Historical context
Convergence Science
Convergence Science is a broad prior research framework by Sergiu Margan, spanning domains well outside AI systems. Andaluz represents it here only as historical context for the emergence of Kernel ANN. It is not the Andaluz programme.
The current Andaluz engineering focus is Kernel ANN: protected effect authority for mutable optimisers, tested through bounded experiments.
Relevance
The strands that carried into Kernel ANN
Only a narrow part of the framework bears on the current programme. These are the strands the architectural commitments were reached through — provenance, not evidence.
Non-bypassable constraint
What structural conditions make a boundary survive optimisation pressure rather than remain advisory.
Evaluator preservation
Whether the machinery that judges an action can be preserved against the system it judges.
Closure and repair
When a harm path is genuinely closed, and what counts as repair rather than apparent closure.
Authority placement
Where decision authority sits, and what depends structurally on that placement.