Now group B decides not to deliver any water to group A. In fact, group B decides that since they don't want to waste their water on the infidels of group A they would much rather just wait untill they all die out.
You make a lot of assumptions and emotional appeals. But no, group B cannot be forced to deliver water to anyone. They can trade it voluntarily with others.
This example has little to do with assumptions or emotional appeals. It is how humans interact. History is absolutely full of examples of groups unwilling to share their natural resources and
many people died in wars fought over resources.
If this group B cannot be forced into cooperation then what use do your DRO's have? Exactly, zero.
Group A
will start a war for water because they have no future anyway. Instead of stabilizing this plateau of trust you put societies in a ring and go "Fight!". Sure, a winner will emerge, but at what cost? And who would be willing to pay this cost?
The logic is that science allows us to check the premises of our gut feelings of ethics.
It makes ethics more rational.
With universally preferable behavior the scientific method is used to come to a rational framework for ethics. This also allows to check the "gut feelings of ethics".
But then it must be wrong because as i've explained before there are no universally preferable behaviours. What is or is not preferable behaviour is defined by a point of view. Points of view are not universal.
Maybe you can give an example of one of those universally preferable behaviors?
PS. you still haven't explained what the actual function is supposed to be for these DRO's.
They are these mystical entities that somehow manage to fix these big issues between groups of people that in the end have to fix their own problems.