Why the no‐miracles argument fails

Abstract

The chief argument for scientific realism is the no‐miracles argument, according to which the approximate truth of our current scientific theories can be inferred from their success through time. To date, anti‐realist responses to the argument have been unconvincing, largely because of their anti‐realistic presuppositions. In this paper, it is shown that realists cannot pre‐emptively dismiss the problem of the underdetermination of theory by evidence, and that the no‐miracles argument fails because it does nothing to dispel the threat posed by underdetermination, although it may be effective against other anti‐realistic arguments.