Why it is a question
The best fit in THEORY-163’s source uses a need distribution built from the naming data themselves, averaged and cross-validated over languages. With a uniform source the systems are still near the curve, but the fit is worse. The only structured alternatives tested are hue rotations of each system. Until 2026-10-20 this was THEORY-163’s promote_when.
What would answer it
A test in which the need distribution over meanings is measured independently of the naming data (for example from usage frequencies) and the languages scored are held out from every fitted quantity, so that attested systems are compared with the bound and with structured alternatives other than hue rotations; or the same result in a second semantic domain with its own independently grounded meaning space. A better fit with a source estimated from the same naming data cannot settle it.