The point the article is making is not that the messages are "Good" or "Bad", it's that an automatic rating system that intends to allow some usages of profanity rates messages with the type of profanity that it purports to allow as very bad, towards the end of the scale that it uses.