I think you missed the angle of the article. It's just pointing out the isomorphism between Church encoding, the visitor pattern and GADTs. You wouldn't use any

I think you missed the angle of the article. It's just pointing out the isomorphism between Church encoding, the visitor pattern and GADTs. You wouldn't use anything but GADTs in Haskell today, it's just an interesting curiosity unless you're actually a type theorist or something.