There's a third aspect to that coin.
The target of a given key reference might have any of three relationships to the referencing element, and it might vary from context to context.
The target might be
1. In a different root map.
2. In a different scope in the same root map.
3. In the same key scope.
By using standard keyref syntax and not imposing a new keyref syntax component that forces the assumption that the target is 'somewhere else,' we give map authors the flexibility they need to configure all three cases.
Chris
