There are two answers I can think of: 1. You know what the code is doing to the extent that you know how monads work. As in, you know that you’re going to do s

There are two answers I can think of: 1. You know what the code is doing to the extent that you know how monads work. As in, you know that you’re going to do some monadic action that returns a, then another one that returns b, etc. The code as written is generic so it’s not specified what the actions are. 2. And this is the much more common case; the code in question is being used within a particular context (I.e. for a particular monad that the programmer is working with) and the context makes it obvious what is actually being done by the monadic actions. Here the genericness is not in writing a function that can be used in multiple places, but in having a portable concept of “a pipeline of actions that return things and could fail” which you can use wherever you need it.