Right. And we can at least consider the reductio ad absurdum case of accepting any input and just outputting “No”, or the equivalent generic word of negation in various languages, assuming they have one. But even then there’s the question of whether blank negation is equivalent in all languages. As in, do some languages have polite versus casual ways of indicating negation, and so in choosing one have you introduced a feature for one language that isn’t present in the others? And that kind of thing.
Or to approach this from a slightly different angle: I’m typing this on a QWERTY keyboard. It is quite well-suited to entering the Latin alphabet, Arabic numerals, and common English punctuation marks. With slightly greater effort it can be used to enter accents and so on: sauté. And with somewhat more effort than that I can enter more or less any character supported by whatever I’m typing things into: 叉烧包 (char siu bao, pork bun).
But if I was doing most of my typing in Chinese I wouldn’t want to be using this keyboard. Because although I can use this keyboard to produce essentially anything I could produce with any other keyboard, this doesn’t mean that a QWERTY keyboard is equivalent to every other kind of keyboard.
That is, in some narrow technical sense I can accept that I have a keyboard that can type any language (angels and ministers of UTF-8 permitting) but I’m extremely skeptical of the claim that this is a universal keyboard for any language in any practical sense.
Of course this keyboard wasn’t designed to be a universal keyboard, it was designed for US English. So you could sit down and start making modifications to it to make it better suited to other languages. Perhaps just sorta bolting together a bunch of keyboards for different languages. But even once you’ve done that, my assumption is that you’re still going to have a different experience in some languages versus others. And I don’t just mean in abstract qualia or something like that, but in terms of basic nuts and bolts mechanics. As in some languages are just going to be better suited to keyboard-like input semantics than others.
And that’s the kind of thing I’m talking about with parsers. I’m skeptical (although willing to be convinced otherwise) that the whole parser paradigm or whatever you want to call it maps equivalently onto all languages. And that’s to say nothing of the tacit assumptions built into the typical parser-based game world model.
To be clear, this isn’t me trying to terminate discussion or anything like that. I’m just trying to clarify what I meant when I said that I think any argument about whether or not you could build a universal parser eventually boils down, in practical terms, to an argument about synonymy. As in, because both my QWERTY keyboard and a pinyin keyboard can produce arbitrary UTF-8 characters, are they equivalent? Is a romanji keyboard equivalent to a kana keyboard? And so on.
My point being that these kinds of things tend to end up not being questions of narrow technical capabilities, but squishy just-draw-a-line-somewhere definitional things.
