General thinking tools · 0 connections

First-principles thinking

If you cannot rebuild a belief from things you can verify yourself, you don't understand it — you've only received it.

Explanation

Take any belief you're confident about and try to rebuild it from scratch — from pieces you can verify, without borrowing the conclusion. Most of the time, the confidence deflates. What felt like understanding was a feeling of understanding, not the thing itself. First-principles thinking runs that test deliberately. It works in two moves: break the idea or problem down past every assumption that feels obvious but wasn't independently checked — past received rules, inherited wisdom, and "how it's done here" — until you reach something you can verify yourself. Then rebuild from those verified pieces. If the rebuild holds, you understood it. If it doesn't, you've found the gap. The same method handles hard questions where gut instinct would otherwise take over: break the question into parts you can measure or estimate, assign what you know to each, and reason up from there. People who do this on scored predictions outperform those who don't.

Watch

No video has yet earned its place here.

When it applies

Reach for it when the situation is genuinely new and no reliable pattern covers it, or when patterns you've been using are visibly failing. It applies when you catch yourself sorting a problem by what it looks like rather than how it works. If your answer to "why?" is "because that's the kind of thing it is," you're reasoning from a label, not a mechanism. It applies when a hard question would otherwise be settled by gut instinct. Break it into parts, assign what you can verify to each, and reason up from there.

Where it stops

It reaches its limit when you need a starting point, not a derivation. Every demonstration has to start somewhere, and that starting point cannot itself be shown to be correct from within the system it holds up. Look for a foundation below that level and the search loops.

The misuse

Rejecting inherited views, questioning everything, announcing a first-principles approach — this is the conspicuous part. The quiet, decisive part — rebuilding on premises you've actually verified, then testing the rebuilt model against reality — is where people stop early, and where stopping early leaves no mark. The second version is subtler. You break things down faithfully and halt at the first premise that feels basic — usually the most familiar one — and treat it as a foundation without testing whether it actually is.

A worked example

At Toyota's production plants, a machine jamming at a specific point in its cycle prompts the obvious question: what jammed? If the answer is a particular part — a worn bearing, say — the natural response is to fix it. The machine runs again. The standing instruction was to ask why five times before accepting any answer as the cause. A bearing wears because the lubrication interval is wrong for the actual load. The interval is wrong because the maintenance specification was inherited from the original equipment design and never updated when the production cycle changed. Fix the bearing and the machine runs for a while. Fix the maintenance specification and a class of failures across the line stops happening. The right depth to stop at is the depth where fixing the identified condition prevents recurrence — not the depth where you first find something you can explain.

Push

Before you act on a belief, try to rebuild it from things you can independently verify. If you can reconstruct it, rely on it. If you cannot, the feeling of understanding is not evidence of it — set the decision down until you can trace where the reconstruction fails, or at least name the step you can't supply.

Veto

Don't declare a root cause for a failure that involves multiple interacting conditions. Where many things have to go wrong together to produce an outcome, naming one cause as the answer is not a simplification — it's a misrepresentation. The confidence that single answer produces is proportional to how wrong it is.

Connects to

Connections appear as the deck grows.

Go deeper

Mastery question

**Question:** You decompose a belief, reach a premise that feels obviously basic, and rebuild from there with new confidence. What single thing must you check about that stopping point before the rebuild is worth anything — and why can't your confidence tell you whether you've checked it? **The answer:** You must check that the stopping point is genuinely something you can't break down further and verify independently — not just the most familiar place to stop. Your confidence can't confirm this, because a foundation is exactly what cannot be verified as one from inside the frame it supports. Feeling like you've hit bedrock is identical to stopping at an inherited assumption you've decided not to question. The only real test is external: can you derive the premise from something more basic, or does it hold up when you actively try to break it? If you haven't tried to break it, you haven't confirmed it's a foundation — you've only declared it one. **The answer that misses it:** "Check that the rebuild is logically consistent — that each step follows from the one before." Or: "Make sure you've questioned your assumptions." **Why the difference matters:** Consistency only tests whether the chain holds together — and a chain built from a false starting point is perfectly consistent and entirely wrong. Whoever has genuinely understood this concept knows the dangerous moment is the declaration that something is a foundation, not the reasoning that follows. And they know the danger is self-concealing: the more confidently the rebuild proceeds, the less likely the starting point was ever checked.