Programme
Research
The active Andaluz programme is Kernel ANN: whether consequential effect authority can remain structurally outside the writable state of a mutable optimiser, and what bounded experiments can establish about that.
Earlier theoretical and formal work establishes the research lineage. Experimental claims for Kernel ANN depend only on the evidence produced by the Kernel ANN experimental programme.
Current programme
Active research
- Kernel ANN architectureARCHITECTURE
- Protected semantic authorityACTIVE RESEARCH
- Authenticated execution boundariesDESIGN PHASE
- Optimiser redirection under constitutional unavailabilityBOUNDED EVIDENCE
- Semantic provenance and long-horizon reasoning failuresACTIVE RESEARCH
Provenance
Research lineage
This is intellectual and technical provenance — where the architectural commitments came from. It is not a proof chain and not a validation chain. Only the high-level sequence below is the direct lineage; other migrated work is related formal, architectural or authority research rather than a link in this chain.
- TRO
The Redemption Optimization
Event-valued treatment of harm, freedom and repair. Foundational work in the lineage; not experimental evidence.
- MKO
Moral Kernel Optimization
Non-bypassable constraint structure: preservation, anti-domination, repair.
- Closure
Closure and kernel work
Margan Closure Theorem, kernel transfer and Coherence Wall results; Ground as an operational closure audit layer.
- Kernel ANN
Protected effect authority
Architectural separation of a mutable optimiser from the authority governing consequential effects.
- M1…
Bounded experimental programme
Small, adversarially tested experiments. Current evidence is bounded to what those runs establish.
The number of prior research records does not strengthen any Kernel ANN claim. Claims advance only when experimental evidence advances.