Just as a FYI, it is working only because the incorrect parts are being skipped over by the files. As long as you have "enough right", then it should still work.
This is basically what I learned from reversing the certification stuff back in the day:
There are many ways to get working configs, the way "official" screenshots are setup is not the only way. In fact, non of my configs match official screens because I removed extra nodes not even needed that they use.
The direction of the links matters.
The order of the nodes in the packt.dat matters (at least, the first two)
The extra fields are a mixture of visual data (such as icon id state), important data, such as ids to link to, service type, etc..., and a lot of extra data that is not useless, but does not need to be figured out to get the files working.
Editing files by hand is both inefficient and impractical for managing multiple server/shard layouts for the long term.