Yeah, I don't get it either. In isolated testing, redefining the position of the pinned object has zero effect (since pin is basically "set position every tick", so it can't be overridden), so it's not like that's what's buggering up the behaviour.
But then, you shouldn't be able to reposition the object at all, in that case.
It must be the mirror function that causes this hiccup, but I don't know if it's a bug or just a result of the the way that the function works.
In truth, I think I've confused myself now :P
Perhaps @Ashley or someone else more familiar with things could shed some light on this?