What have we learned from large spectroscopic surveys?

An essential use of stellar population synthesis methods is to interpret the observations of galaxies, to infer their star-formation and assembly histories. 1 describe here some of the uses of these techniques over the past few years, primarily in the context of local galaxy redshift surveys. For tins purpose, such surveys have both advantages and disadvantages relative to higher signal-to-noise but smaller studies, which 1 discuss. After discussing general issues, I describe how the substantial model uncertainties impact the analysis of galaxy spectra.. Then, I discuss the special case of elliptical galaxies, arguing that investigators have made interesting discoveries even in the absence of perfect (or even good) models. I then describe the desiderata for massive higher redshift surveys and the trade-offs that must be made between cosmological and galaxy-focused science, in qualitative terms.