From reading his later book, I got the impression that he let himself get carried away easily with his ideas. Bayesian thought hadn't yet become respectable; so while he talks a lot about "the logic of probability", and much of what he says about it seems exactly right, he never seems to use the laws of probability or lessons drawn from them. He'll talk about the importance of recognizing uncertainty, and then suggest we should have known a priori the Universe was finite but unbounded. I don't know if he ever really tried to disprove his more unique ideas, or find evidence to distinguish them from alternative hypotheses.

On the other hand, I wonder if mainstream semiotics committed even greater crimes against probability, in which case the honest people in the field might have benefited from some Korzybski.