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).