For what it's worth, the book has been published and should answer anyone's questions on the subject. I have it, but I've only just began to read it. The book might be somewhat disappointing to some people in the sense that not everything falls in place once you hear the theory. The theory is rather blunt, but sounds convincing so far.

They have a summary of the theory in the introduction:

"Our brains are engaged full time in real-time (risky) heuristic search, generating presumptions about what will be experienced next in every domain. This time-pressured, unsupervised generation process has necessary lenient standards and introduces content --not all of which can be properly checked for truth-- into our mental spaces. If left unexamined, the inevitable errors in these vestibules of consciousness would ultimately continue on to contaminate our world knowledge store. So there has to be a policy of double-checking these candidate beliefs and surmisings, and the discovery and resolution of these at breakneck speed is maintained by a powerful reward system --the feeling of humor; mirth-- that must support this activity in competition with all the other things you could be thinking about."

In fact, they argue that such facility might be necessary for truly intelligent computational agent:

"... We propose to tackle this prejudice head on, arguing that a truly intelligent computational agent could not be engineered without humor and some other emotions."

And I got a link to this post - which is something I find funny. Not that I feel like I'm debugging anything...

A book about the subject is coming out early next year, called Inside Jokes.

There's also a video from Dennett's talk, but that too ends too short. Nevertheless, Dennett manages to get into the subject matter. You can get the gist of it looking these videos and the book excerpt, but still not quite enough.