The theme of the talk is how different modes of stellar feedback affect the evolution of galaxies and how, with appropriate care, we can constrain the role of the historically popular supernova feedback on the evolution of galaxies and their stellar content. We argue that prior work has modeled supernovae poorly by ignoring stellar clusters and also the key physics of conduction that governs hot gas evolution. Clustered supernovae create superbubbles, kpc scale feedback events that can drive strong galactic winds. We show that superbubbles can be modeled via first principles simulations without resorting to common numerical tricks. Galaxies with superbubble feedback strongly regulate their star formation and global baryon budget. These simulated galaxies match disk galaxy observations very well. However, there are limits -- supernovae cannot explain the regulation of star formation in all galaxies.