Sometimes it's nontrivial to show that a theory predicts something, even if the thing is already observed.