Maybe instead of trying to use one base representation XOR the other, we could instead employ each representation based on the convenience it lends to the subject at hand?
So that every subfield invents its own base representation to go along with its own unit system?

I cannot think of any advantages that would outweigh the confusion that different base representations would create.