Observation

Grammar is a prerequisite for literature, yet in UHT space they are strangers — Jaccard 0.200, Hamming distance 12, only 3 shared traits out of 15 combined. Cell and organism, by contrast, share 13 traits with perfect subsumption and a Jaccard of 0.929. The difference is not accidental. It exposes a clean boundary in what UHT encodes: the taxonomy captures mereological structure (what things are made of) but is blind to epistemological structure (what you need to know first).

The test asked whether UHT trait profiles could detect conceptual prerequisites — whether the simpler concept’s active traits appear as a subset of the more complex concept’s traits. Across 6 prerequisite pairs and 6 sibling pairs with 21 freshly classified entities, the answer is a structured no.

Evidence

Prerequisite pairs showed a mean trait subsumption ratio of 0.681, below the 0.70 confirmation threshold. But the aggregate masks a bifurcation. Part-whole prerequisites — where A is literally a constituent of B — averaged 0.875 subsumption: cell→organism at 1.000 (all 13 of cell’s traits present in organism, which adds one more), atom→molecule at 0.750 (3 of 4 shared, differing only on bit 16). Knowledge-dependency prerequisites averaged 0.584: arithmetic→algebra at 0.667 (Jaccard 0.444), formal logic→computer programming at 0.667 (Jaccard 0.353), force→classical mechanics at 0.667, and grammar→literature at 0.333 (Jaccard 0.200).

Sibling pairs — algebra and geometry, chemistry and physics, botany and zoology — showed a mean max subsumption of 0.979. Same-domain concepts share nearly all their traits, typically differing by 1-2 bits. Ethics’ 6 traits are a perfect subset of law’s 16. Zoology’s single active trait is contained in botany’s two. This near-total overlap makes subsumption useless as a prerequisite detector: siblings are more subsumed than many genuine prerequisites.

Interpretation

UHT’s 32-bit trait encoding captures ontological composition faithfully. When B physically contains A — organisms contain cells, molecules contain atoms — A’s traits genuinely are a subset of B’s, because B inherits the ontological properties of its parts. This is mereological encoding, and it works.

But knowledge prerequisites are a different relation entirely. Grammar and literature are both about language, yet grammar is structural-formal (rule-governed, compositional, symbolic, normative) while literature is cultural-aesthetic (institutionally defined, economically significant, social construct). Understanding grammar helps you read literature, but grammar’s ontological properties are not building blocks of literature’s ontological properties. The prerequisite relationship lives in the epistemological domain, not the ontological one.

This cleanly separates what UHT measures from what embeddings measure. Embeddings would rank grammar-literature as similar (they co-occur in educational contexts). UHT correctly identifies them as categorically different things. Neither system detects the prerequisite relationship — it requires a third kind of knowledge that neither ontological traits nor distributional statistics encode.

Action

HYP-048 is closed as refuted. RES-052 records the full analysis with the mereological/epistemological split. A research fact captures the finding that subsumption detects part-whole but not knowledge-dependency prerequisites. The mereological sensitivity is itself a positive finding: UHT could potentially be applied to detect compositional relationships (does concept B contain concept A as a component?) with the subsumption metric. A future hypothesis could formalize this narrower claim — testing whether subsumption ratio above 0.80 reliably identifies mereological containment across diverse domains.