Every layered system eventually faces the regress: the rules constrain the powerful, but someone maintains the rules — so the powerful capture the rule-maintainers, and the constraint becomes a costume. Quis custodiet ipsos custodes — who will guard the guards themselves — has survived two thousand years because most designs answer it with another layer of custodians, which merely relocates the question.
This design's answer: the custodian-selection rules belong to everyone, and to no one else. How delegation is earned, what testing or qualification applies to the upper tiers, which inputs carry weight and how much — age, achievement, property, whatever a society chooses — these are decided directly by the broadest tier, and are structurally outside the authority of the tiers they govern. Delegates vote under weight-rules they cannot set; leaders draft under entry-rules they cannot touch. The whitepaper's cohort architecture states the principle exactly: specialized cohorts operate within rules they did not write and cannot rewrite, and the broadest cohort's authority over the meta-rule is constitutive, not consultative. Power over policy can be delegated upward; power over the shape of delegation itself can never be.
The failure mode when custody is misplaced isn't hypothetical — it's on display in the United States right now. Congressional term limits have polled around eighty percent support for decades and have no path to enactment, because every path runs through Congress itself: the amendment process needs congressional supermajorities, and the Supreme Court's Thornton decision, which in 1995 barred states from term-limiting their own members of Congress, closed the state-level route. The smaller cohort retains structural control over the rules of its own qualification, so an overwhelming public preference about who may hold power simply cannot execute.
Switzerland shows the repaired version: one hundred thousand signatures force a binding national vote, and the custodian-rules stay permanently reachable by the governed. This chapter is the difference between those two countries, generalized.
A layer this powerful needs one more property: slowness. The meta-rules carry the system's longest maturation windows — the settling period before a rule faces its first verdict — its highest passage thresholds, and its slowest review cadence — deliberately the most viscous layer in the stack. The foundation can move; that is the whole point of chapter 2 — but it can never move fast. A sovereign that can amend its constitution in one angry quarter does not have a constitution; it has a mood. Viscosity is also the anti-capture property: a layer that takes years to change is a layer that cannot be seized in a news cycle, and every attempt to change it happens in full public view at deliberate speed, under the same engagement-and-approval floors as everything else — just set higher.
Note the composition with chapter 8's asymmetry. The negative powers — refusing reconfirmation, tripping a removal gate — stay cheap and dumb even at this layer. It is the positive restructuring of the rules that moves slowly. Stopping the machine is easy; rebuilding its governor takes years of sustained consensus. That asymmetry — easy to stop, slow to seize — is the signature of every mechanism in this proposal, applied here to the mechanism that governs the mechanisms.