If you look at this from an "expression problem" point of view, this version with the records-of-functions is very similar to OO programming. But without inheritance, classes and so on.
I'm very much a fan of that POV. I think OO has a lot of mythology, but the technology is relatively similar to something like OCaml or Haskell. Classes are essentially just functions which return "structures".
Inheritance and open-recursion/dynamic binding are usually what remain. I'm not terribly sure I miss them.
Indeed classes are great for these kinds of encodings. It is one of the use cases for which classes are the nicest approach in OCaml.
For example, see this section of Real World OCaml:
https://realworldocaml.org/v1/en/html/classes.html#open-recu...
For instance, here we have Option
For Option, since it's non-recursive, the Scott Encoding and the recursor/inductor/Church Encoding are identical. Here's a linked list, though Anyway, the pattern should be more clear now. These provide effectively "functionalized" pattern matching which you can apply whenever you need. In particular, you can think of these as expressing a (potentially partial) "view" of the abstract type. For instance, my linked list might have not been a linked list exactly but instead some kind of tree, but `scott` and `fold` let me expose a "view" of that tree as though it were a linked list.