It's a “bug” (and yes, I'm using that deliberately in an opinion-as-fact sense) because it is a blatant violation of the Principle of Least Surprise, especially for those coming from C-style backgrounds. If you are a C-style language (which JS is), and you do something in a non-C-style way, you should have a very good reason (besides implementation artifacts and/or not thinking robustly about scope). I haven't seen anything indicating that var's behavior is a well-reasoned, principled choice - as far as I know, it's just how JS 1.0 worked, and we're stuck with it.

In my opinion, unnecessary surprise as a result of historical implementation artifacts is a design bug.

Yep, need to learn about it. Yep, you can, and write good JavaScript. But that doesn't mean the design is good.