In the riveted system the lever is held to the side of the post by a rivet passed through both. This is inherently asymmetric and over the course of a few million button depressions will cause the hole in the lever to wear to the point where the lever becomes liable to break. This is more prevalent where the spring pressure is greater so the first to go is the air button lever, and yes I have experience of this.
The hook under system can still be asymmetric depending on the alignment of the components with the pad position but does not suffer wear in the same way.
It seems to me that the hook system does not ensure such a quick return of the pad to its sealing position when the button is released, though this may just be my subjective impression. The whole lever can move away from the fulcrum when depressed and so take longer to return to its rest position. You have to be playing really fast pieces to detect anything but I think that I have done on occasion.
Perfection would be a frictionless pivot in a split lever that passed both sides of the post.

Interested to hear other opinions.