Collective intelligence, by choice
Top-down institutions were built for a world where every actor was a person and effort was slow.
5 October 2026 · 7…
Why this matters now
Top-down governance assumes one voice…
Top-down governance assumes one voice comes from one person at human speed; capable AI breaks that assumption unless each person has a personal agent to participate and verify on equal terms.
Inspect technical derivation & primary-source evidence ▾
Public consultations can be flooded, identity checks can be passed by synthetic applicants, and the people meant to review all this cannot keep up. We do not think the answer is to make top-down systems bigger, faster or more watchful. We think the answer is to give each person better tools to take part, and to let institutions that use those tools well earn people's trust in the open.
Your agent as your participant and shield
Your personal agent ( running…
Your personal agent (running on your own device) translates complex proposals into plain impacts for your household, never votes without your explicit permission, and acts as an anti-coercion filter against every governance system—including ours.
Inspect technical derivation & primary-source evidence ▾
- It helps you take part on equal terms. Everyone's agent has the same standing, so a well-funded lobby gets no louder voice by owning more software.
- It tells you when something departs from your needs. When a proposal, rule change, or budget cuts against what matters to you, it flags it with its reasons.
- It speaks only with your permission. It drafts; you decide.
- It acts as an abuse filter. If an institution pressures you or buries a demand in fine print, your agent points it out and helps you refuse.
Values by voice, facts by track record—and voluntary migration
We strictly separate moral choices…
We strictly separate moral choices (decided one person, one equal voice) from factual predictions (weighed by verifiable forecasting track record and sandbox simulations), allowing member-owned co-ops to win purely by voluntary choice.
Inspect technical derivation & primary-source evidence ▾
- Values are decided one person, one voice. Agents help people express their values; they do not get votes of their own.
- Factual questions are weighed by forecasting track record. Forecasts of those who have been right before on verifiable questions count for more.
- Rewarding institutions that open up & Outcompete, never overthrow: New co-operatives and DAOs sit alongside existing institutions; people move only when they choose, with full data portability.
What others have learned & how this could fail
We build on deliberation research on Polis (Small et al., 2023), collective constitutional input (Huang et al., 2024), and generative social choice (Fish et al., 2023).