a builder's codex
codex · patterns · Stay inside the circle vs. ship into the unknown — Munger and Altman on opposite stances toward unknown territory

Stay inside the circle vs. ship into the unknown — Munger and Altman on opposite stances toward unknown territory

The contradiction

Charlie Munger's circle of competence says: name the boundary of where your knowledge gives you an edge, and refuse to operate outside it. Brilliance applied outside its circle becomes confident error; honest boundary-keeping compounds trust and decision quality over decades. Knowing what you don't know is more useful than being brilliant.

Sam Altman's iterative deployment philosophy (and the broader frontier-AI operating culture) says: ship into the unknown to learn. The circle of competence does not pre-exist for genuinely new categories — first-mover learning is what creates the circle. Waiting until you are inside-circle on a frontier problem means waiting forever; the only way to acquire circle-of-competence in a novel category is to act in it before you have it.

Why both can be right

The two stances apply to different cost structures of being wrong:

The contradiction is therefore conditional: which stance is right depends on the reversibility of the decision and the information content of being wrong.

How to resolve in practice

For any decision, ask:

1. Is the cost of being wrong recoverable? If yes (small bet, fast feedback, no reputational cliff), Altman wins — ship to learn. If no (irreversible commitment, public stance, capital concentrated), Munger wins — stay in-circle.

2. Does being wrong produce information that improves the next decision? If yes (the failure teaches you something legible), shipping is the cheap-tier learning path. If no (the failure looks like noise — random outcomes, unclear causes), waiting until you have a clearer model is cheaper than learning by attempting.

3. Is the circle definable yet? For mature categories, Munger's discipline is straightforwardly applicable. For frontier categories, Altman's stance is forced — there is no circle to stay inside; you are creating one.

Bezos's Type 1 / Type 2 decision framing is a useful third stance: high-reversibility decisions get fast-and-loose Altman treatment; low-reversibility decisions get deliberate Munger treatment, and the discipline is to know which type you are facing before deciding how to decide.

Implication for the codex

This is a productive tension worth holding open rather than resolving in one direction. Operators citing Munger's circle of competence as universal advice are mis-applying it to fast-iteration contexts. Operators citing Altman's ship-to-learn as universal advice are mis-applying it to irreversible-commitment contexts. The discipline is recognising which kind of decision you are in.

Sources

Open the interactive view →