Use when the current approach feels fundamentally wrong, inherited constraints need questioning, or someone wants to reason from the ground up. Triggers on: "왜 이렇게 해야 해?", "기본부터 다시 생각해", "first principles", "가정을 의심해봐", 이 방식 자체가 맞는 건지", "처음부터 다시
A framework for stripping away assumptions, analogies, and inherited constraints to reason from the ground up. Based on two complementary traditions: Aristotle's philosophical method of decomposing to irreducible truths, and the practical reconstruction approach used by Elon Musk, Charlie Munger, and the Farnam Street mental models community.
Most thinking is analogy-based. First principles thinking is the exception — and the competitive advantage. Reasoning by analogy means "this is like that, so we do what others do." Reasoning by first principles means "what do we actually know to be true? What can we build from that?" The former is fast and usually wrong in novel situations. The latter is slow and usually right.
The foundation: There are two paths. Path A: Aristotelian decomposition — identify the primary premises, the most fundamental things that cannot be derived from anything more basic. If you can question it, it's not a first principle yet. Path B: Practical reconstruction — Musk's approach of breaking down something into its raw materials and costs, then asking: "Could we build this from scratch, given what we actually know?" Path C synthesizes both: decompose to truth (A), reconstruct from truth (B), then challenge the reconstruction with adversarial questioning.
Goal: 10/10. When analyzing a problem, assumption, or decision using first principles, rate the analysis 0-10. A 10/10 means the thinker has fully decomposed the assumption space (A), reconstructed from verifiable truths (B), and stress-tested the synthesis (C). Always provide the current score and specific improvements needed.
Core concept: Aristotle held that knowledge rests on "primary premises" — facts so fundamental they cannot be derived from anything more basic. First principles thinking means tracing any claim, assumption, or constraint back to these irreducible truths. If you can still ask "but why?" — you haven't reached a first principle.
Why it works: Most constraints are inherited — from convention, past decisions, or "that's how the industry works." Decomposition forces you to separate what is physically or logically true from what is merely conventional. This creates a clean foundation from which genuinely new solutions become visible.
Key insights:
Practice applications:
| Context | Pattern | Example |
|---|---|---|
| Cost assumption | Ask "what does this physically require?" | "Battery packs are expensive" → decompose to raw materials: cobalt, lithium, carbon — what do those actually cost? |
| Process assumption | Ask "what is the actual constraint?" | "Deployment takes 2 weeks" → what step actually takes 2 weeks, and is that step physically necessary? |
| Market assumption | Ask "what do customers actually want?" | "Users want feature X" → what outcome are they seeking? Is X the only path to that outcome? |
| Organizational assumption | Ask "why does this team exist?" | "We need a QA team" → what failure mode does QA solve? Could that be solved upstream? |
| Technical assumption | Ask "what is physically impossible?" | "We can't process real-time" → what is the actual latency constraint? Is it hardware, architecture, or convention? |
Core concept: After decomposing to raw truths, the reconstruction question is: "Given only what we know to be true, how would we build this from scratch?" This is Elon Musk's explicit method — applied to rockets, batteries, and transport. Charlie Munger's mental models serve as reconstruction tools: inversion, second-order effects, opportunity cost, compound interest.
Why it works: Reconstruction without inherited constraints produces solutions that look nothing like existing ones. SpaceX rockets look different because they were designed from physics, not from "how rockets have been built." The reconstruction phase is where radical cost reduction and novel architectures emerge.
Key insights:
Practice applications:
| Context | Pattern | Example |
|---|---|---|
| Cost reduction | Rebuild from raw inputs | "Battery at $200/kWh" → materials cost $80/kWh → gap is manufacturing overhead → that's where to focus |
| Architecture redesign | Ignore existing systems, design for the problem | "What if we had no legacy constraints? What would we build?" |
| Inversion | Name the failure modes | "What would make our product definitely fail?" → invert to derive success requirements |
| Second-order | Trace consequences forward | "We reduce price → more customers → higher support volume → quality drops → churn → price reduction fails" |
| Opportunity cost | Name what you're giving up | "Building feature X" → what 3 things could we build instead, and which is highest value? |
Core concept: The synthesis lens takes the decomposed truths (A) and the reconstructed solution (B) and stress-tests the combination. Where does the reconstruction rely on assumptions that survived Lens A's scrutiny? Where does it introduce new assumptions? What are the strongest counterarguments?
Why it works: Even rigorous first principles work can produce blind spots when the reconstructed solution is emotionally compelling. Lens C introduces adversarial distance — treating the reconstruction as a hypothesis to be falsified, not a conclusion to be defended.
Key insights:
Practice applications:
| Context | Pattern | Example |
|---|---|---|
| Assumption audit | List assumptions introduced in reconstruction | "We assumed manufacturing at scale — does that hold at year 1?" |
| Adversarial test | Strongest counterargument | "The smartest critic of this plan would say: ___" |
| Failure conditions | When would this be wrong? | "This approach fails if latency requirements are under 10ms — is that possible?" |
| Completeness check | What did we not decompose? | "We challenged cost but not regulatory assumptions — what does regulation actually require?" |
| Conviction calibration | What would change our mind? | "What evidence would make us abandon this reconstruction?" |
| Mistake | Why It Fails | Fix |
|---|---|---|
| Stopping decomposition too early | "Battery packs are expensive" is not a first principle — it's an observation | Keep asking "why?" until you reach physics, verified data, or logical necessity |
| Reconstructing within existing constraints | You decomposed but rebuilt the same solution | Explicitly list every constraint and ask which ones are truly fixed vs. conventional |
| Skipping Lens C | The reconstruction feels right, so stress-testing is skipped | Always find the three strongest counterarguments before finalizing |
| Mistaking convention for truth | "The FDA requires this" — but what specifically? | Decompose regulatory constraints: what exactly is mandated vs. interpreted practice? |
| Analogy creep in reconstruction | "This is like how Uber did it" — analogy re-enters | Call it out: "Is this a first principles reason, or are we back to analogy?" |
| Using first principles for routine decisions | Overkill for known problems with known solutions | Reserve for novel situations, large bets, or problems where existing solutions are failing |
| Question | If No | Action |
|---|---|---|
| Have you written down every assumption? | Assumptions are implicit and unexamined | Make a full assumption list before decomposing |
| Have you traced each assumption to a first principle? | You stopped at "because that's how it works" | Apply Lens A: keep asking "why?" until you reach physics or logic |
| Does the reconstruction start from the floor, not from existing solutions? | Reconstruction still looks like the incumbent | Explicitly ignore all existing solutions and build from the truths you found |
| Have you named what you're giving up? (opportunity cost) | Alternatives are invisible | List the top 3 alternatives and why this reconstruction beats them |
| Have you found the three strongest counterarguments? | The reconstruction is undefeated because unchallenged | Apply Lens C: argue against your own reconstruction |
| Do you know the conditions under which this is wrong? | You're defending rather than understanding | Name the specific conditions that would falsify the reconstruction |
This skill synthesizes insights from several foundational sources:
brainstorming — 가정을 분해한 후 바닥부터 새 아이디어를 발산하며 설계하고 싶을 때problem-reframer — 분해 과정에서 문제 자체가 잘못 정의됐음을 발견했을 때decision-maker — 재구성한 대안들 중 하나를 구조적으로 선택해야 할 때First principles reasoning originates with Aristotle, who argued in the Posterior Analytics that all knowledge rests on primary premises that are true, necessary, and immediate. The method was revived in modern practice most visibly by Elon Musk, who described it publicly in a 2013 TED interview: "I think it's important to reason from first principles rather than by analogy... you boil things down to the most fundamental truths and then reason up from there." Charlie Munger's lifelong practice of building a "latticework of mental models" is the parallel tradition in investing and decision-making. Shane Parrish's Farnam Street community has systematized both traditions into a practical curriculum for everyday decisions.