The bone that moves the whole rig is generally called the root. Whether it’s good or bad depends on what you want to do. For a lot of purposes, though, it’s bad. (Really depends on whether you have some other center marker for the armature-- what purpose the root bone serves.)
So here’s someplace where it might be bad: you want a walk cycle for a video game. In that case, you should probably have a “walk-in-place” kind of animation, because it’s really only about graphics, right? Walking doesn’t actually mean anything in particular. But your game code needs to know where your character still is, so it can check collision, etc. So you want your game code to be able to control the object (which might be as an object, or might be as a root bone, doesn’t really matter) using game code at the same time that you tell it, okay, play this frame on the model, then play this frame.
If you’re not doing game engine work, that can still matter-- because maybe you want a walk-in-place so you can just play a walk animation while you have the root bone follow a path or something, right? So you can edit the details of the animation separately from the gross structure of it, so when the director/client says, “Big change of plans! He needs to curve around to the table and pick up the banana!” you’re not totally screwed.
So I said, structure of model matters. Because what if you had a root bone, and it had one child, only one child? That’d be like another root bone, right? Not technically at the top of the hierarchy, but it’ll do exactly the same thing. So if you were animating from that bone, it’s fine-- you tell your game code, that’s where your model is; you have that bone follow the path.
But also, most of that stuff is kinda hacky-- it’s about minimizing effort, not about maximizing the quality of the animation. A really great animation isn’t a walk-in-place with a follow path on the root. That’s a lazy way to do it. So if you’re working for Pixar, you’re probably not worried about that.