Since when?
The assumption here is that you can tell a comment is going to be stupid by testing whether or not the person understands homonyms. The only thing you check with that is language ability. I'd argue you'd be better off searching for common internet memes and cliches instead of original thinking -- since that's a sign of a lazy or uninteresting mind. The author is kind enough to inadvertently supply us with one "...If Fox Nation implemented something like this, they’d have zero commenters..." But you could test stuff like this based on any preconceived viewpoint, such as Obama-socialist, Obamacare, etc. The use of shortcuts and blindly-repeated jokes and phrases is a good sign that you're not going to be getting much from the comment.
Not sure how you'd code it, though, but I'm certain you could come up with some semantic magic given enough input text. I'd imagine you'd use n-grams and some Bayesian logic. You'd have to have a pre-existing corpus of the person's writing, though.