I believe what you’re pointing to throughout most of the article is the idea of classical testing, as opposed to a Mockist approach. A classical approach defines an input and an expected output, and doesn’t care how you get to that answer. Martin Fowler has an interesting article comparing the two and goes deep into the alternative, mockist: https://martinfowler.com/articles/mocksArentStubs.html

Would be interested to hear your thoughts on the mockist approach, do you think there is room for both?