Research Programs
Radiant programs begin with a domain problem that can be tested inside the field's own methods.
Artificial intelligence and policy-governed systems
A theory and research program for routing a proposed action to policy regimes that warrant authoritative review. It is not an adjudicator and reports no empirical efficacy result.
Information retrieval and natural language processing
Fixed-budget evidence retrieval: dual-bounded relational context allocation recovered complete HotpotQA supporting-evidence sets more often than matched flat top-k retrieval. The reported outcome is not answer accuracy.
Control systems and systems engineering
Goal governance for systems whose defended reference can become inappropriate as environments change, with a preregistered negative result for one mature-evidence mechanism.
Evolutionary biology and genomics
Endosymbiotic origins, organellogenesis, and biological transitions modeled as threshold events with measurable precursors and residue.
Finance, economic history, and organizational measurement
Instrument proliferation, stakeholder narrative divergence, belief-pool movement, and coherence fracture as testable signals before ordinary measures register distress.
Humanities and social systems
Historical linguistics, narratology, textual transmission, and creative structure treated as domain-native research problems rather than decorative examples.