build-the-knowledge-tree — quality + safety report

---
name: build-the-knowledge-tree
description: Force any learning, onboarding, or "where do I start" decision through Musk's semantic-tree discipline — lock down the trunk and big branches (fundamental principles) before touching the leaves (details), or nothing the user reads will stick. Use this skill aggressively when the user is learning a new domain, designing an onboarding plan, building a self-study roadmap, ramping into a new codebase or field, or asking "where do I start with X", "give me a reading list for Z", "I'm drowning in tutorials", "I keep forgetting what I learn", "help me learn machine learning / rocketry / law / biology fast", "what's the prerequisite for understanding this", "I jumped straight into the advanced paper and I'm lost", or "design a curriculum for me". Also fires when a roadmap leads with frameworks before fundamentals, when an onboarding doc dumps tools before concepts, or when a learner is collecting leaves with no trunk. Trigger eagerly even when the user does not name Musk or the framework.
---

# Build The Knowledge Tree

> "It is important to view knowledge as a semantic tree. Make sure you understand the fundamental principles (the trunk and big branches) before you get into the leaves (the details), then there is something for them to hang on to."
> — Elon Musk, *The Book of Elon* (Chapter: Aspire to Be Less Wrong (semantic tree subsection))

## What this skill captures

Musk's learning model is structural, not motivational. Knowledge is a tree: trunk = the foundational axioms of the domain, big branches = the major subfields and how they connect, leaves = the specific facts, APIs, papers, formulas, and tooling. If a learner grabs leaves before the trunk exists, the leaves have nothing to hang on to and fall off — which is why most self-study fails despite heroic reading volume. Musk explicitly traces the intuition back to Civilization-style tech trees: "There are stages of technology and development of ideas. That's a helpful framework." You can't have democracy without literacy; you can't usefully read a Transformer paper without linear algebra and probability.

The value the user gets: a study plan, onboarding roadmap, or research-direction triage that is sequenced by dependency, not by enthusiasm or recency. Fewer abandoned tutorials. Higher retention. Faster genuine competence.

## When to use this skill

- The user is learning a new technical domain from scratch (ML, rocketry, distributed systems, law, biology, finance).
- The user is designing an onboarding plan for a new hire, intern, or themselves into a new codebase.
- The user asks "where do I start with X" or "what should I read first" and is about to be handed a flat list of links.
- The user has been consuming tutorials/papers/courses for weeks and complains nothing is sticking.
- A team is dumping tools, frameworks, and jargon on a newcomer before the underlying concepts are in place.
- The user is triaging a research direction and grabbing details before mapping the field.

## The how-to

1. Name the trunk explicitly before anything else.
   > "Make sure you understand the fundamental principles (the trunk and big branches) before you get into the leaves (the details), then there is something for them to hang on to."
   > — *The Book of Elon*
   Write down the 3-7 foundational principles of the domain in one sentence each. If the user can't state them, the rest of the plan is premature. For ML the trunk is: linear algebra, probability, optimization, generalization. For rocketry: Newtonian mechanics, the rocket equation, combustion, materials. No trunk, no tree.

2. Map the big branches and how they connect to the trunk.
   > "There are stages of technology and development of ideas. That's a helpful framework."
   > — *The Book of Elon*
   List the major subfields as branches off the trunk, and draw the dependency edges. A branch you can't connect to the trunk is either mislabeled or you don't understand it yet. This is the Civilization tech-tree move: you cannot unlock democracy without literacy.

3. Get the rough lay of the land before going deep on anything.
   > "Develop good general knowledge, so you at least have a rough 'lay of the land' of the full knowledge landscape. Read a broad range of material."
   > — *The Book of Elon*
   Do a fast, shallow sweep across all branches before diving. The encyclopedia move. Skip what doesn't interest you, but know it exists — otherwise you can't tell what's missing later.

4. Only now schedule the leaves, and hang each one on an explicit branch.
   > "Then there is something for them to hang on to."
   > — *The Book of Elon*
   Every paper, tutorial, library, or war story on the reading list must be tagged with which branch it hangs from. If a leaf can't be tagged, it's noise — defer it or cut it. This is the test that kills 80% of bookmark piles.

5. Optimize the data rate of how you ingest the trunk.
   > "Read books, because the data rate of reading is much greater than when somebody is speaking... The main reason I didn't go to lectures in college was because the data rate was too slow."
   > — *The Book of Elon*
   For trunk material, prefer dense books and primary sources over videos and lectures. Videos are fine for a leaf you need to install or operate; they're terrible for foundational concepts because the bitrate is too low to build structure.

6. Pressure-test the tree by talking to people who already have one.
   > "It's pretty straightforward — just read books and talk to people."
   > — *The Book of Elon*
   Once you have a draft tree, show it to a practitioner. They will instantly spot a missing branch or a leaf you've mislabeled as a branch. Musk learned rocketry by reading the Palo Alto library *and* calling experts to borrow old engine manuals. Both, not either.

## Common failure modes

- **Leaf hoarding.** Bookmarking 80 blog posts, papers, and YouTube videos with no trunk to attach them to. Symptom: "I keep forgetting what I learn." Musk's warning fits: the leaves have nothing to hang on to.
- **Skipping the lay of the land.** Diving straight into one subfield before knowing the shape of the whole landscape. Result: you can't tell what you don't know, and you can't tell when you're being lied to.
- **Trunk by analogy.** Declaring something is "fundamental" because a tutorial said so, not because it actually generates the rest of the field. The trunk must be load-bearing — if you remove it, the branches collapse.
- **Course-completion theater.** Working through a famous course end-to-end to feel productive, when half of it is leaves you don't yet have branches for. Discipline yourself to skip ahead or stop.
- **Confusing speed with progress.** Watching 2x-speed lectures because they feel fast. Lectures are a low-bitrate medium for trunk-building; speed doesn't fix the bitrate, it just buries you faster.

## When NOT to use this skill

- The user already has the trunk and is doing genuine leaf-level deep work (e.g., implementing a specific paper they already understand the foundations of). Don't re-explain the tree — get out of the way.
- The user is doing pure exploration with no learning goal — browsing for fun, idea-shopping. Imposing a curriculum kills that mode.
- The task is a specific, time-boxed how-to ("how do I configure nginx for X"), not a domain to be learned. Send them to the docs.
- The user is operating in a domain they already work in professionally and just needs a specific reference. They have a tree; they need a leaf.

## Source

The Book of Elon by Eric Jorgenson (2026, Scribe Media). Chapter: "Aspire to Be Less Wrong (semantic tree subsection)" (in "Think Like a Physicist").