What Would Show You Wrong?

A colleague asks what would show your claim wrong. He wants to know where to push, and he is offering to do the pushing. Someone else asks the same six words and is not waiting for a list: his question expects the answer nothing, and has a use for that answer already prepared. And a third thing asks nothing at all — it moves from I can find no test for this to this is not so. The first is a courtesy, the second a criterion, and the third a conversion. Only the first has anything to do with what exists. This article separates the three and defends an ontology that proceeds by disclosure against all of them. It argues that such an inquiry can carry the discipline in full; that the criterion is contested on its own ground and is not owed; and that the conversion has a name, an ancient origin, and a limit that blocks it. The discipline is met by five things rather than by testability: admissible presentations, negative controls, defeaters, public warrant, and the revisability of the instrument. The fifth is load-bearing, and the argument for it is a record rather than an assurance: three regional trials in which the instrument lost — a region that restricted what could be brought to it, a case that returned no verdict where a negative would have been the easy course, and a control that returned a verdict against the trial's own instrument, barring it from becoming a criterion. Access that cannot lose is construction; access that can lose is access. A second difficulty concerns claims that appear immune to evidence of every kind. Three levels are distinguished, with three registers of assessment: that a constitutive character operates, which is not a claim about texts or tests; that it appears to whoever encounters the phenomenon, which is falsifiable by a thinker who plainly faced it and shows no trace; and that it appears ontologically, as the differentia of a category of being, which is falsifiable by a single text. Demanding one register for all three is the error, and it is what makes claims of the first kind look evasive when they are claims of another kind. The conversion is then traced. Parmenides forbids knowing or uttering what is not, and the ontological conclusion follows without a step being visible, because thinking and being have been identified: at the founding there are not two ends for an inference to run between. The Megarians make the same move where no identification is available — a thing is capable only while acting — and it is therefore a thesis, and refutable, which Aristotle refutes while conceding the priority of actuality. The verification criterion performs it in a further register, and the falsifiability criterion in a fourth — though the fourth is not Popper's, who distinguished demarcation from a criterion of meaning and declined the identification. The conversion already has a name. Roy Bhaskar called it the epistemic fallacy in 1975, connected it to actualism, and later traced it to Parmenides. The article credits the diagnosis rather than claiming it, and states what it adds: the mechanism at the origin, the Megarian step by which a condition becomes a refutable thesis — a step outside the concept of actualism Bhaskar was working with, as his own footnote on the modal sense shows — and the fourth register. Against the conversion stands a limit: what has occurred cannot become never-having-occurred. Having-occurred is not a hypothesis awaiting trial, and a criterion admitting only future-testable claims has no register for it. Rather than record that its instrument does not reach, it tends to say there is nothing there. The rule that blocks this — non-disclosure licenses openness, not negation — is the same rule that keeps a regional trial's verdict open, which is why the diagnosis is not imported to defend the doctrine but is the doctrine's own discipline recognised in a wider setting. Five conditions under which the argument fails are printed, and the one likeliest to succeed is named.