A cosmotopological relation for a unified field theory

I present an argument, based on the topology of the universe, why there are three generations of fermions. The argument implies a preferred unified gauge group of SU(5), but with SO(10) representations of the fermions. The breaking pattern $SU(5) \to SU(3) \times SU(2) \times U(1)$ is preferred over the pattern $SU(5) \to SU(4) \times U(1)$. On the basis of the argument one expects an asymmetry in the microwave data, which might have been detected already.