Estimating Success is Not Simple and the No Miracles Argument May Still Be Unsound: A Reply to Nyström

Howson (2000) argued that the No Miracles Argument is a Base-Rate Fallacy: It only works if truth is prevalent among theories in the first place. Dawid and Hartmann (2018) offered a fix: If one calibrates the success of the next theory on the rate of success in the given field of research, then success does confirm the expectation of truth. In (Boge, 2020), I pushed back on this approach, by nurturing skepticism about Dawid and Hartmann’s success-transfer and by showing that, in certain fields of research, it leads to the rejection of other premises of the No Miracles Argument. Nyström (2025) has reacted to my paper, defending Dawid and Hartmann’s approach by means of a deeper Bayesian analysis. I here offer a response by further improving the depth of the analysis and by commenting on the scope of Nyström’s arguments.