He claims to have an argument that P=NP. He's a philosopher, so "argument" != proof. Although approaching P=NP as a philosophical argument does strike me as kooky.

Better proof of kookhood is that he was at AGI mainly to present his work on hypercomputing, which he claimed was a computational system with more power than a Turing machine. One element of his argument was that proofs using hyperset logic (which he said is an entire field of logic nowadays; I wouldn't know) use a notation that can not even theoretically be represented by a Turing machine. These proofs were published in two-dimensional journal articles, in black-and-white print. I did not notice any fractal fonts in the proofs.