The person you're replying to meant "is intuitive to us" when they said "making sense". As for systems where it's hard to come up with a complete and consisten

The person you're replying to meant "is intuitive to us" when they said "making sense". As for systems where it's hard to come up with a complete and consistent model... - If coin flips were decided by a cryptographic random number generator, it would be intractable for us to extract the seed or even to distinguish the output from true randomness or go-both-ways indexical uncertainty. - Probably-approximate-correct learning is not possible for all models [1]. - If we're unable to eliminate Boltzmann brains [2] from cosmological predictions, that would seem to imply agents should constantly assign near-certainty to being surrounded by heat death and ignoring that is more of an optimization-for-the-cases-where-you're-not-and-things-matter. - If the number of rules is larger than the number of atoms in the universe, we're never going to be able to remember them all or write them all down. 1: http://en.wikipedia.org/wiki/Probably_approximately_correct_learning http://en.wikipedia.org/wiki/Probably_approximately_correct_... 2: http://en.wikipedia.org/wiki/Boltzmann_brain http://en.wikipedia.org/wiki/Boltzmann_brain