There are a number of inadequately explained phenomena in addition to downsizing. We do not understand the high frequency of giant disk galaxies that are bulgeless (Kormendy et al. 2010). Many of these are massive disks. The problem here is that early merging, intrinsic to hierarchical formation, results in torquing and angular momentum loss. Not only are bulges inevitable, but essentially all models, with any combination of cold stream/minor merger/major merger scenarios for gas delivery to drive both star formation and SMBH feeding, produce overly massive bulges compared to the disk masses. The fundamental plane of spheroids displays remarkably little scatter (Zaritsky, Zabludoff and Gonzalez 2010). This is a challenge for SAMs in which merging plays an important role. As for SMBH, we lack any theory of their formation, or perhaps more accurately, we have too many theories. Star formation remains the outstanding problem. We have no fundamental theory of how stars form. The early dream of Eddington, "imagine a physicist calculating on a cloud-bound planet and ending with the dramatic conclusion, `What happens is the stars' " has not so far been fulfilled in any quantitative sense, such as deriving the IMF, even though it remains qualitatively correct. We are tempted to extrapolate star formation laws, as measured locally, to the extreme conditions found in major mergers. There are already indications that this extrapolation may be premature. But we have no robust alternative. The one ray of hope is that physics was simpler when the first stars formed, when there were no heavy elements, no dust, and no magnetic fields. This makes computations almost tractable. The crucial ingredient in star formation is the baryons. But we cannot fully account for the discrepancy between the initial baryon abundance and that observed in galaxies. Ejection certainly occurs, as evidenced by the enrichment of the IGM, but the details remain elusive.
Perhaps the greatest uncertainty lies in the nature of star formation. Is the star formation mechanism the same in all environments and at all epochs? Is the IMF universal? Improved resolution in theory and observation is needed. The great projects of the future, including the ELTs, JWST and LSST, will surely play key roles in this endeavour.