Just to be sure, I went over it in as much detail as it has to see if I still agree with it, and I so.
There are at three two points on the roadmap though, that's open for contention because they're unclear:
Faster Block Propagation
The roadmap gives an example but makes it clear that it's just an example. Block propagation is something that can be done either within a single node ecosystem, or across the entire networks ecosystem. At this moment, the only contenders I know of is graphene and xthinner, both of which have displayed impressive stats.
However, graphene is old and mature at this point, so would be a safe bet that we could do right away to improve performance. It doesn't touch concensus, and it synergizes with good transaction propagation and mempool convergences, traits that are desirable to work for as they also help with double-spends, user experience and could synergize with pre-consensus systems as well.
Pre-consensus
