The perhaps most important criticism of the nontransitive approach to semantic paradoxes is that it cannot truthfully express exactly which metarules preserve validity. I argue that this criticism overlooks that the admissibility of metarules cannot be expressed in any logic that allows us to formulate validity-Curry sentences and that is formulated in a classical metalanguage. Hence, the criticism applies to all approaches that do their metatheory in classical logic. If we do the metatheory of nontransitive logics in a nontransitive logic, however, there is no reason to think that the argument behind the criticism goes through. In general, asking a logic to express its own admissible metarules may not be a good idea.Until recently, philosophical common sense had it that, on pain of triviality, you cannot accept all of classical logic, allow for self-reference and let your language contain its own truth predicate (satisfying all Tarski biconditionals). The nontransitive approach to semantic paradoxes has proved common sense wrong. It offers a supraclassical logic, called STT, with a transparent truth predicate (i.e., a truth predicate, T, such that A and T⟨A⟩ are everywhere intersubstitutable) and self-reference (Cobreros et al. 2012(Cobreros et al. , 2013Ripley 2012Ripley , 2013. Transitivity, that is, Cut, fails in STT. But it only fails for paradoxical sentences, like the liar sentence or Curry sentences. That seems like a small price to pay for the full strength of classical logic together with a transparent truth predicate. With that much to recommend itself, I think we should try to hold on to the nontransitive approach as long as we can. My goal in this paper is to show that, contrary to recent criticism (Barrio, Rosenblatt, and Tajer 2016; Rosenblatt 2017), we shouldn't reject the nontransitive approach because it cannot express the admissibility of its own metarules.Along the way, we can observe some general facts about the possibility of expressing, in the object language, which metarules are admissible. In particular, I point out that no one who does her metatheory in classical logic and allows for self-reference can allow her logic to truthfully express which metarules are admissible, according to her own logic.