Considerable progress was made in semantics/pragmatics by postulating covert operators that enrich the meaning of some constituents as part of compositional semantics. Three cases in point are Chierchia, Fox, and Spector’s exhaustivity operator O, used to compute local implicatures; Bochvar’s assertion operator A, used to compute local accommodation of presuppositions; and Blumberg and Goldstein’s non-redundancy operator R, used to capture cases of intrusion of non-redundancy constraints in truth conditions. But are these operators syntactically real? We argued earlier that A isn’t, and that O must have such a powerful semantics that one can, in near-equivalent fashion, insert it everywhere in the syntax, or encode its effects, operator-free, as part of the semantic procedure. Here we consider R, which purports to derive three types of inferences in the scope of operators: diversity inferences, ignorance inferences, and free choice inferences. Using ellipsis-based tests, we argue that R isn’t syntactically real, and we derive its desirable effects from different operator-free mechanisms. 1. Diversity inferences are lexically triggered presuppositions and not the by-product of a general non-redundancy mechanism. 2. Ignorance inferences reflect pragmatic non-triviality conditions, not the effect of an operator. 3. Free choice inferences can be derived without R by relying on existing and independently motivated theories.