How Can False Assumptions Yield Successful Predictions? Defending Deployment Realism Against Supposed Counterexamples

Scientific realists argue that the truth of theories is the only plausible explanation of the success of novel predictions. Laudan has countered that various false theories made successful predictions, and T. Lyons points out that some of those were novel predictions. The answer of “deployment realism” is that novel predictions do not warrant the truth of a whole theory, but only of the claims essentially involved in those predictions. But Lyons lists a number of false claims from the history of science which supposedly were crucial in deriving novel predictions. This is troublesome, for successful novel predictions cannot be explained by showing that they were derived from false tenets. I argue that they were possible partly because the essential role in their derivation was not played by false claims, partly because they were enough probable (either a priori or in the light of available evidence) to be guessed. There is thus warrant for the truth of claims which are actually essential in the development of predictions which are novel in a precisely specified sense.