Roberto,The formalism and math part is definitely an area where I am weak in understanding. :) I think I did adopt the set-theoretic view only because I am used to seeing it from Rosen.Yes, I think that a self-referential structure would be a possible solution. But would that satisfy the view presented in the sci-am article, of some deepest theory with no inputs? I am not sure, myself, but it seems that self-reference is a different way of avoiding the infinite regress than what the author proposes in the article, since each theory still receives inputs in the self-referential case. Perhaps this is what you meant, I am not sure.Wouldn't the self-referential structure make the sequence of (mathematical) theories essentially noncomputable, since - as a dynamic system - they must all be solved simultaneously?Regards,Tim