If anything, his analogy between the impracticality of fully deterministic NLP and H-M type inference is weak. You want static typing to be deterministic and quickly resolved, or else you'll get nasty C++-like compile times. (Of course, it also has to be smart and consistent, or you just get something that will just make you go through the motions for little gain. Smart static typing can be very useful, but dumb static typing is a waste of time.)
I understand people hating on static typing, though - when most people think of static typing, they think of C++ and Java. I didn't see the point of static typing until I learned OCaml, and that still has room for improvement. (OCaml + Haskell-style typeclasses + error messages that weren't obviously transliterated from French as an afterthought would be a nice start.) There are only a few languages with well-designed static type systems, and the most obvious one insists on being all indie rock about it ("avoid success at all costs").
The Haskell / indie rock comparison is remarkably cogent. It makes all the more ironic that I think it really was better in 2005 before it got 'popular', before the age of dons and reddit-spam monad tutorials.
It's unfortunate that OCaml sucks as hardcore as it does, seemingly everything they added to ML is either useless or harmful (except for camlp4), but unfortunately even fewer people use ML. SPJ expressed interest in a new Haskell-replacement with eager evaluation, but I don't see that coming anytime soon.
I don't think OCaml "sucks hardcore", I just find it a bit frustrating, the same way that Common Lisp is not as good as my ideal Lisp (which doesn't exist, and therefore isn't burdened with any compromises). I'd still rather use it than several other languages. Everything has quirks.