I think he argues that any methododology—not just any simple methodology—will fail in some cases. The reason is that there is something “irrational”, that is, irreducibly sociological, about scientific progess. I disagree because I think there is an optimal methodology for intellectual progess (Bayesian inference), and successful inference is ultimately reducible to approximations of it.
Bayesian inference only functions within known solution-space. Spotting things outside of known solution space, while rare, is essential for the progression of science – and can’t be modelled simply as Bayesian inference.
I think he argues that any methododology—not just any simple methodology—will fail in some cases. The reason is that there is something “irrational”, that is, irreducibly sociological, about scientific progess. I disagree because I think there is an optimal methodology for intellectual progess (Bayesian inference), and successful inference is ultimately reducible to approximations of it.
Bayesian inference only functions within known solution-space. Spotting things outside of known solution space, while rare, is essential for the progression of science – and can’t be modelled simply as Bayesian inference.