It would be great if you could read the actual proposal and show where the problem is specifically, rather than just list your a priori assumptions.
One challenge/complexity would be that there is an time-bound asymmetry between the network cost, and the pay for sending (or for relaying) a transaction.
The Bitcoin network cost = Data Size * decentralization.
The transaction must be stored on every node and forever, whilst the fee is paid once and to only one recipient. Bitcoin solves this using a free market solution of the block solving lottery. The assumption here is that the block reward is sufficient for the miners to "tip out" to the nodes. To make sure that they keep at it.
Peter Todd twitter-quipped that the block size is a barrier to entry, he is right, but this is balanced by another barrier, the data size. This grows at the rate of the block size. So the bigger the block sizes, the more transactions can be processed, but also the greater ongoing cost for storing all those transactions for an indefinite future. So a cost-payment asymmetry. The one-time payment must cover costs for an indefinite future.
Getting this balanced is not easy within the paradigm of a protocol. This is how we became stuck with a block size limit which we hope to outgrow over the next few years.
Justus's brilliance is that he sheds complexity by simply ignoring it.
Yes, that is the way to do it. The complexities get simplified at the time the code is written. Unless they can't be simplified, and then the issue doesn't get solved.