Ok, but rather than "thinking outside the box" substitute "thought outside a minimization of a weighted graph". Scientists do not give enough thought to the differences mentally between what nonEuclidean shape the box remands in the difference of optimizations from a given graph weighted subjectively or more or less objectively matched to givens (minimal spanning trees in biogeography, macroeconomic models, causal patterns any way sugessted etc).
It is a hard thought. That is all. Kelvin was doing the best anyone could at his time.