For Artificial Humor we'd need to add a confidence matrix of
expectations. - with different layers of certainties.
Humor isn't hard to decode. It's changing context on several
levels simultaneously. Some we expect, which makes it funny
(fits pattern-of-funny, and the laugh typically comes after a
certain set of parameters are met, which requires timing) and
then there is the unexpected change of context, where we are in
a safe environment and the context / world frame change isn't
deadly.
Safe/unsafe - that's another level that needs to be added to the
confidence.
In an unsafe environment, unless one has cognitive differences,
"THIS ISN'T FUNNY!" is a common response; yet then so is
laughter... it "lightens the mood" because one laughs at the
cognitive dissonance to help reconcile it (laughing your way
into acceptance) - turning the XOR into an == as it were.
Anyway, I can visualize the humor program sending and receiving
in pseudo-code in my head but I'm sure somebody else has drawn
it out, labelled it, has humor programs that can comprehend
jokes already.
If they don't, they will, because if I can see it, I'm sure
somebody else 'out there' has too, or will
