apply-first-principles

Apply First Principles

Decompose any problem to bedrock truths and rebuild solutions unconstrained by inherited convention.

Why This Is Best Practice

Adopted by: SpaceX, Tesla, Amazon (Bezos), Basecamp, McKinsey — and rooted in Aristotle's epistemology and Descartes' method of systematic doubt, both foundational to Western scientific reasoning.

Impact: SpaceX applied first principles to rocket manufacturing: instead of buying rockets at market price (~$65M/launch), Musk's team priced raw materials (carbon fibre, aluminium, titanium, copper) at commodity rates — achieving ~$6M/launch, a 10× cost reduction (Musk, TED 2013).

Analogical thinking — reasoning from "how others do it" — imports all of an industry's historical constraints and compromises. First principles cuts through accumulated convention to reach what is actually true, enabling solutions that would be invisible to anyone anchored to precedent.

Steps

1. State the problem precisely — write it in one sentence without referencing any solution.
2. List every embedded assumption — enumerate what the current approach takes for granted (cost, materials, process, timeline, regulation, user behavior). Write them all down before questioning any.
3. Interrogate each assumption to bedrock — ask "Is this actually true?" and "What would have to be true for this to be necessary?" Stop only when you reach physics, verified data, or a logical identity. "Industry standard" is not bedrock.
4. Identify which assumptions are false or contingent — mark assumptions that failed interrogation. These are your leverage points.
5. Rebuild the solution from verified fundamentals only — construct an approach using only what survived step 3. Ignore inherited form.
6. Compare rebuilt vs. inherited approach — the delta is the opportunity. Implement the delta.

Rules

• Never start from analogy ("how does everyone else do this?"). Start from the problem.
• Write assumptions down before questioning them — unlisted assumptions go unchallenged.
• Challenge one assumption cluster per session; depth beats breadth.
• Stop interrogation when you hit physics, mathematics, or verified empirical data — those are bedrock.
• The rebuilt solution may look absurd by industry standards. That is expected, not disqualifying.

Examples

SpaceX rocket cost:

• Inherited assumption: "Rockets cost $65M because aerospace is expensive."
• First principles: Price each raw material at commodity market rates. Carbon fibre + aluminium + copper + titanium = ~$2–3M of materials. The rest is markup, inefficiency, and legacy supply chain.
• Rebuilt solution: Manufacture in-house from raw stock. Launch cost drops to ~$6M.

Software architecture:

• Inherited assumption: "We need a microservices architecture because our team is large."
• First principles: What problem does microservices solve? Independent deployability. Do we actually need independent deployability? Interrogate whether the team structure requires it or merely suggests it.

Common Mistakes

• Confusing first principles with brainstorming — brainstorming generates options; first principles eliminates false constraints. Different operations.
• Stopping at "industry best practice" — industry practice is a conclusion reached by others under their constraints, not a ground truth. It is always an assumption to interrogate.
• Skipping the written assumption list (step 2) — assumptions not written down survive interrogation by staying invisible.
• Rebuilding too quickly — the value is in the interrogation phase (steps 2–4), not the rebuilding. Rushing to "rebuild" before fully questioning reinstalls old assumptions under new labels.