Reserved Jurisdiction

Layer 7 · Judgment and escalation · v0.1 DRAFT · 2026-07-30

Invariant

Escalation to human judgment must exist, and it must be constitutional and plural — written into the system's rules, held by more than one hand. The invariant is inherited from Chapter Eleven of Purposive Accelerationism, and it cuts both ways: a stack with no escalation path has bet everything on its own code, while a stack whose escalation runs to a single hand has installed its own capture vector. The backdoor form of human override is the vulnerability; the constitutional form is the safeguard.

Constitutive choices

• What is reserved. Which decisions may never be fully automated — the contested exit, the disputed identity, the rights conflict the rules did not foresee. Naming the reserved list is the founders' work; a list that grows silently is a jurisdiction being captured, and one that shrinks silently is a jurisdiction being deleted.
• Who judges. A standing panel, a drawn jury, a professional arbiter, or a court imported across the territorial seam — and how those judges are selected, compensated, recused and removed.
• What judgment can reach. Whether a ruling can move funds, rewrite an entry in a registry, or only unlock a next procedural step. The narrower the writ, the less the jurisdiction is worth capturing; the wider, the more it must itself be checked.
• The finality rule. Where appeal ends. Every system needs a final word, and choosing it in the open is what separates a constitution from a terms-of-service document.

Forces and capture risks

The pattern tilts toward the Collective vertex — judgment is a common good, and someone must hold it. Shipped alone, the risks are the classical ones with new plumbing. A single arbiter is the general will in a robe: whoever holds the override becomes the most attractive capture target in the stack, which is the empirical finding the book cites at this seam — the subordinated system is the most capturable by whoever occupies the authority position. Judges paid by the party they judge tilt to the Market. Automated pre-screening that quietly decides what ever reaches a human tilts the whole jurisdiction back into the code it was meant to check.

Pattern space

• Staked arbitration. Arbiters post value forfeited for rulings later overturned; disputes route to panels whose incentive is coherence with the written rules. Scales cheaply; justice tends toward what the median staker believes.
• Drawn juries. Random selection from the affected membership, personhood-gated, serving once. Hardest to capture in advance; costliest per case, and it needs the identity layer to hold.
• Tiered escalation. Code decides the routine; a cheap fast panel hears objections; a slow deliberate body hears what the panel certifies as constitutional. Most systems need all three speeds, and writing the certification rule is the real design work.
• Imported courts. The territorial world's judiciary, reached through the wrapper and the oracle seam. Sometimes unavoidable where imported assets are at stake; imports the territory's politics, fees and pace with it.

Reference implementations (2026 — dated, replaceable)

Staked arbitration as implemented by Kleros and its descendants; appeal-tiered dispute clauses in mainstream DAO frameworks; optimistic mechanisms whose challenge windows are, in effect, a standing invitation to escalate. Configurations, not recommendations.

Failure modes

The captured bench: one hand over the override, then over everything downstream of it. The priced-out plaintiff: escalation fees that make the jurisdiction real only for the large. The silent filter: automated triage that decides what a human ever sees, reserving the jurisdiction in name and deleting it in flow. The forum-shopped ruling: plural arbiters devolving into a market for favourable judgment, plurality without independence.

Composition

Requires the identity layer (juries and recusal need persons), the money layer (stakes and fees need settlement), and escrow from the parts inventory (a ruling needs something held to rule over). Fluid collectivism depends on this layer for contested exits; the polity layer depends on it for everything its written rules did not foresee; the meta-governor names it directly — a right that cannot be appealed to a plural human jurisdiction is a right in name only.