Enjoyed the article as a functional programmer and category theory enthusiast. Haskell does not have to be a strict encoding of Category for it to be useful. We

Enjoyed the article as a functional programmer and category theory enthusiast. Haskell does not have to be a strict encoding of Category for it to be useful. We are grown ups and can understand that not everything is perfect and we can overlook some warts in order to benefit from a beautiful set of rules that help us think about composition as a first class citizen in our work. The first thing mathematicians tell you when talking about categories of, say, sets, we have to do some hand waving to avoid Russel's Paradox. In other words even mathematicians have caveats when talking about different aspects of category theory. In the case of the Functor, we use fmap and not a "function that lifts functions" because it is much more ergonomic. Yet, it is just as "correct" since it is isomorphic to lifting a function and applying it. With monads we use bind and flatmap although the most obvious categorical implementation would be a simple compose function. That maps more directly to the "category of endofunctors", but again it is isomorphic to the same code that uses flatMap/bind, and the same laws hold. A lot of utility has been gotten from analogies with category theory and as long as we hold our noses we can ignore the occasional whiff of in-authenticity. I don't believe that you need to learn category theory to take advantage of the concepts we have borrowed, but it does make everything become more coherent to dig into it a little. It's also a lot of fun if you like that sort of thing.