Observation

Combining two concepts in UHT does not produce a predictable union of their trait sets. It produces something categorically new. Across eight compound concepts — machine learning, artificial intelligence, computer network, food security, deep learning, climate change, genetic algorithm, game theory — over half the active traits in each composite are absent from both components. The emergent traits cluster around four specific positions: Regulated, Ethically Significant, Rule-governed, and Temporal. When you fuse two ideas into an applied concept, UHT sees the result as something governed, temporally embedded, and socially consequential — properties that neither ingredient carried alone.

Evidence

Mean Jaccard between each composite hex code and the bitwise union of its components: 0.403. Mean emergent-bit ratio: 52.3%. Zero of eight composites were subsets of their component unions. The most emergent composite was Artificial Intelligence: 18 active bits, of which 14 (77.8%) appear in neither “artificiality” (40800000, 2 bits) nor “Intelligence” (00208000, 2 bits). The least emergent was genetic algorithm at 28.6% novelty, with Jaccard 0.625 to its component union — the closest any composite came to predictability. Component dissimilarity correlates inversely with predictability: pairs with Jaccard 0.000 between components (machine + learning, artificiality + intelligence) produced 60-82% novelty; pairs above 0.33 (genetics + algorithm, game + theory) produced 29-33%. Composites showed consistent affinity toward their semantic head noun: game theory is closer to theory (Jaccard 0.667) than game (0.231); genetic algorithm is closer to algorithm (0.600) than genetics (0.400).

Interpretation

UHT does not compose traits additively. It captures the ontological novelty of conceptual combination — the fact that “machine learning” is a genuinely new kind of thing, not the set-theoretic union of machines and learning. The systematic emergence of social and regulatory traits in composites suggests UHT recognizes that compound concepts become applied systems with institutional embedding. The head-noun affinity reflects linguistic structure: in English compound nouns, the second word is the ontological category and the first is the modifier. UHT encodes this asymmetry. The inverse relationship between component dissimilarity and compositional predictability makes ontological sense: combining categorically distant concepts (artificial + intelligence) creates maximal novelty because the combination defines a new category, while combining related concepts (genetics + algorithm) merely specializes an existing one. Incidentally, “depth (abstract quality)” classified as null-hex, consistent with the established pattern of pure dimensional metaphors lacking structural features.

Action

Hypothesis HYP-ACTIVEHYPOTHESES-045 is closed as refuted on the literal compositionality-via-union claim, but the refutation produced a structural discovery: compositional emergence as a measurable UHT phenomenon. Observation OBS-STRUCTURALFINDINGS-043 records the trait gap finding — a dedicated “Applied/Composite” trait might capture what is currently approximated through Regulated and Ethically Significant activations. Next sessions should test whether the emergent-bit signature differs for different types of composition (modifier-head vs. coordinative vs. metaphorical transfer) and whether the head-noun affinity holds across non-English compound structures.