Absolutely - numerically is they way to go in some fields, but without a deep understanding of the underlying limitations there will be no expansion understanding/knowledge. It's the old academic vs. practitioner argument.