If all you are worried about is in-principle verification, then, so the argument goes, the way to check is to see if there is any empirical content to "statements which are not testable are meaningless." What is there in the world that we would see if that statement is true?