I initially agreed with that article, but it than further and further focused on criticizing cargo culting in specific OOP languages (mostly Java). The actual problem is that there is an abstraction boundary introduced at every tiny thing. The fact that these abstraction boundaries are structured with OOP, is only incidental. I see that this is relevant to OOP, but only in so far as this seems to be a disease of OOP-only/first languages and OOP cultures. What I say is that it is true, that what he criticizes is OOP, but OOP doesn't requires any of that and the actual problem is the existence of the abstraction boundaries, not their shape.

You can't make the all generic implementation, than you will get a more complicated formulation of a turing machine. Software is useful in that it narrows down the expressiveness of computation to a single problem. A generic implementation is able to express anything and thus nothing, i.e. it doesn't contain the information of the problem anymore.