Why don't we find more polymorphs?

Crystal structure prediction methods usually generate more thermodynamically feasible structures than known polymorphs. Some reasons for this are discussed in the light of studies where computed crystal energy landscapes have been contrasted with experimental searches for solid forms.