One of Graham Harman's commitments that is most difficult for many people to accept is the view that there are no perfect recipes.
It's one thing to say that the chair and a description of the chair are two different things. But Harman holds that any description of the chair will leave something out. So even God could not have a cookbook that would allow him to produce the chair such that she can predict all of the chair's possible interactions, and the properties that emerge as a result, with everything else.
There are two connected ways to take this, both of which I think Harman endorses: (1) as a claim about metaphysical emergence, where there is always the potential for genuinely novel, genuinely surprising (even to God) properties called forward when the chair interacts with other things, and (2) a claim about the limits of representation.
I'm very interested in the extent to which these two claims mutually re-inforce one another.
Mark Silcox and I have a paper in American Philosophical Quarterly that I think is relevant to further working out (1) (or at least working out a Harmanian, perhaps to be contrasted with Deleuzian, theory of emergence).
But (2) itself is a little bit delicate. Again, could an infinite representation (shades of Maimon and Hegel here) of the chair allow one to predict all of its possible interactions with other elements (by "interactions" I don't just mean movement through space, but all of the properties that emerge as a result too)? For Harman, such a representation would just be another identical universe! So it would not in fact be a representation.
Another way to put the objection, not involving a mirror universe, might be to say that the "infinite representation" would be nothing like actual human languages, with recursive syntaxes. I mean there would be no computer program to determine whether something was a sentence of the language. But then the "representation" would be so transcendent that it would at the very least be as weird as the paired universe.
I lean towards this view, and it in fact follows from my piece with Silcox, but it is one that scientistic philosophers would lean against. They would say that the representation might be in some sense infinite, but in the kind of way computers can generate infinities through recursion.
I don't know of a knock down argument against the scientistic philosopher here. On the other hand, I can't think of one plausible reason to believe it, and naturalism's bad track record with respect to mathematics, the advent of matter, normativity, and run of the mill emergent phenomena such as colors, makes me deeply skeptical here too. I mean, I do not think that the null hypothesis should be that Harman is wrong. Rather, I think one can (and I intend to, actually) argue that the burden of proof is on Harman's opponents here.