I'm a little confused because I feel like the original post is really talking about how the abilities are depicted in the fictional game world.
I don't think the fact that a halfling could shift a Tarrasque in game terms is the same thing as the halfling literally shoving a gigantic lizard in the game fiction. You can describe it any way you want in the fiction, as long as it in some way matches up with the rules outcome.
Off the top of my head, "I slash at the tarrasque with quick cuts around its ankles, so that it rears back to strike." There ya go, end result, halfling took an action and the end result is the Tarrasque shifted.
In fact, that move could also be described as "I attack at the exact moment that the Tarrasque rears back". Yes, in that case you've got a player describing something that happens to something outside of his PC (that the tarrasque moved, right then), but there's nothing inherent to roleplaying or to the rules that prevents this.
The stuff in the "game world" is imaginary, and while it should be plausible, it's often pretty easy to imagine plausible results if you don't restrict yourself to only picturing an exact description of the final rules outcome.
As far as the cat example ... it only applies if you have a cat PC. "Just a cat" is scenery, nobody has any investment in it as a character. Applying the rules doesn't seem appropriate there. Whether or not you think that's how the game should have been designed, that's the deal with 4e. The rules are there to help referee interactions between the PCs and "everything else". If the PCs aren't involved, then why are the rules involved? Just for fun?
Like the minion debate, it comes down to whether or not rules should act as physics or as constraints on outcomes. 3.x acted as physics, but I always felt like older editions didn't, so in that respect 4e "clicks" for me.