[MUD-Dev2] The Future of Quests

[Mike Sellers]

Eric Lee wrote: 
> > Cruise wrote:
> >
> My NPC's can decide on what they need, then plan out from the 
> actions available to them the best of way of achieving it 
> over a long period of time. They're becoming better at 
> evaluating how much a particular action or item means to 
> them, which in turns allows them to barter, reward or be 
> rewarded by each other for actions and objects.
> This can range from an NPC asking you to fetch them a sword 
> to the King's brother asking you to assassinate his sibling. 
> Or asking any other NPC.
> >
> 
> I'd be delighted to see a system that can actually deliver on 
> that promise in a robust and scalable way.  It's badly 
> needed, that's for sure!  

I don't know Cruise's technology, but we have been working on a solution
that sounds similar: autonomous agents with their own personalities,
motivations, goals, emotions, relationships, etc.  They learn over time, so
they evolve relationships with each other and with players.  (This also
leads to a unique and robust reputation technology that we've productized
separately.)

I agree with you that this is sorely needed, but you might be surprised to
learn that's hardly a majority view.  A lot of players -- and more to the
point, a lot of MMOG developers/publishers -- like the certainty afforded by
quest-trees (the balance, the quantifiable nature of the gameplay), even
with the increasing cost of content creation, and with the unending grind
this produces.  I believe this view is changing, but not overnight.  

> The problem is that the quest goal has to be something that's 
> measurable in the game system; visit X location, move this 
> object from here to there, find N objects, kill N monsters . 
> . . that's pretty much it.  There's no emotional component in 
> the core quest goals, so they get boring pretty quickly if 
> they're left in their raw state. 

The issue here is that such quests are meaningless.  Quests that occur in a
static world where the NPCs don't care quickly become sterile and ultimately
boring.  To overcome this, the players have to have relationships with each
other, with the NPCs, and with the world that become meaningful to them.  In
part this requires that the players' accomplishments have real impact on the
world -- thus not just a sandbox, but a sandbox that changes over time.  And
in part it requires that the players care about the NPCs -- they have to
have bona fide relationships with the NPCs, which in turn requires that the
NPCs have enough emotionality (among other things) for the players to care
about them.  

Not surprisingly, emotion and resulting sociability are at the core of our
NPC-AI technology. :)

> ... The problem is that creating high-quality 
> emotional content has (so far) not been possible to automate; 
> it has to be created by hand.  ... You may 
> have a king in your game, and the king may view his brother 
> as an obstacle, but the only way the king can ask you to 
> assassinate his brother is if the game designers have 
> explicitly built in the concept of "assassination" as an 
> option.  This is one of the fundamentally hard problems of 
> AI; one where we're always 10 years away from solving it and 
> have been for the last 40.

I know the history going back to the overly optimistic pronouncements of the
1960s, but it's not 10 years away. :)  We've been actively working on this
problem for the past six years (on our own, with DARPA, and with other game
developers) and have made a lot of headway.  Enough?  We'll see.  

> ...One lesson we can learn from movies is that often 
> it's better to skip the emotional content altogether than to 
> include it and do it badly.  There's an "uncanny valley" 
> effect at work here so it has to be done by people skilled at the job.

Actually I think the lesson from the movies is the opposite: nothing trumps
emotion.  Emotion can make a bad story credible, and a good story memorable.
Emotion that's faked or forced doesn't work, that's true.  It's also true
that our industry -- mostly young, white, unmarried guys with little life
experience (my typical rant) -- has spent decades almost uniformly running
in terror from anything that smacked of emotional content.  So while a lot
is made of the "uncanny valley" (which Mori applied to figures and
animation, btw, not emotional action), it's a whole lot easier to cross that
valley than you might think.  Take a look at the work of Ken Perlin in
animating emotions with simple characters if you haven't already.
Communicating sincere, evocative emotion doesn't require incredible
graphics, but it does require a depth of meaning, a depth of character and
purpose, that the "vending machine" NPCs and static storylines we have today
simply cannot muster.  

> So yeah, I think the first person to solve the quest problem 
> will be a hero (well, at least to those of us to care about 
> such things) but I also think there's going to be lots of 
> attempts that die in the "uncanny valley" before we get there.

The former would be nice, and I have no doubt of the latter point.  We've
made a lot of headway already though.

Mike Sellers
Online Alchemy