[bitblender]

I think your disagreements are valid, but I don't think it is fair to say this is an amateurish take or infer the author's level of experience. Your example of unnecessary inheritance hierarchies (which I have also faced many times in real world scenarios) may even be a symptom of exactly what the author is saying: what you might call a "bad fitting abstraction" the author would just call "bad code". The implementation details of how code gets shared (composition vs inheritance) is a subtle but still vital consideration to the cost benefit analysis. The author is observing that it might be misleading or dangerous advice to urge developers to choose duplication just because issues with abstraction have been historically observed, which I completely agree with and do not consider myself to be an amateur. I also agree with you and other posts that the author fails to mention the (exponentially higher) costs of abstraction boundaries that also span human organizational boundaries.