I think you misunderstand - it’s not that you shouldn’t be capable of evaluating and improving the implementation, but that it’s usually irrelevant. A very large majority of the time, “good enough” is fine. Fix things people complain about, and use the rest of your time to do more important things.