I don't know if that is the thrust of his argument, but if so, I don't think he demonstrated that. His idea of requiring security deposits may in fact address the NaS issue in that people can no longer vote on multiple chains simultaneously without penalty...but what would stop people from trying to prepare chains in advance and then only pay the deposit when they have a winning chain? If that happens, then you are back to square one: People will be spend spending money to build CPUs (or ASICS or whatever) to find better chains, and you will be back to the level of people spending the same amount as the rewards.
I am not suggesting that his proposal is any good or without flaws. I am not even suggesting we diverge from 100% PoW either. However, we need to be open to learning new things and questioning our own biases from time to time.
I Agree.
His paper addresses your concerns with not only deposits but also TaPoS and exponential subjective scoring to address all these possible attack vectors.
Actually, I don't think it did address my concern.
The bit about ESS is mostly describing how splitting the network would be avoided.
And there's some stuff in there about the consequences of the deposits
(what he's referring to as "inefficieny of sacrafice"), but that is something different.
My concern is that "miners" will try to compete to gain more rewards by
gaming whatever PoS algorithm is in operation...and they will use greater
and greater resources to do so, until is no longer profitable. Therefore,
there is no efficiency advantage.
In other words, take away NaS, but what you are left with is
good old fashioned competition.
If you feel he addressed that concern, please reiterate the solution for me.