Re: open conflicts

[WFreitag@aol.com]

In a message dated 2/3/01 4:49:44 AM, kenlu@MIT.EDU writes:
>>The resolutions of each conflict are authored in advance, so there
>>can be at least some narrative thought/data embodied in each of them.
>The
>>player is not directed but if she happens to create the conditions for
>a
>>resolution of an open conflict through volitional actions or even just
>random
>>mucking about, a resolution event kicks in.

> (snip)
>The reason I bring this up is because this seems to perfectly fit 
>your "open conflict" system.  In EV, you can just keep on being a 
>mercenary if you want, but there are scripted stories ready to happen 
>elsewhere in the galaxy.  Those stories are pretty linear, with maybe 
>a branch or two at most, but that's offset to some degree by the 
>relatively dynamic nature of the galaxy as a whole.

Yes, this is the basic idea. The key shortcomings of the modular scripting in 
games like EV and the example I used, Zelda 64, are that the resolution 
conditions are too highly specific (a characteristic of "missions" that most 
types of conflicts in actual stories don't share), the modules have too few 
distinct resolutions (not that a large number are needed, but "succeed, die 
trying, or give up" often aren't sufficient), and that the granularity of the 
modules is too consistent, giving the whole action an episodic feel. The 
"conflict objects" I was contemplating would improve in these areas by 
allowing a more general definition of a conflict. 

For example, a conflict could be a character not believing he can live up to 
the name of his famous heroic father. One resolution would occur if at any 
time the character's heroic-ness equals or exceeds his idea of his father's. 
This resolution condition could be brought about as a result of any events 
that raise the former (e.g. as a result of heroic successes) and/or lower the 
latter (e.g. learning that the father wasn't so perfect himself). A different 
resolution would occur if the character learns that he doesn't need to judge 
himself against his father at all; this is triggered if the character's 
self-reliance score (raised, perhaps, when the character solves problems or 
forgives others) hits a certain threshold. A third possible resolution occurs 
if the character dies. 

Note that this example would not be a very appropriate conflict for a 
player-character, since players don't like being told how their character 
feels about something. But it would be fine for a supporting NPC. Also, this 
conflict appears to have to be "built into" a specific character, but other 
types of conflict object could be designed to arise dynamically during the 
story. As in EV, missions (blow up the enemy's supply outpost) and goals (get 
a better ship) are simple examples.

>So yeah.. I think this open conflict system can work okay for now, 
>and EV was indeed one of my favorite games, but I think they're only 
>a bridge to more dynamic stories.

I agree that current games are a bridge, although I think my example above 
would qualify as descriptive of the "more dynamic" version you're speaking 
of. It's what I mean by data and process co-existing. Obviously the conflict 
object embodies a lot of process, in the form of rules and mechanisms using 
self-esteem scores and so forth. But it's also "merely" an evolved branching 
story. The nature of the conflict, its milestone events while it is active, 
and the resolutions are all scripted, so it still embodies narrative quality 
crafted by a human author.

There are still some deep issues here. Defining the idea of a conflict object 
doesn't tell us how to use them. Clearly, such a conflict object could not 
exist in a vacuum. Events like "character forgives others," "character solves 
problems," or "character achieves heroic success" would most likely occur as 
part of the resolutions of other conflict objects. So we'd need a whole web 
of them. (Sound familiar?) And that whole web has to live on top of a 
simulation engine in which the individual events take place. (Or does it? 
Maybe it's turtles all the way down.) The key challenge in conflict web 
weaving is how to make a conflict give rise to subsidiary conflicts likely to 
lead toward the resolution of the parent conflict. For example, the character 
in the example shouldn't just wait around for his self-esteem or 
self-reliance to be changed by random events. The existence of the conflict 
should cause him to act in ways that are likely to lead to resolution 
conditions. For example, the conflict might cause him to take risks and align 
himself with his father's cause, which in turn makes it more likely that 
he'll do heroic things, solve problems, and/or get killed. But this is a 
simple example of what is in general a very tricky process. Characters aren't 
always motivated to act in ways that resolve conflics. We may need other 
causal mechanisms as well.

Generalized confict objects are therefore too far for me to go in one leap 
(both as programmer and as author), so I'm working on simpler and more 
constrained forms of "conflict object" that operate in a more limited 
context, such as the simulated world of a typical computer game, and 
interconnect with each other in more systematic (less complex) ways. They 
involve a bit more scripting and a bit less process, so I'm definitely 
working on the data-intensive side of the divide, oddly unperturbed at having 
just learned that it's an evolutionary dead end.

- Walt