Over the last couple of decades, our ability to accurately describe the evolution of galaxies using numerical techniques has improved tremendously, and we now have a fairly strong idea of the basic physical building blocks involved. One building block which remains complex and elusive, however, is the nature of galaxies’ physical relationship with the supermassive black holes (SMBHs) they host. From observations we know they’re present, we know they’re powerful, and we know they coevolve with the galaxies. It also seems to be the case that unless we account for feedback from those SMBHs in some way, our simulations do not produce the observed red sequence of “quenched” massive galaxies which have ceased forming stars, implying that the SMBHs may be responsible for quenching these systems. Thus, modern simulations of massive galaxies generally include some model for SMBH accretion and feedback. Because the fundamental physics is unresolvable, there is a tremendous amount of flexibility in choosing how to implement these models, and models can vary widely between collaborations. I will describe our recent efforts in the Feedback In Realistic Environments (FIRE) collaboration to get a handle on this problematic flexibility to try to understand: how do the choices we make in modeling SMBH accretion and feedback affect the star formation on galactic scales? To what elements of a SMBH model are the galaxies most sensitive, and what are common qualities shared by “successful” models that most accurately describe the properties of real galaxies? And in turn, what does that tell us about the nature of the physical relationship between galaxies and SMBHs?