I really disagree with the comparison drawn here. The question may be ill-posed in one way or another, but it's a category error to equate it with "could a programming language be based on cats". This question brushes up against, not quite coming to the point of asking, multiple legitimate on-topic non-trivial questions, and if it's closed it should be because it doesn't pick just one of them.
There is a long history of use of typographical variants, including typeface, for semantic meaning in programming languages. In fact, there's a longer history of that than virtually all languages that don't do so. The question touches on this point.
There are a number of ways of dealing with same-named or un-named variables, including De Bruijn indices, which this proposal is exactly analogous to. There is substance on this point too, even perhaps an onramp to a learning opportunity made accessible by the comic.
There is an element of how such a program would be edited, and structured editors and block-based or visual languages address these elements too. In fact, it's not even especially uncommon for such languages to distinguish variables or terms through some visual means, rather than just by name.
"Could it work" is a question of language ergonomics and usability, which provide both qualitative and quantitative evidence. It admits answers of the form "only up to N variables, because people can't readily distinguish more than that many fonts [cite]", of user studies with those visual languages — which do have a real question of how many types they can distinguish that way — described above, and of raising issues of accessibility.
It's overly literal to take this phrasing as a trivial yes-no question, and good answers would address when and why it did, or didn't, work, and those answers would have real applicability to real design of programming systems.
All of these points have substance, none of them are frivolous, and all of them are raised within and behind this question. "Could a programming language be based on cats" raises nothing alike that, and such a question would be pointless, trivial, frivolous, and whatever else you like.
Now, is a question based entirely on a joke in a comic likely to be a good one? Probably not in general. Does this particular question make it all the way to any of these concrete points? No, not quite, so it likely is unfocused in that sense, but I don't like to dismiss it as entirely frivolous. I think there could well be solid questions that draw from this comic as an entrance point to serious PL topics, and they wouldn't even be extremely different from this one.
I don't think that questions should be excluded purely because they derive from jokes somehow, or from some pop-culture source. The aetiology of the question isn't relevant to whether it's raising a point of substance or not. There are genuinely good, on-topic, well-formed questions someone could ask about INTERCAL and its derivatives, for example, and we can't exclude those or the AspectJ people will object.
In its current form, closing this question as unfocused is reasonable, but it's not because it's frivolous, it's because it doesn't form a single coherent well-scoped question with a specific concept of "could work". "Motivated by xkcd.com/2309, I'm wondering [ how typography has been / how variable names can be / how users could / how far it's viable to ] ..." could all make for very reasonable questions, given some appropriate details and context.
