Thomas Dalton wrote:
As I said, in some cases (usually where I already know a lot about the subject) I start the stub _before_ I go reading stuff. So that would be impossible.
That's exactly what people are meant to be not doing. You shouldn't be writing based on your own knowledge, you should be writing based on reliable sources. What's the points of writing from your own knowledge if you have to go and check a reliable source anyway? Just check the source first...
I've already explained what the point is, sometimes I find it convenient to throw down a framework before I start digging into sources. If I'm "not supposed to do it" and yet it still results in good articles being written, perhaps there's something wrong with having a rule against it (assuming we actually do)? And I certainly have no intention of changing my approach just because you personally don't prefer to do it that way.