Well done on the amount of encodings!
I'm just wondering if the direction this is going is towards the centralised approach, with one entity doing over half of the throughput.

Just a question, not a fact.
Fact is, it does seem to be as explained in the post.