Freak observers and the measure of the multiverse

I suggest that the factor $p_j$ in the pocket-based measure of the multiverse, $P_j=p_j f_j$, should be interpreted as accounting for equilibrium de Sitter vacuum fluctuations, while the selection factor $f_j$ accounts for the number of observers that were formed due to non-equilibrium processes resulting from such fluctuations. I show that this formulation does not suffer from the problem of freak observers (also known as Boltzmann brains).