It is commonly believed that the meaning of a formal declarative knowledge representation language is determined by its formal semantics. This is not quite so. This paper shows an epistemological ambiguity that arises in the context of logic programming. Several different logic programming formalisms and semantics have been proposed. Hence, logic programming can be seen as an overlapping family of formal logics, each induced by a pair of a formal syntax and a formal semantics. We would expect that (a) each such pair has a unique declarative reading and (b) for a program in the intersection of several formal LP logics with the same formal semantics in each of them, its declarative reading is the same in each of them. I show in this paper that neither (a) nor (b) holds. The paper investigates the causes and the consequences of this phenomenon and points out some directions to overcome the ambiguity.