While some parts of the original hypothesis have merit, how would "Stage IIIa" and "Stage IIIb" have originated? Independently? How then would "Stage IIIa+b" have originated? Sequentially? Then "Stage IIIa+b" and the later one of the two others would be coincident. It does not compute.
The way I understood it, both of these stages were hypothetical, and Prum & Brush couldn't decide if the evolutionary pathway ran through IIIa _or_ through IIIb. Only one of these stages would ever have existed then.
It is the lack of "Stage II-but-nothing-higher" taxa or even periods in the fossil record that is puzzling.
I thought *Sinosauropteryx* has stage I and probably II feathers and lacks all higher stages?
As a compsognathid, it fits very nicely in phylogenetic terms. In stratigraphic terms, I don't see your problem, what with *Anchiornis* being apparently Oxfordian in age or so.
The point where I differ is that Prum invoked a convoluted and partially genetically implausible, partially evolutionarily nonsensical two-step mechanism for first- and second-order branching and another mechanism for the rachis, while I'd rather see a branching "algorithm" superimposed on the proto-rachis (which needs to be nothing more than a protofuzz fiber) to get first-order branching and a proper rachis, and superimposed on the first-order branching to get second-order branching.
Ontogenetically, AFAIK, the barbs _are_ rachides that move (literally move) onto the biggest rachis. (Or onto the second-biggest one, the aftershaft.) They do not originate from branching, and they're arranged in a circle before the "conveyor-belt mechanism" starts. In Prum's model, I think, the entire circle is supposed to be homologous to a stage I feather.
Out of curiosity: did mammalian guard hair evolve from the undercoat hair?
Completely unknown, but the opposite seems likely; it has even been proposed that all hairs evolved from whiskers, and various old books say there's evidence for whiskers on the snout of *Dimetrodon*. Of course, neurovascular foramina can mean anything, can't they. :-/