But it turns out that you can live in the IO Monad and just write Haskell as if were a strongly-typed kind of Python, and quickly get to the point where you are doing something useful, and benefiting from the type system and lazy evaluation (e.g. being able to define an infinite list then just consume as much of it as you want) and the nice syntax for pattern matching, and so on. Then you can gradually add in more and more Haskell concepts as and when a) you feel ready and b) you encounter situations where they'd be useful.
Jumping straight in with monads is the wrong way to do it (and monads are actually easy if you let them sneak up on you rather than trying to chase them, like cats) and the monad tutorials you find on the web are all written by idiots who think they a lot cleverer than they are. Burrito Spacesuit guy, I'm looking at you.