great work & presentation, shame the implementation got put on hold! we now have workarounds to make emulating this not unpleasant (the defun library), but I'd love to see it in the base language. also dorchard jumpscare at 20:10

Very nice, yet that last comment was spot on: you want to have more than "just than just that dualism of abstraction". In effect you want as many higher order type level abstraction "splits" as you need, not just two, something similar to type indexing but at the meta-type level.

His slides are interesting. It seems to be a piece of Haskell program but with a slide title. How can it be done?