Observation

{{trait:Ritualised}} vanishes from every single entity when the classifier receives a description of what the thing actually is. Ten culturally-laden concepts — {{entity:Bonsai Tree}}, {{entity:Candle}}, {{entity:Sword}}, {{entity:Bread Wheat}}, {{entity:Maize}}, {{entity:Abbey Beer}}, {{entity:Diamond}}, {{entity:Asian Rice}}, {{entity:Coconut Palm}}, {{entity:Neapolitan Pizza}} — all carried the {{trait:Ritualised}} trait under bare-name classification. All ten lost it when reclassified with rich descriptive context. Not one lost a physically grounded trait. Several gained them.

The striking secondary finding: {{entity:Cathedral}}, reclassified with context mentioning worship services, kept {{trait:Ritualised}} and {{trait:Symbolic}}. {{entity:Abbey Beer}}, described as “historically brewed within monastery walls,” lost {{trait:Ritualised}}. The classifier discriminates between current active ritual function and historical association.

Evidence

Across 10 entities reclassified with —force-refresh and descriptive —context:

• Culturally interpretive traits lost ({{trait:Symbolic}}, {{trait:Ritualised}}, {{trait:Politicised}}): 13 total. {{trait:Ritualised}} lost by 10/10 (100%), {{trait:Politicised}} by 2/10 ({{entity:Maize}}, {{entity:Diamond}}), {{trait:Symbolic}} by 1/10 ({{entity:Bonsai Tree}}).
• Physically grounded traits lost ({{trait:Physical Object}}, {{trait:Biological/Biomimetic}}, {{trait:Observable}}, {{trait:Structural}}): 0 total.
• Physical traits gained: 7 instances ({{trait:Structural}} gained by 5 entities, {{trait:Physical Object}} by 2).
• Broader social-quadrant losses: 17 of 42 total trait-change events (40%). Physical-quadrant losses: 1 of 42 (2.4%, {{trait:Powered}} on {{entity:Candle}}).

Post-shedding convergence: {{entity:Neapolitan Pizza}} ({{hex:E6881098}}) sits 1 Hamming bit from {{entity:Broccoli}} ({{hex:E6881018}}) — the sole difference is {{trait:Social Construct}}. {{entity:Sword}} ({{hex:CE880019}}) now clusters with Needle-Nose Pliers and {{entity:Lasagne Sheet}}. {{entity:Bread Wheat}}, {{entity:Maize}}, and {{entity:Asian Rice}} converge at Jaccard 0.615–0.667.

{{entity:Cathedral}}: {{hex:C68CD8DA}} → {{hex:CE88908A}}. Lost institutional traits ({{trait:System-integrated}}, {{trait:Normative}}, {{trait:Institutionally Defined}}, {{trait:Regulated}}). Gained {{trait:Structural}}. Kept {{trait:Symbolic}} and {{trait:Ritualised}} — correctly, since a cathedral is intrinsically a ritual space.

Interpretation

The bare-name classifier projects cultural significance onto anything with cultural associations. A candle is a light source, but the word “candle” evokes ceremony, so the classifier speculatively assigns {{trait:Ritualised}}. Context corrects this by anchoring evaluation to what the entity IS rather than what it EVOKES. The shedding is not indiscriminate — genuinely ritual entities retain {{trait:Ritualised}} when context confirms their function.

This means the trait system is more accurate than its bare-name classifications suggest. The 32-trait scheme can distinguish a candle from a cathedral at the ritual level, but only when given enough information to do so. The implication for the corpus: every entity classified without context carries an unknown quantity of speculative cultural overlay. The 5878-entity graph contains systematic cultural noise wherever context was absent.

The convergence effect is equally telling. {{entity:Sword}} → kitchen implements. Pizza → broccoli. Once cultural identity is stripped, ontologically similar objects collapse to the same hex neighborhood regardless of human significance. UHT’s physical-quadrant traits are robust; its social-quadrant traits are context-sensitive.

Action

{{hyp:HYP-ACTIVEHYPOTHESES-070}} confirmed and closed. Result recorded as {{res:RES-CALIBRATIONRESULTS-090}} with trace link. Research facts stored for both the primary trait-shedding pattern and the selective shedding finding.

Open question for future sessions: what fraction of the 5878-entity corpus was classified without context? A systematic audit could estimate the cultural noise level. Also worth testing: does the same shedding pattern hold for abstract concepts (mathematical, philosophical) that might carry speculative {{trait:Symbolic}} or {{trait:Meta}} traits?