Many factors detract from MQA being a support-worthy goal (@Bill_Janssen’s comment above about the compression being one, the origami work being another). Still, one should take note that there is a vision that they have, and that vision is not yet complete.
Many people buy into that vision, and not it’s the current state of affairs. The current state is pretty weak, but it’s still the early days. MQA is closer to an objectively better model of provenance than anyone before them. It isn’t just “empty marketing”, it’s just not completed yet.
As for the other pieces, Roon changed the course of DSP + MQA. There exist possibilities to change the course of the future impact of MQA in other ways.