Carl,
I proposed a scenario, if you choose to agrue with the scenario then nothing in the message has any meaning to you. The only way the recundant address data would not be denormalized is if the attributes in the customer table had a different domain definition than the ones inthe address table, which in my scenario is not the case. I don't understand how arguing as to whether or not denormalization was called for in the secnario has any bearing on what denormalization is.
I said, "Testing showed that ..." now that is a premise. It may or may not be valid in any given situation, but in my example it was valid because I said it was. That was the scenario with its constraints. The purpose was to demonstrate actual denormalization as opposed to a normalized structure that is very often called denormalization.