But Haskell_2005 presented more difficulties (and research opportunities) than Haskell_2014. It was always interesting but now it's also proven to be powerful and meaningful in a way nobody had gotten around to at that point.

I disagree; I think, at the time, functional programmers were primarily using OCaml and various lisps. This is purely anecdotal, mind you, and haskell was definitely on the FP radar big time.

I dunno. I remember it being all over LtU and most every programming language discussion was that or scheme, but always treating Haskell like it was the ideal (and it had become default on research papers instead of OCaml)

Also, I remember when this came out the first time and the discussion on proggit was replying using mainly Haskell examples.

Then he's not thinking of Haskell when talking about referential transparency and variable assignment, or the other points of his post.