Evolution

  1. Experiment with visual formats
  2. Build a topic map
  3. Remove all publish dates from primary navigation

Common Mistakes

Mistake 1: Waiting for the “perfect system”

Mistake 2: Organizing by date or category only

Mistake 3: Never updating published notes

Mistake 4: Treating AI as an answer machine

Resources

Maggie Appleton’s Work:

Related Frameworks:

Tools to Explore:

Next: Building with AI Tools — Maggie’s AI design principles in practice.

Topics: knowledge-management workflow open-source prompting