abstract: It is well known that there should be an analogy between the arithmetic of rational points on varieties over global function fields and over number fields. This analogy should be confirmed also by some deep conjectures, like for instance Vojta conjectures. Nevertheless there are some substantial differences. I would like to make some comments on some of these differences. It is my opinion that once we better understand the reason of these differences, we will be able to push forward the analogies between the theories and the involved methodologies.