> What does a coarse-grained effect type get you?
For monofunctors: reliable error handling, an ability to re-interpret the same IO structure multiple times, better reasoning during refactorings due to referential transparency.
For bifunctors: the above plus explicit domain (expected) error encoding and even more reliable error handling.
My view has been that reinventing imperative programming (IO monad) on math (pure FP) on imperative programming (the machine) brings me no benefit.