But I wasn't trying to argue that "low level" does mean "close to the machine".

I didn't think, and didn't mean to imply that I thought, you were. I mentioned it for the same reason you did: to help describe my meaning of "low level" by its connection to something related.

I was after a language which is, as far as possible, free from prejudice towards particular applications and I was arguing that it can be non-trivial to get one. You might dispute my use of the word "low level" for this, but I would say that this is largely a semantic issue and that there is still a need for knowing that we can get a language with these properties.

I don't think that's what you're really after. When you describe what you want, it sounds like a language that is prejudiced against describing things that are complicated in reality, so the complexity of the description matches the complexity of the reality.

It's not just a semantic problem that you're calling it "low level." "Low level" means it's far from how humans think, which tends to remove human prejudice. You call it "low level" because you think you can find it by removing prejudice. You actually need to switch from one prejudice to another to get what you want.

(Also, thanks for the reply. Sorry I didn't read the whole thing, but I got to the list of methods you had rejected, and it was just too much. It feels a lot longer to someone who thinks the basic idea behind all the methods is off base.)