The rhombic dodecahedron may be a new shape to some, but it has existed on earth for millions of years, long before any humanoids were around to ignore it. Even among those humans who trouble themselves over pesky matters in such fields as science and mathematics, the rhombidodec has a long tradition.
Kepler, of planetary motion fame, had noted the significance of the shape centuries ago. Stephen Hales, an English clergyman, once took some time off from matters of damnation to see what would happen if peas were compressed together, and found they tended to form rhombic dodecahedra. The French zoologist G.L.L. Buffon thought this was news worth mentioning and told his colleagues of these developments (while conveniently omitting that the results came from across the channel). On matters of growth and form, D'Arcy Thompson classically had much to say, including that clay pellets will compress into shapes much like rhombic dodecahedra, while J.W. Marvin compressed lead balls and found the same shape yet again. Thus, to many experimenters of old, the rhombidodec had established its celebrity rather firmly. It may not have been the only applicant for ideal close packing, but there is no question that it has been one of the most prominent, if not the most prominent, close packing shapes.
Even in modern times, rhombic dodecahedra are seen yet again in bubble experiments, being the most common of all foam shapes as reported by Philip Ball in "The Self Made Tapestry."
Closer to home, in Synergetics, page 828, supplement F by Arthur Loeb on close packing, Loeb echoes the idea we shall see momentarily in point 7, namely, that if bees were to use their economic building methods in several dimensions beyond their hexagonal honeycombs, they would construct rhombic dodecahedra.
Thus one might very reasonably ask: is it really so unlikely that rhombidodecs might best describe the shape of nucleons, too?