I believe I have figured out what Fuserleer's design is doing based on re-reading the descriptions above.
I believe what he is attempting is to define a data structure wherein he can partition double-spends so that they can not cross-partition each other. In other words, once there is a double-spend, instead of discarding it (and unrelated transactions in the same chain), he isolates those transactions which depend on the double-spend and prevents them from cross-pollinating each other in derivative transactions.
The problem with this of course is it ruins the incentive to converge. It becomes a divergent block chain where the incentive is to double-spend and create forks (within the same system).
I cannot speak for Fuserleer, but I can say that this is how the design I am writing up works. Double spends
do not create orphaned branches, they simply become invalid and subsequent, unrelated transactions process as normal. Therefore, you cannot create a divergent mess by double spending, because they can coexist within the DAG - this is only possible because of the eventual total order.
The system must have some means of converging on a consensus choice amongst competing double-spends.
is a discussion about that. We don't need to repeat that discussion here. I urge readers to click that link if they want to read what is being discussed about whether there could be an alternative to Iota/DAG, which is mathematical model which I assert the control of must be centralized in order to not diverge.