Is anyone familiar with any sort of theoretical explanation of the number of chiral families? Any model or idea in this direction would be very interesting and useful.
The most official answer I have always heard is that three is a minimum for CP violating phase. But not a strong argument.
Extra dimensions seem to like three generations, as it is known from compactification of superstrings, usually getting the number of families from some topological parameter. With extra dimensions, but without superstrings, some of my university teachers did a paper years ago: https://inspirehep.net/record/11555?ln=es (Phys.Rev. D26 (1982) 691-697)
My own guess is a very retorted mechanism: use the quarks as preons for the yet-to-be-found susy scalars. This is only possible with three generations.
Article Origin of inertia at rest and the number of generations