If C is the same for all K, then can’t he extrapolate the trend from K = 2 (which you’re right, I did misread) to all other values of K? That seems to be his justification for the claim you’re arguing.

I do think some more rigor is needed here (especially wrt to whether his derivation for C is sensical, as you allude to, since it obviously can’t converge to 0) but I hope we can agree that we are well into the “amateur who has trouble presenting as such” territory rather than the “AI slop” territory.