── ── Decision-making · Decompose downward to bedrock
First Principles
First-principles thinking strips a problem down to the foundational truths that cannot be reduced further—physical law, definition, or verified fact—and rebuilds the answer using only those. It rejects reasoning by analogy or inherited convention, asking not what is normally done but what is actually, irreducibly true.
How it works
Take the claim or design you are evaluating and keep asking 'why is that true?' until each remaining statement is something you can prove independently: a law of physics, a contractual definition, a measured number. Discard everything that survives only because 'that's how it's done.'
Then reason upward from that bedrock, building the conclusion fresh from the verified pieces. Because no inherited assumption is smuggled in, the result is often radically different from—and cheaper or simpler than—the conventional answer.
When to use it
- A 'best practice' or industry standard is treated as untouchable and you suspect it inflates your costs or constraints
- Pricing a product where everyone copies competitors instead of computing the true cost-to-serve
- Entering a market where incumbents share an assumption no one has tested
- Your team keeps saying 'it can't be done' without naming the specific law that forbids it
When not to use it
For routine, low-stakes decisions where the conventional answer is already good enough—rederiving everything from scratch wastes time you don't have.
Worked example
The Wright Brothers, 1901
When their 1901 glider underperformed, the Wright brothers stopped trusting the published lift and drag tables every aeronautics pioneer relied on. They built their own small wind tunnel and measured the behavior of more than a hundred wing shapes themselves. The inherited data was wrong; their rederived numbers let them design a wing that actually flew.
Why it matters for founders
The most dangerous assumptions a founder inherits are the ones no one in the industry questions—the 'normal' margin, the 'standard' go-to-market, the cost everyone treats as fixed. First-principles thinking forces you to separate what is physically or economically true from what is merely customary, which is where defensible advantages hide. It is the same discipline deciqAI's agents apply: decompose the problem to verified facts before committing to a move.
Install this skill (free, MIT)
npx skills add deciqAI/knowledge-skillsFAQ
How is first-principles thinking different from just doing research?
Research gathers what others have concluded; first-principles thinking discards those conclusions and rebuilds from facts you can verify yourself. You use research to find the bedrock truths, not the inherited answers.
Isn't reasoning from first principles slower?
Yes, and that's the tradeoff. It is worth the cost only when an inherited assumption is expensive or wrong—on routine decisions, leaning on convention is the rational choice.