First Principles Thinking
A rocket is not expensive because aluminum forgot how to be cheap. First principles thinking asks what the thing is made of, what each part costs, which constraints are real, and which constraints are inherited theater. The phrase goes back to Aristotle's archē, a starting point that is not derived from another claim. The modern danger is that the phrase now appears in pitch decks where nobody has decomposed anything.
How it works
Analogical reasoning says: this problem resembles an old problem, so copy the old solution and adjust. That is how most human judgment works, and it saves time. First principles reasoning is slower: break the problem into claims, test which claims are directly verifiable, then build the answer back up.
The SpaceX example is sticky because it put numbers under a sacred industry. Elon Musk has said he looked at rockets as piles of aluminum, titanium, copper, carbon fiber, electronics, labor, and assembly rather than as mystical $60 million objects. The exact arithmetic is debated, but the move was clear: compare the market price of the finished artifact with the input cost of its materials and processes.
That does not prove rockets should be cheap. It proves the right question is not "what do rockets usually cost?" The right question is "which part of this cost is physics, and which part is procurement, regulation, failure tolerance, monopoly, risk pricing, and habit?"
The operating procedure
A useful version has 5 steps:
- State the question in measurable form. Not "how do we improve hiring?" but "which observable signals predict 12-month performance better than interviews?"
- List every assumption inside the current answer.
- Sort claims into directly testable, derived, and unknown.
- Verify the testable claims with numbers, artifacts, or experiments.
- Rebuild the answer and inspect the gap between the rebuilt answer and the conventional answer.
The gap is the prize. If the rebuilt answer differs by 5%, convention may be fine. If it differs by 10x, someone is either wrong or sitting on a business.
Where it pays
First principles thinking works best when convention has had time to calcify.
| Domain | Conventional question | First-principles question |
|---|---|---|
| Pricing | What do competitors charge? | What does delivery actually cost at each unit? |
| Engineering | How is this usually built? | Which physical constraints must be obeyed? |
| Careers | What path did successful people take? | Which compounding mechanisms are available to me? |
| AI systems | Which model is popular? | What task, data, latency, and failure cost define the system? |
Richard Feynman's reputation partly comes from this habit. During the Challenger investigation in 1986, he did not begin with institutional language. He put a piece of O-ring material in ice water during a televised hearing and showed that it lost resilience at low temperature. The demonstration did not explain the whole disaster, but it cut through pages of abstraction.
What's contested
The hard part is knowing when you have reached bedrock. A founder may say "the customer wants speed" when the deeper fact is "the customer wants fewer status meetings." An engineer may say "the model needs more data" when the deeper fact is "the loss function rewards the wrong behavior."
There is also a cost problem. Decomposing every decision is intellectual vanity. Analogy is not the enemy; it is the default tool. First principles thinking earns its cost only when the decision is rare, expensive, path-dependent, or trapped inside inherited assumptions.
Why this has to do with other realms
In overview sanskrit grammar, Pāṇini compresses Sanskrit into roughly 4,000 rules. That is first-principles behavior before the modern phrase existed: reduce surface variety to generative machinery. The same instinct appears in concept computation, where a messy activity becomes tractable once you ask what operations are sufficient to produce it.
The bridge to mission voyager 1 is less obvious. Voyager 1 did not become immortal by being maximal. It became durable because engineers reduced the mission to constraints that mattered: power, mass, communication, trajectory, redundancy. First principles thinking is not "ignore history." It is asking which parts of history are load-bearing.
An open question
If first principles thinking is expensive, what is the reliable signal that a problem deserves it before the payoff is visible?
Key Sources
- Aristotle, Metaphysics - classical source for first principles as starting points of knowledge.
- René Descartes, Discourse on the Method (1637) - reduction and reconstruction as a method of inquiry.
- Richard Feynman, Surely You're Joking, Mr. Feynman! (1985) - useful record of physical intuition and decomposition.
- Report of the Presidential Commission on the Space Shuttle Challenger Accident (1986) - source context for Feynman's O-ring demonstration.
- to verify: exact original interview source for Elon Musk's rocket raw-material cost breakdown.
Further Reading
- person feynman - the clearest human example of cutting through formal language with a test.
- concept decision frameworks - when decomposition is worth the time and when heuristics win.
- Thinking, Fast and Slow by Daniel Kahneman - why analogy and shortcuts are not bugs in human cognition.
- The Beginning of Infinity by David Deutsch - why explanations matter when rebuilding from primitives.
Abhishek's take
I use this when a vendor says a style cannot land before a 100-day lead time. I break the delay into fabric, dyeing, approval, booking, and warehouse handoff, then ask which step is physics and which step is habit. The answer often changes the buy from “drop the style” to “change the fabric and keep the bet.”
See Also
- person feynman
- concept decision frameworks
- concept computation
- overview sanskrit grammar
- mission voyager 1