Trust, Values & Intent

A seven-part series on what “value” really means in a democracy where AI takes a growing role — and how citizens’ assemblies rebuild the ground under it. Plain-language throughout; about 41 minutes end to end.

Politics keeps offering us “values” that would forbid no decision, while the machines we are building can out-argue us on everything except what matters. This series pulls those two problems apart, one plain idea at a time, and ends with something you can build: a room designed to hold disagreement rather than fake agreement. Read it in order, or drop in anywhere — each part says where it sits and where it is going.

The seven parts, in order

  1. Values That Never Say No · ~7 min
    Most “values” — on office walls, in party charters, in mission statements — are pleasant leanings that would forbid no real decision, so they guide none. Test any of them with one question: name one decision this would have forbidden. The answer to hollow values isn’t better words; it’s rooms that make the test unavoidable — which is where this series is going.

  2. The Stranger Who Never Dies · ~6 min
    We never trusted people for their values; we trusted the shared conditions — a body, mortality, kin, skin in the same game — that made betraying them costly. A new kind of agent has none of those, so the old trust instinct fires at nothing. The sane response isn’t to trust it harder but to stop needing to: check its working instead.

  3. Who Made Up Your Mind? · ~5 min
    Democracy counts your preferences on the assumption they’re yours; but much of what you want is now built for you, before you ever vote. That hollows the ballot at its source. Rooms that give people time, checkable evidence, and each other are slow and hard to manufacture — which is exactly why they matter.

  4. Smarter Is Not in Charge · ~6 min
    No amount of knowledge tells you what you ought to do: a machine that knew everything still couldn’t say what matters. Its brilliance isolates the values question rather than answering it, leaving legitimacy — the right to decide — more ours, not less. The real danger is handing that over out of convenience; the guard against it is a plain rule — the software rehearses, people decide.

  5. A Room Built for Disagreement · ~6 min
    Lasting disagreement between good people over good things is normal, not a failure — so a good room hosts it rather than faking consensus. A room built for that must do five things: keep the disagreement verbatim, show its working, mark every AI touch, leave absent voices empty, and decide nothing itself. Design for disagreement.

  6. Don’t Trust Us — Check the Working · ~6 min
    One working build, audited against piece 5’s five requirements — including where it falls short. It refuses to fake agreement (that’s a runtime error, not a setting), leaves barred seats empty, decides nothing itself, and seals a tamper-evident record. Don’t trust us; check the working.

  7. Many Rooms, One Map · ~5 min
    Run one question in many rooms at once, each sealing its own record, drawn into a single map that never averages — the breadth of a national vote with the reasoning of an assembly. The series leaves you three tools — the test, the checklist, the fairness scorecard — and one invitation: check our working, and build a room instead of a wall.

Begin at part one →