I think trying to reduce the complexity of error handling is the original sin. Thinking of error handling as a separate case requiring separate mechanisms is the original sin. The structure of your code should not reflect any difference between "error" and "success" cases. They are equal, and neither should be subordinated to the other.