Quote by mitchell porter
There is also a secondary "rule" that you can add two ordered pairs which both have nonzero weak isospin, only if one has isospin +1/2 and the other has isospin 1/2.

It seems reasonable, as then we can look for some symmetrization argument to justify the idea. But is is also peculiar. It means that the uu and dd combinations only happen for R type quarks.
Looking at the reference of Huerta, I note that in
http://math.ucr.edu/~huerta/guts/node10.html the previous section he takes some pains to discuss the adjoint representation of U(1) and its role in the hypercharge. A subltle point here is that U(1)hypercharge is still chiral (as Distler likes to stress) and then it needs complex representations, while U(1) electromagnetism is not.