While I don't have internal insight to Jane Street, some of their talks and public information seems to support the idea that they are actually doing a lot of similar work to compiler development. They have stated that they tend to do a lot of symbolic logic and analysis, and they make it sound like they create DSLs that traders can more easily use to convey trading strategies.
So OCaml does really well at getting things done, but historically it has been within very narrow definition of the types of things to be done.
We use OCaml as a general purpose tool, and it really excels in that role.
Log analysis compiles logs into an analysis. A GUI repls user interactions.
It all has to do with how you define the abstractions.
It also does seem that OCaml is more popular in Europe than North America but I don’t have any data on that.
"It was pointed out to me in a private communication that the tuple function \x->(x,x) is actually a special case of a diagonalization for biapplicative and some related structures monadicially."
It received 39 pretty enthusiast replies.
I don't expect this sort of discussion on Ocaml mailing lists (please correct me, Ocaml mailing list subscribers!) And I think the reason is that Haskell makes this sort of abstraction very very easy, and makes it pay off, while Ocaml does not. In Ocaml, it's not even useful to abstractly define functors, which are the most basic data of category theory. In Haskell, functors are a typeclass that comes with the prelude and are hard to avoid. And there's no reason why you wouldn't use a biapplicative either:
https://hackage.haskell.org/package/bifunctors-5.5.7/docs/Da...