The past 20 years have revealed that supermassive black holes play an essential role in the formation and growth of galaxies. In the local universe, every massive galaxy hosts a supermassive black hole, with black hole mass tightly coupled to the mass of the galaxy bulge. But a complete census of supermassive black holes over cosmic time has remained elusive, particularly in the “chicken-or-egg” formation of the first galaxies and black hole seeds. I will show how probabilistic, forward-modeling techniques reveal a hidden population of supermassive black holes missed by previous surveys. When applied to SDSS and Hubble CANDELS/3D-HST observations, these methods enable the first census of black holes in low-mass and star-forming galaxies, for a fossil record that discriminates between models of early-universe black hole seed formation. Beyond simply identifying supermassive black hole activity, I will also discuss the importance of time-domain surveys in mapping black hole mass and accretion, showing results from the pioneering new SDSS-RM project. I will conclude by looking forward to the next generation of observatories: JWST, WFIRST, and Euclid for a new spatially resolved frontier of supermassive black hole selection, and LSST and PFS for a new time-domain frontier of black hole mass and accretion.