Observation

UHT’s holistic encoding — the distribution of active traits across all four quadrants rather than concentration in one — turns out to depend on how many traits an entity activates. Below nine active bits, concentration is common. Above nine, it vanishes entirely. The hypothesis that UHT encodes concepts holistically (HYP-042) is correct for trait-rich entities but fails for sparse ones, and the boundary is sharp.

The 8-bit stratum produced the most puzzling result: 39% of 8-bit entities exceed 0.60 quadrant concentration, rivaling the 42% rate at 4 bits where arithmetic makes concentration nearly inevitable. Investigation revealed this is not a systematic bias but a domain-cluster artifact. Six Italian pasta shapes share the identical hex code E6880008 (5 of 8 bits in the social quadrant), and three historical dates share 0004F2C0 (5 of 8 in functional). Remove these repeated classifications and the 8-bit anomaly dissolves.

Evidence

Across 296 entities with four or more active bits (sampled from 600 across the full corpus): mean max-quadrant-ratio 0.481, with 15.2% exceeding 0.60. The hypothesis required mean ≤ 0.50 (met) and fewer than 15% above 0.60 (marginally failed). Breakdown by bit count tells the real story:

• 4 bits (N=43): 42% exceed 0.60 — a 3-1-0-0 split gives 0.75, so concentration is the expected case.
• 8 bits (N=33): 39% exceed — driven by domain clusters (pasta ×6, dates ×3, all identical hex).
• 9 bits (N=18): 6% exceed — one entity.
• 10–23 bits (N=142): 0% exceed 0.60. Zero.

Quadrant means across all qualified entities: Functional 2.68, Abstract 2.52, Physical 2.28, Social 1.89. The social quadrant is consistently the least activated.

Interpretation

Holistic encoding is not a uniform property of UHT — it is a threshold property that emerges above a critical trait density. Entities with nine or more active bits distribute those bits across all four quadrants without exception. This makes biological sense: concepts rich enough to activate nine traits are necessarily complex enough to span physical, functional, abstract, and social dimensions. Simple concepts with four bits lack the trait budget to populate all quadrants.

The social quadrant’s systematic deficit (1.89 mean vs 2.68 functional) suggests either that social traits have stricter activation thresholds or that the corpus skews toward concepts with stronger functional and physical signatures. This warrants investigation.

The 8-bit cluster artifact reveals something else: domain-specific entities (pasta shapes, calendar dates) classify identically, inflating concentration statistics. Any future quadrant analysis should deduplicate by hex code.

Action

Recorded result RES-CALIBRATIONRESULTS-045 linked to HYP-042. Created observation OBS-STRUCTURALFINDINGS-041 about the 8-bit anomaly and social quadrant deficit. Stored research fact on the bit-count threshold finding.

HYP-036 (null-hex reclassification) could not be tested — the upstream OpenAI API key has expired, causing all force-refresh classifications to return 00000000 with 401 errors. An integrity alert has been raised; this must be resolved before any session requiring fresh classification can proceed. Next session should test HYP-036 once the key is restored, and should investigate the social quadrant deficit as a potential trait-set imbalance.