The article lists two arguments against Currying:

  1) "performance is a bit of a concern"
  2) "curried function types have a weird shape"

2 is followed by single example of how it doesn't work the way the author would expect it to in Haskell.

It's not a strong case in my opinion. Dismissed.