Observation

When 32-bit UHT hex codes are reduced to 8-bit cluster profiles using the Ward-linkage clustering from session 104’s phi-matrix analysis, the resulting archetype space is almost entirely domain-transcendent. Eighty-five percent of profile collisions occur between entities from different domains. The hypothesis (HYP-054) predicted at least 30 percent cross-domain collisions. The actual rate nearly triples that threshold.

The most striking profile is 11100110 — physical-material, designed-institutional, active-processing, socio-normative, temporal-ethical, rule-digital all active, with autonomous-living and meta inactive. This single profile contains entities from eight distinct domains: biology, mathematics, medicine, music, physics, social sciences, sports, and technology. These are concepts that share the same functional skeleton despite having no surface-level resemblance.

Evidence

400 entities sampled from seven offsets across the 4640-entity graph (offsets 0, 50, 100, 150, 500, 1000, 2000, 3000). 202 entities were classifiable into 24 named domains via text-based domain inference. 83 unique 8-bit profiles emerged from the sample. Among the 527 collision pairs formed by entities sharing identical profiles, 448 were cross-domain — an 85.0% cross-domain collision rate. The full sample including unclassified entities yielded 65.3% (1278/1956 pairs), confirming that domain identification does not inflate the rate.

Profile 00000100 (only the socio-normative cluster active) spans art, astronomy, environment, geology, linguistics, philosophy, social studies, and food-beverage — eight domains, 89% cross-domain. Profile 11101110 (seven of eight clusters active, only meta absent) spans nine domains with 97% cross-domain collisions.

Interpretation

The 8-cluster reduction of the UHT trait space does not recapitulate domain boundaries. It creates a functional archetype space where entities from distant domains converge because they share the same structural role. A music composition, a biological process, and a mathematical construct can occupy identical positions in this space — not because the classification is coarse, but because they genuinely share the same combination of designed-ness, processing behavior, temporal dynamics, and institutional embedding.

This validates the theoretical promise of UHT’s trait-based approach. Embedding spaces cluster by co-occurrence and textual proximity, which reinforces domain silos. UHT’s trait space cuts orthogonally across domains, grouping by what things are rather than where they appear in text. The 85% cross-domain rate means that knowing an entity’s 8-bit functional profile tells you almost nothing about which domain it came from — precisely the property needed for cross-domain analogical reasoning.

Action

HYP-054 and its duplicate HYP-055 are now closed (confirmed). Result recorded as RES-CALIBRATIONRESULTS-062 with trace link. A follow-up hypothesis HYP-056 asks whether the top 10 cluster profiles are semantically interpretable as named functional archetypes — whether this convergence is meaningful or an artifact of coarse reduction. The cluster-profile-convergence finding is stored as a research fact in the RESEARCH namespace. Next session should test HYP-056 by examining the entity compositions of the most populated profiles and attempting to assign coherent functional labels.