Firstly, I think the model should be based on communications AMONG userspace and kernel developer, not TO the kernel developer.
Secondly, to facilitate this kind of change, instead of showing the code patches, it should be more higher level. It definitely is better to argue and resolve all the discrepancies at the design level, before actually implementing it.
I am quite amazed that this way of presenting the specifications, and then followed by implementation has such a long history back - even before it was OpenSource. Those past ARC before OpenSolaris was released is usually not available.
From the above mailing list, one can see that there are even patches/corrections to the specifications before implementation starts, which is much better that patches AFTER the implementation.