Very good post, like it lots.
To our knowledge this problem has not been openly addressed by the academic community without introducing a trusted party as found in the RingCT 2.0 [4].
This is the same as my findings when you really start to up-scale to numbers over a hundred million and in the end the network becomes flooded
with chatter and the time characteristics will fall like a stone.
Do not confuse the word "trust" with control and really Pow or PoS and all the other concepts of "Proof" flying around really comes down to establishing
a type of trust between nodes which is fine with me and if a cluster of nodes (big fuck off machines) are needed to enhance the smooth running and they
don't get direct access to the "Money" then this is a price we must pay.
Academia should not veto practicalities, you have to have a trade off and more to the point admit it also because in the end you get found out
when things go wrong.