"Entropically more favorable to construct" was my first thought. If you have the option of a structure to do X with 8 copies of one molecule in a radially symme

"Entropically more favorable to construct" was my first thought. If you have the option of a structure to do X with 8 copies of one molecule in a radially symmetric arrangement, and a structure to do X with 5 different molecules that have only a single possible arrangement, having a random assortment of those molecules in solution (or cytoplasm), you're a lot more likely to get working copies of the symmetric structure, and non-working copies of the particular structure. Think viral nucleocapsid self-assembly in cells. The decreased coding required is a nice bonus, which probably contributes.