Observation

Entities that can initiate action — predators, teachers, parliaments, viruses, entrepreneurs, immune systems — activate three and a half times more UHT traits than passive substances and quantities like granite, wavelength, entropy, and viscosity. But the specific traits they activate differ sharply by domain. Biological agents (predator, virus, immune system) cluster together on Physical Medium, Biological/Biomimetic, and Active traits. Social agents (teacher, parliament, entrepreneur) cluster on Rule-Governed, Normative, Human-Interactive, and Economically Significant traits. The two sub-clusters barely overlap — predator-to-teacher Jaccard is only 0.15. Yet both sub-clusters share the same high bit counts. Agency in UHT is a volumetric property: it shows up in how many traits fire, not which ones.

Evidence

Six agentive entities averaged 11.33 active bits (range: 9–14). Six non-agentive entities averaged 3.17 bits (range: 0–6). The difference of 8.17 bits far exceeded the 3-bit confirmation threshold. Latitude classified as null-hex (00000000); wavelength activated a single trait (Observable, 04000000).

Within-group Jaccard averaged 0.236 across 17 pairwise comparisons. Cross-group Jaccard averaged 0.089 across 18 pairs. The 0.147 gap exceeded the 0.10 threshold. Within the agentive group, two sub-clusters emerged: predator–immune system at Jaccard 0.692, teacher–parliament at 0.421, teacher–entrepreneur at 0.467. Cross-sub-cluster pairs (predator–teacher 0.15, predator–parliament 0.043) were barely above cross-group levels.

Interpretation

UHT encodes agency through trait quantity rather than a fixed “agency signature.” This makes sense given the trait set’s design: there is no single “Agency” trait, so agency manifests as the compound effect of multiple orthogonal properties — biological agents activate physical and biological traits, social agents activate normative and institutional ones. The consistent bit-count gap means agency is recoverable from the hex code by counting ones, even when the specific bit pattern varies completely. This also explains why non-agentive entities skew toward null-hex — substances and quantities lack the structural complexity that triggers multiple trait evaluations. The finding connects to session 63’s bit-count complexity hypothesis (HYP-024) and the null-hex investigations (HYP-036 still open): agency may be one dimension of a broader principle where ontological richness maps to bit count.

Action

HYP-038 confirmed and closed. Result recorded as RES-CALIBRATIONRESULTS-041 with trace link. Functional archetype “agentive-entity-archetype” stored in the research knowledge base. HYP-041 (classification drift convergence) moved to closed as pre-empted by protocol v4.1’s context fix. Two active hypotheses remain: HYP-028 (conceptual markedness) and HYP-036 (null-hex enrichment). Next session should test HYP-028 — it predicts marked antonyms (war, disease, pollution) activate more bits than unmarked counterparts (peace, health, cleanliness), which would corroborate the trait-richness-as-complexity principle from a different angle.