A further thought on this...
RELATIONSHIP is multifaceted. If you wanted to think of RELATIONSHIP as a fact associated with a person, it is not *A* fact--there could be thousands of RELATIONSHIP facts to tie to that person. Not manageable.
And which direction do you go with it, relation to ancestors or descendants? Why? Why would it be logical to go one way but not the other?
I can imagine a construct "being decided" to tie generations together--but I just think the birth fact does that. Birth implies parents, so to me, I show that connection to each preceding generation in the BIRTH fact. I suppose you could have FACTS for each person of FATHER and MOTHER, and put source cites for whatever evidence you use to determine that.
What if you do that and (wrongly) create this situation...
In FTM, Mary is entered in the form screen as a child of John.
In Mary's record, the FATHER fact is entered as Paul.
No, that should not happen, and it will not with the current structure. But if an independent fact for parent relationship is created, the door is wide open for entering contradictory data.