There is nothing minding blowing here. Just more of the same centralization failure. Well the concept of a DAG is mind blowing in the sense that it allows proof-of-publishing to be orthogonal to consensus finalization, which is one facet of increasing the TPS rate. But the Byteball "stability points" algorithm solution is incomplete. It hasn't solved the holistic problems entirely.
Exactly. It's not trustless. They took the various incomplete research into DAGs (e.g. Bob McElrath) and implimented it in node (node!?). They couldn't find a consensus algorithm that works so they go with a sibyl attack prone trusted node system instead. I bought 1BTC to get a real stake in it though because the community is strong which means they might end up solving these issues, if they do it will be unstoppable.