[MUD-Dev2] [Media] Informal poll on MMOWatch

[John Arras]

On undefined, Mike Rozak <Mike at mxac.com.au> wrote:
> 
> A small (informal) poll on the mmo.rpgwatch.com web-site just caught
> my eye:
> 
>     Favourite RPG element?
> 
>     NPC interaction  (33.77%)
>     Character development  (29.87%)
>     Combat  (4.16%)
>     Exploration  (24.68%)
>     Puzzles  (1.3%)
>     Questing  (6.23%)
> 
> 
> If these numbers are right, why do MMOs (except perhaps Bioware's
> secret MMO) spend so LITTLE time on "NPC interaction" and
> "Exploration", and so MUCH time on "Combat" and "Questing"?

I don't think it's that people spend so little time on it, it's that it
takes so much time to get something that's good compared to spending the
same amount of time on combat. Combat systems can be developed
incrementally and even the most advanced combat today in an FPS can
trace its roots back to the Combat game on the Atari 2600.
However, NPC interaction gets into the realm of hard AI where a simple
solution you might start with such as conversation trees won't keep
leading to deeper NPC interaction. It is possible to do a very good job
with this such as in games like Kotor, but there isn't anything going on
that's emergent or changing because you've already mapped out all of the
options for where the conversation can go. You might get one or two good
brief interactions with an NPC before moving to the next one, but could
you come back to it 20 hours later and have it say something meaningful
to you about your last 20 hours of play?
 
So what other possibilities are there? One option might be to give the
NPCs some kind of internal state and motivation, and also a memory and
knowledge about the world they live in. This would include knowledge
about each player they've interacted with and relationships between
entities. Then the world itself probably has to probably be dynamic and
changeable by the players to make the NPC interactions meaningful as
time advances. This gets into other issues such as knowledge
representation and search and learning and many other areas of AI.
 
Much better NPC interaction requires a whole new way of looking at
things and a lot of machinery must be built to go beyond something like
conversation trees. You can't just keep adding more branches to your
conversation trees in the hopes of getting richer and richer
interaction. People have been batting this around for decades so a huge
amount of work has gone into it, but it's a really hard problem to
solve. I think the issue is that it requires a certain level of
complexity to be good enough to use and the jump needed to reach that
complexity is so high that people give up before they get there and
often end up with a broken, unsatisfying system. Peter Molyneux has
spent years working toward this with B&W and is going down this path
with Fable2.
 
John