Feferman (1960) demonstrated that the choice of provability predicate is conventional, since differently defined but extensionally equivalent predicates yield divergent unprovability results.
?Rate how convincing each reason is below to see the overall strength.