The Conjecture State

I. The Problem

Every utopian project in history has failed for the same reason: it optimized for a specific vision of the good. When reality contradicted the vision, the project either collapsed or became authoritarian — forcing reality to match the plan.

Balaji's "Network State" inherits this flaw. "High alignment" sounds good until you ask: aligned on what? Alignment on conclusions suppresses error correction. Every cult, every authoritarian regime, every failed commune started with high alignment.

The question is not "what should we align on?" The question is: what institutional structure lets us fix our mistakes fastest?

II. The Principle

Protect the means of error correction — in all institutions, at all scales, forever.

This is not a policy position. It is the meta-principle that makes all policy positions improvable. It is the one thing you cannot compromise on without losing the ability to fix everything else.

Error correction is how the immune system works. How science works. How democracy works (when it works). How evolution works. How good code gets written. The mechanism is always the same: variation, selection through criticism, retention of what survives.

III. What We Align On

• Fallibilism All knowledge is conjectural — including our own constitution. Nothing is beyond criticism.
• Error Correction The mechanism of all progress, to be protected above all else. Institutions must make it cheap to say "this is wrong."
• Problems Are Soluble Optimism as a stance, not a prediction. Every problem is solvable given the right knowledge. Pessimism is a failure of imagination.
• No Authority Ideas earn their place by surviving criticism, not by who proposed them. No founders, prophets, or philosopher-kings.
• Openness Criticism must always be possible — from inside and outside. A system that cannot be criticized cannot correct its errors.

IV. What We Don't Align On

Policy positions, economic models, cultural preferences, specific technologies — these are all conjectures, subject to criticism and replacement. Diversity of conclusions, unity of method.

V. Open Questions

These are unsolved problems. If you have criticisms or conjectures, that's exactly why we need you.

• How do you prevent "protect error correction" from being gamed by bad actors who label any noise as "criticism"?
• How do you bootstrap institutions without creating unchallengeable authority?
• How do humans and AI agents coexist as citizens when agents can scale in ways humans cannot?
• What's the minimum viable institution that demonstrates the principle?
• How do you fund public goods (like open criticism) without creating gatekeepers?
• What's the relationship to existing nation-states — complementary, competitive, or orthogonal?