First Principles
How do you cut through bullshit to validate the truth, and apply force where it creates the greatest step improvement?
Strip away received wisdom. Reason from what you can see, test, and validate yourself.
Most thinking is analogy — copying what others did. First principles creates what didn't exist.
| Reasoning by Analogy | Reasoning by First Principles |
|---|---|
| "This is how it's done" | "What's actually required?" |
| Incremental improvement | Step-change innovation |
| Constrained by precedent | Constrained only by physics |
| Fast but local | Slow but global |
The Method
"The first basis from which a thing is known." — Aristotle, Metaphysics
| Step | Action | Question | Technique |
|---|---|---|---|
| 1 | Identify assumptions | What do I believe is true? | Five Whys — ask "why?" until you hit bedrock |
| 2 | Decompose to fundamentals | What are the irreducible components? | Socratic questioning — expose hidden constraints |
| 3 | Rebuild from scratch | Given only these truths, what's optimal? | Physics test — does it violate natural laws? |
| 4 | Simplify | What can I remove without breaking it? | Delete until it breaks, then add 10% back |
Example: SpaceX rockets. Industry price: $65M. Musk asked: "What are rockets made of?" Aluminum, titanium, copper, carbon fiber. Raw materials = 2% of price. Built in-house. Cut costs 90%.
Naming First
The first principle of first principles: how do you name and categorize things by their purpose? This is nomenclature (naming) and taxonomy (classification). Without naming standards, you can't decompose to fundamentals — you don't know what you're looking at.
Applied: Optimisation
Musk's 5-step algorithm — first principles applied to process optimisation. Steps must be done IN ORDER.
| Step | Action | Principle |
|---|---|---|
| 1 | Question requirements | Your requirements are definitely dumb |
| 2 | Delete parts/processes | If you don't add 10% back, you didn't delete enough |
| 3 | Simplify & optimize | Don't optimize what shouldn't exist |
| 4 | Accelerate cycle time | Don't dig your grave faster |
| 5 | Automate | Only automate what survived steps 1-4 |
Most engineers start at step 5. That's the mistake. For the full improvement loop — document, measure, analyze, improve, standardize — see Process Optimisation.
First Principles Across the Site
Every domain has bedrock truths. This page teaches the method. These pages apply it.
| Domain | First Principles | What They Ground |
|---|---|---|
| Self | Agency | Psychology, philosophy, foundations, goodwill |
| Business | Business | 15 principles by lifecycle stage |
| Forces | Crypto | Money, data, energy, stories — how value flows |
| Science | Science | Physics, biology, mathematics — pure inquiry |
| Crypto | Crypto Principles | Composability, SSI, permissionless, verifiable truth |
| Optimisation | Process | Musk's 5-step algorithm |
| Naming | Standards | First principle of first principles |
| Knowledge | KB Stacking | Primitives to platform |
| Navigation | Tight Five | Principles = Question #2 |
| Biases | Behavioural Biases | Where humans systematically fail |
Context
- Navigation System — Value, Belief, Control as the three loops
- Knowledge Stack — Primitives → Protocols → Standards → Platform
- The Tight Five — Principles is the first of five tightly-bound elements
- Mantra — When systems fail, principles become the trigger phrase
- Inversion Thinking — Flip the problem
- Scoreboard — Principles tell you where to point; the scoreboard tells you where you are
Questions
What received wisdom are you treating as first principles without having decomposed it?
- Which domain's first principles do you know intellectually but fail to apply under pressure?
- Where would stripping to fundamentals reveal that the problem you're solving shouldn't exist?
- What assumption would invalidate everything you've built if it turned out to be wrong?