In the study of galaxy evolution, we do not care about the physical parameters of any one galaxy: we care instead about the distribution of galaxies in physical parameter space as a function of cosmic time. The framework of hierarchical Bayesian statistics allows us to connect the measurement of this population distribution directly to pixel-level observational data, and to extract all available information without bias even in the limit of large numbers of objects observed at low signal-to-noise. I will describe this approach as we are currently applying it to the spectroscopic data set of the Baryon Oscillation Spectroscopic Survey (BOSS) of the third Sloan Digital Sky Survey (SDSS-III), and present our latest results on the joint distribution of galaxies in stellar mass and velocity dispersion at redshift z ~ 0.55. I will conclude with some provocative yet quantitatively justified claims about optimal experimental design for studying galaxy evolution and cosmology together.