I'd recommend you, too, read the article I linked in the comment above. To characterize Wallace's words as an "argument", as if he were a lawyer in the court of being—as if such a thing exists—, is to employ cognitive decoupling, and to miss the point.
There's nothing wrong with C.D.—it's absolutely useful and necessary to have contexts in which words mean one thing and one thing only, but it's equally if not more necessary to have contexts in which that constraint does not hold, in which we are free to make meaning by use of metaphor and non-literal comparison. Indeed, it's the only way in which we can point to dimensions of experience that do not yet have names; it's how language evolves and expands, it's how technical domains come into existence.
I'll bet you're an engineer, data scientist, or other technical professional, not simply b/c you're here, but b/c your response to Wallace's use of ambiguity and multiple layers of meaning reveals as much about the nature of your own relationship with language, which I'd guess prefers that things should be clear, precise, and have singular meaning. (I'm certainly oversimplifying you, so apologies.)
Look into "cognitive decoupling", and you may find a dichotomy that illustrates some of your reaction to Wallace.
https://everythingstudies.com/2018/05/25/decoupling-revisite...