The Consistency of Negation as Failure

Abstract Clark's attempt [1] to validate negation as failure in first order logic is shown to contain some fundamental errors. In particular, we show that the motivation for the completed database, the definition of the completed database, and the attempt to validate negation as failure in terms of it are illogical, that the completed database cannot be regarded as the intended meaning of the database, and that the closed world assumption is generally absurd and, in any case, irrelevant. A validation is given using a consistent first order extension of the database and hence in the only terms which appear to make any sense, namely, consistency with the database. However, it seems that the query evaluation process, with negation interpreted as failure, is of no practical use as a theorem prover.