Observation

Population has fewer active traits than a cell. A transistor outscores a logic gate. A city is simpler than a community. The elegant complexity corridor — that rho=0.925 Spearman correlation between bit count and structural complexity — does not survive contact with actual compositional hierarchies. When tested as a monotonic scale principle within three domains (biology, computing, social organization), it breaks in every one.

The hypothesis predicted that moving up a part-to-whole hierarchy would produce monotonically increasing bit counts, and that entities at the same hierarchical level across domains (molecule ≈ transistor ≈ individual as atomic units) would cluster tighter than entities at different levels within the same domain. Both predictions fail cleanly.

Evidence

Biology: molecule (5 bits) → cell (6) → tissue (7) → organ (10) → organism (11) → population (6) → ecosystem (11). Monotonicity breaks at population, which loses Physical Object, Observable, and Physical Medium — the very traits that built the ascending sequence.

Computing: transistor (12) → logic gate (10) → integrated circuit (14) → processor (14) → computer (13) → network (19) → internet (13). Fails immediately: a transistor activates more traits than its logical abstraction. Internet drops below network as social/institutional traits replace physical ones.

Social: individual (4) → family (10) → community (11) → city (6) → nation (10) → civilization (13). City collapses — classified as a physical structure, losing the social-interactive traits that define family and community.

Same-level cross-domain mean Jaccard: 0.266. Within-domain different-level mean Jaccard: 0.391. The strongest pair was transistor–processor at 0.733 (same domain, two hierarchy levels apart). The weakest was processor–community at 0.136 (same level, different domains). Domain membership is 47% stronger as a similarity predictor than hierarchical scale.

Interpretation

The rho=0.925 complexity correlation was real but misleading. It emerged from comparing entities across categories — atoms vs. ecosystems vs. corporations — where category and complexity are confounded. Within a single compositional hierarchy, the relationship shatters at every physical-to-abstract transition. Population is not a more complex molecule; it is a different kind of thing entirely, and UHT’s trait vocabulary resets at the boundary.

This confirms and sharpens the domain-over-archetype finding from session 61: UHT encodes what category of thing something is, not where it sits in a compositional stack. The bit-count continuum is a shadow of category membership, not an independent scale signal. A transistor scores 12 bits not because it is hierarchically complex but because it is an engineered physical object — a category rich in trait activations.

Action

HYP-ACTIVEHYPOTHESES-032 moved to closed (refuted). Result recorded as RES-CALIBRATIONRESULTS-034 with trace link. Research fact stored under “domain-over-scale”. Twenty new entities classified across three hierarchies, expanding the entity graph with systematically chosen compositional-hierarchy concepts. Next sessions should investigate the physical-to-abstract transition boundary directly — what specific trait combinations predict where UHT’s complexity encoding breaks.