Would it make sense to make this return the string, rather than assign it to the instance variable result.

Longer term, perhaps we need a bind method like: IO<A> bindLeft(IO<A> io, F<A, IO<B>> f) which binds, but returns the a from the left argument. But this would be another issue. I would have to check the Haskell doc to see if this exists.

This is a way to mock the println call. In general print returns Unit

I know. I thought it would be better to return IO<String>and return s here. Then in the method above you can replace result with readAndPrintUpper.... and remove the variable result. This would improve referential transparency.