RESEARCH

The unresolved questions stay visible.

Qualified results are the baseline. Untested generalisations remain research rather than being promoted into product claims.

PARTIALLY QUALIFIED / ACTIVE RESEARCH

Sovereign Control

Question
How narrowly can authority be bound to one operation, purpose and domain without accidental widening?
Why it matters
Narrow authority limits blast radius and prevents one approval becoming general power.
Already qualified
Independent authority boundaries and the common ALLOW / DENY / ESCALATE model are established.
Still open
Overlapping grants, broader operation-scoped authority and cross-domain generalisation.
Related evidence / papers
RMV1 branch evidence; common control contract.
QUALIFIED BASELINE / ACTIVE GENERALISATION

Federation & Projection

Question
How far can independent domains cooperate while remaining independently authoritative?
Why it matters
Shared state is useful; merged authority is a different and much riskier architecture.
Already qualified
Bounded two-domain projection and canonical joined-fabric composition are qualified.
Still open
More domains, longer chains, degraded links and stronger physical/network separation.
Related evidence / papers
Bounded projection contract; Shared Sovereign Fabric.
ACTIVE RESEARCH

Selective State & Scaling

Question
What is the minimum local state needed to decide safely, and when should missing state be hydrated?
Why it matters
A large represented estate cannot require every domain to materialise every fact for every decision.
Already qualified
Bounded selective state/hydration exists; 409,600 / 15-kernel is a closed tested stress point.
Still open
Hydration policy, freshness rules, minimum sufficient state and scaling laws beyond tested rungs.
Related evidence / papers
RMV1; data-centre SIL evidence.
QUALIFIED BASELINE / ACTIVE RESEARCH

STOP & Containment

Question
How far should STOP propagate, how quickly, and where can failure remain local?
Why it matters
A useful interrupt must limit further consequence without unnecessarily collapsing unrelated domains.
Already qualified
Common bounded software STOP semantics are qualified.
Still open
Federated propagation, latency, failure locality and domain-safe physical behaviour.
Related evidence / papers
STOP constitution and bounded STOP evidence.
ACTIVE RESEARCH

Recovery & Reconciliation

Question
How should partial cross-domain effects be resolved without inventing success or causing duplicate consequences?
Why it matters
Distributed systems routinely lose certainty before they lose state.
Already qualified
Bounded receipt and reconciliation contracts exist.
Still open
Partial cross-domain effects, safe resume/replay and uncertain-effect finality.
Related evidence / papers
Receipt contract; institutional handoff, distributed-compute and bounded agent-execution patterns.