During the last decade, photoionization models for low-metallicity BCD galaxies have been systematically unsuccessful, in that the model temperature was smaller than the observed one, with important implications for the global energetics of the Early Universe. I shall analyse the origin of this problem and propose a solution.