Why Do B+ Trees Have a Minimum Occupancy of 50%?

Question

As far as I can tell, the basic operations (add, delete, search) on a B+ Tree would function the same if minimum occupancy was 1, or 1/3, or any other function on node size.
All the sources available give the minimum occupancy as 50%. Why 50%?

Answer 1

Minimum occupancies of greater than 50% don't work, because when you need to split a node because the node is too full, it would be impossible to cut a node such that the two resulting nodes were greater than 50% full. (You could do 60/40 or something like
that but the smaller node would always fail the data structure's invariant of each node being greater than 60% full, for example)
Minimum occupancies of less than 50% waste space because the goal is to need to read in the fewest number of nodes from secondary storage / memory, and increasing the amount of empty space in each node means that the average request needs to read a greater
number of nodes from disk. (It would also be tricky to implement effectively given that the split operation needs to split into 1/(ratio) chunks instead of 2 chunks)