Can we truly predict crystal structures?

This post was stimulated by a nice talk that Richard Catlow gave at MM2009 last night.
He began with the fact that twenty-one years ago, the Editor of Nature, John Maddox made the provocative claim,

"One of the continuing scandals in the physical sciences is that it remains impossible to predict the structure of even the simplest crystalline solids from a knowledge of their composition."

Woodley and Catlow recently wrote a nice review in Nature Materials, entitled, Crystal structure from first principles, which highlights recent progress and future challenges. I acknowledge the impressive progress, particularly with the developing of ingenious new algorithms for searching configuration space that has been made, but most of this seems to be using semi-empirical force fields and/or empirically based cost functions based on bond valence sums.

I think I would rewrite Maddox claim as,

One of the greatest scientific and philosophical challenges in the physical sciences is that it remains impossible to consistently predict from Schrodinger's equation the structure of crystalline solids solely from a knowledge of their composition and without knowledge of the structure of related materials.

I include "philosophical" challenge because this strikes at the heart of issues relating to emergence and reductionism.