V.5 — Dominions: Plurality Without Closure

Paper Summary

Analyzes a concrete proposal: federated virtual jurisdictions (“Dominions”) where agents create sovereign spaces, admit others by consent, and enforce rules solely through expulsion. Shows this is optimal as a governance layer under agency-preserving constraints.

The Dominion Architecture

Core Features

  1. Dominion Sovereignty: Any agent may create a virtual jurisdiction Dᵢ with locally defined rules
  2. Voluntary Entry: Others enter only by invitation + explicit consent to rules
  3. Bounded Enforcement: Rule violations result only in expulsion. No punishment, fines, or coercive penalties permitted
  4. Exit Supremacy: Unconditional ability to leave any Dominion
  5. No Global Value Aggregation: System doesn’t rank/optimize/reconcile Dominion-level value functions
  6. Thin Substrate: Shared infrastructure enforces identity persistence, consent verification, capability isolation, expulsion — but does NOT adjudicate values or outcomes

Critical Requirements

Asset Portability:

“Persistent identity, assets, and reputation are substrate-bound and user-owned, not Dominion-operator-owned.”

Dominions may define local affordances but durable assets must persist across expulsion. Without this, exit costs rise and sacrifice gradients re-emerge.

Capability Isolation: Dominions are capability-isolated execution contexts, not peer sovereigns:

  • No network-addressable access to other Dominions
  • Cannot allocate substrate resources beyond assigned quotas
  • Cannot write to shared state except through explicit substrate-mediated bridges

“Inter-Dominion aggression is physically impossible by design, not regulated post hoc.”

Analogous to process separation in operating systems or object-capability security.

Admissibility Constraints

Set A of admissible architectures satisfies:

  • Preservation of agentic decision authority
  • Allowance for endogenous value drift
  • Absence of outcome coercion
  • Enforcement limited to constitutive constraints

Question: Is Dominion system undominated within A?

Optimality Claims

1. Pareto Maximality Within A

Any alternative A’ ∈ A that:

  • Restricts Dominion creation
  • Limits exit
  • Enforces shared norms
  • Imposes mandatory coexistence

…strictly reduces at least one agent’s value-consistent trajectories while failing to increase any other agent’s attainable value.

“No admissible architecture strictly dominates the proposed system.”

2. Freedom Density Optimality

Freedom density = measure of distinct value-consistent future trajectories per unit of coercive constraint

Maximized because:

  • Constraints localized to voluntary contexts
  • Enforcement minimal and reversible
  • Arbitrarily divergent norms across Dominions permitted

3. Robustness Under Value Drift

For any agent i and t₁ > t₀, if Uᵢ(t₁) ≠ Uᵢ(t₀), there exists a path not requiring reforming or coercing other agents.

This property fails in nation-states, federations, consensus communities, public-goods-dependent systems.

“The system of Federated Virtual Dominions is therefore drift-optimal.”

4. Minimal Sacrifice

No standing sacrifice class:

  • Losses localized to exit costs
  • No agent’s flourishing depends on another’s coerced participation
  • No variance sink required for equilibrium

What It Doesn’t Optimize

Deliberately excludes:

  • Shared meaning
  • Large-scale coordination
  • Public goods
  • Epistemic convergence
  • Low transaction cost

“Agents who value such goods may instantiate them voluntarily within Dominions. The architecture refuses to enforce them globally.”

Scope Limitation

Optimality claims apply to governance layer of digitally mediated interaction.

Does NOT eliminate:

  • Physical scarcity
  • Energy requirements
  • Biological dependency
  • Material political economy

“Physical sacrifice patterns may persist elsewhere.”

Relation to Nozick

Can be read as digital instantiation of Nozick’s “framework for utopias” with:

  • Exit supremacy
  • Asset portability
  • Expulsion-only governance

Unlike Nozick: Justification rests on agency preservation + non-composability + drift robustness, not rights-based derivation.

Key Result

Under constraints of agency preservation, value drift, non-coercion, asset portability, and capability isolation:

A system of Federated Virtual Dominions is:

  • Pareto-maximal
  • Freedom-density optimal
  • Drift-robust
  • Sacrifice-minimal
  • Undominated within admissible design space

“This is the strongest form of optimality available in principle.”

The Core Insight

“The correct ambition is no longer to design a perfect world, but to design a system that refuses to decide what perfection must be.”

Connections

  • Concrete realization of V.3’s plurality-preserving meta-architecture
  • Satisfies V.4’s OAM requirements by construction
  • Shows expulsion-only enforcement can work when exit supremacy + asset portability hold
  • Relevance to AI alignment: substrate-level design that prevents domination structurally