I've not given it much thought so far, but there's at least one thing that stood out:
The main advantage of this protocol is to protect against 51% attack. In order to overtake the
existing network, an attacker will need to increase the network capacity in proportion to the ratio
of the number of UNIs in a public network to the number of UNIs in a false network. Since the
number of UNIs depends on the number of coins, there is no point in attacking the network in
which the attacker has significant property, so the amount of UNIs available to an attacker will
be significantly less than the number of UNIs in the public network.
I guess that's a logical flaw.
If an attacker were to try a double spend, the network he would attack would always be the one in which he has less coins, i.e. he will have an even greater incentive to attack that network.