It does look more like a random walk through the Why's. Most worthwhile problems have multiple Why's, some of which you can do something about, some which you can't. The examples where Five Why's actually works are ones where the person doing it secretly knows what the problem is, and just picks a path that will terminate on their desired problem, I guess.

But could there still be some value to this sort of exercise? At least it terminates at something. If it is possible to get rid of that convenient or culturally acceptable blame attractor, at least one problem has be exorcised from the organization, right?