Comment by mazesc

IBMs response seems strong regarding "linear scaling", but could they give a complexity-theoretic argument rather than an empirical one?