Many times I wonder if it's possible to collapse abstractions? The idea of making the whole stack simpler. I'm not overly optimistic of LLMs achieving this, but it they do, I will be paying more attention.