By comparing different ideas and using the data to draw conclusions. For example it is logical to me that the accountabilities of PO and SM should be separated, but has anyone really tried to proof this? Another example, why 15 not 20 minutes? How should it depend on product domain / team size / complexity?

It seems to me that many of the topics discussed in community are opinions and experiences that can be highly influenced by cognitive bias of individuals.

I would expect defining “quality of proof", known from evidence-based medicine, and trying to apply it to the Scrum Guide content.

In my opinion “The result is not Scrum” has one disadvantage — it closes a discussion how to make Scrum even better.