I'm not sure what he means by the naive scientific method. It has always been a fundamentally adversarial process, nothing new is being discovered here. The sc

I'm not sure what he means by the naive scientific method. It has always been a fundamentally adversarial process, nothing new is being discovered here. The scientific method as it has always existed requires two heuristics from you, the scientist: a hypothesis generator, a test generator (and an all important simplicity measure). The point of the method is that by making these things work adversarially, we can turn our very messy, fallible creativity, ie these heuristics, into facts about what the model isn't. Moreover no additional hypothesis of what the model is, or test which contradicts it, can derive a false fact of this form, relegating the entire concept of adversary to the meta who-gets-their-name-on-the-paper game. If Daniel has only recently started "using science adversarially", then he has only recently started doing science.