One brain or three?

[WFreitag@aol.com]

Andrew, thanks for giving me the feedback to guide me to some sort of 
understanding of what you're doing. I've read the papers you referenced as 
well. From this I see that my "pizza" examples weren't very enlightening, as 
Hap includes concepts (specifically, success tests, context conditions, 
conflict lists, and goal priorities) that address those exact problems. The 
points I was trying to make using those examples _might_ still be valid (in 
other words, these mechanisms may prove to be inadequate, or difficult to 
apply, in a realistic story domain), but that would need to be explored 
through more thoughtful examples or through actual research.

Since the key issue seems to be the information flow between the "reactive" 
portions and the "plot-minded" portions of the behavior hierarch(y)(ies), I 
have to revisit one of the basic questions we started with, in more technical 
terms. That question is, does your system have a single Active Behavior Tree 
(ABT) that includes all of the characters' behaviors as well as the story 
manager behaviors, or do characters have individual ABTs separate from the 
story manager's ABT?

I ask because there are default mechanisms for interactions between behaviors 
within an ABT (such as succeed- and fail-behavior messages and conflict 
lists) that weren't originally designed to be used for communication between 
one agent's ABT and another's. It's also possible that the question is 
unimportant in the long run because you'll be building way beyond the 
defaults so you might ultimately reach the same results either way. (There 
are lots of possible mechanisms such as the emotion model that can interact 
globally with behaviors regardless of which character's ABT they're in.)

For those not familar with Hap, what I'm asking is essentially whether the 
(non-player) characters each have their own minds which are being manipulated 
(as if by telepathic mind control) by the storyteller, or whether the mind of 
the story manager subsumes everything, like a novelist creating internal 
mental models of characters within her own mind. Conceptually the question 
may be unimportant, since either case could probably model the other without 
the audience knowing the difference. But it's structurally important, like 
knowing whether you're building a high-rise apartment building or a block of 
row houses.

- Walt