We have played with SDSS spectra for almost two decades and enjoyed performing all sorts of statistical analyses with them. In this talk we will take a critical look at these explorations and discuss what has worked well and what has not, what mistakes were made and some of the fundamental limits arising from the calibration of the data — all of which are informative to prepare ourselves for the DESI era. Finally we will also discuss new data analysis ideas and emerging techniques that are likely to impact the way we think and manipulate spectra in the near future.