No one is arguing reuse is a surprise - there used to be a joke in the early days of protein fold prediction - that if the protein amino acid sequence was a certain length - you just predict TIM barrel and you'd be right.
the question is a separate one - how much of protein universe of folds have already been seen?
