Comment to a post from Lubos in his blog. Copy paste for reference.
As here you could be stressing the criticism of a comment ten years ago, allow me just a reminder for the people who was not here at that time. We set V as a five, (u,d,s,c,b), of SU(5). Then we go for composite scalars: the 24 out of V x V* = 24 +1 has six of charge +1, twelve neutrals and six of charge -1. The 15 out of VxV=15+10 has six of charge +2/3, six of charge -1/3, and three of 4/3. So this is the scalar content of three generations of the Supersymetrical Standard Model, except for the chiral 4/3 beastie.
This “Nuclear superdemocracy” actually fixes the number of generations because you can not make the same with any, only with N=3 and one “top” quark out of the set. Not a thing that Chew could have attempted in the sixties, with only u,d,s and no argument to look for squarks nor sleptons.
A peculiar coincidence is that the “fake sleptons”, this is, the actual mesons bound with SU(3) colour, have approximately the same mass that the actual fermions, which is amusing when one considers that the origin of the mass is completely different: yukawas + higgs on one side, pure QCD on the other.