Numerical simulations of galaxy formation, based on the cold dark matter model, explain the observed large-scale structure and abundance of massive galaxies. However, current hydrodynamic simulations predict incorrect structure and sub-structure of individual galaxies. Many more dwarf galaxies around large galaxies are predicted than observed, leading to the "missing satellites problem". Much more cosmic gas inside galaxy halos is predicted than observed, leading to the "missing baryons problem". Much larger bulges inside disk galaxies are predicted than observed, leading to a new, yet unnamed problem. I will describe current attempts to resolve these problems by improving the modeling of star formation and its feedback on the galactic structure.