[MUD-Dev] RE: Realistic Ecological Models, Differentiating Areas by Difficulty, and Socialization

[Sean Kelly]

From: "shren" <shren at io.com>

> What I want is more complex systems.  Fixed spawning is an
> ecology.  The monsters grow out of the ground at points.  The
> issue is that it's a dull, dead-end ecology, where nothing the
> players actually do matters.

> The oppisite, which I guess I'll call free spawning, is:

>   1) There are a fixed amount of monsters in the world.

>   2) Slain monsters respawn out of living monsters.

>   3) Monsters form groups and wander outwards seeking lower
>   population density.

> Now, everything the players do matter.  Every slain monster causes
> a population shift.  The monster population center of the world
> will be kicked around like a soccer ball as players hunt the best
> waves of monsters like surfers look for the best waves.

> Neither system, in my mind, is ideal.  What I'm wanting after is
> deep play.  That's why I find myself looking at cellular automata
> of different types and trying to see how the technology and
> methods there can be adapted.  My foe is predictability.  In both
> of the simple examples above, the monsters are predictable.

Definately.  And I would really consider either aforementioned case
a spawn-based system.  The only thing that changes is the spawn
location.  And the second example is in many ways worse than the
first, as the monsters will end up concentrating in the least
traveled (and thereforelikely the least "fun") parts of the world.

When I think of an ecology, it's more along the lines of UO's
original model, but expanded.  Monsters breed more monsters,
regardless of location.  They compete for space, the smarter ones
have goals and agendas.  They are, for the most part, social
creatures and work to preserve their group.

To me, the most interesting parts of the ecology is the AI,
particularly the social aspects.  I'm interested in how creatures
work together in a group, share knowledge, etc.  And this applies to
townspeople as well as monsters.  Especially so, in fact, since a
player may not notice if a pack of wolves is displaying social
behavior, but they certainly would notice if NPCs did.  But by
extension, anything that was developed for an NPC should be able to
be applied to monsters as well.

I've been thinking about knowledge graphs and how they could be
applied to arbitrary situations to produce a reasonable decision.
And how this knowledge could be transferred between individuals or
accessed centrally as group knowledge.  It seems, as Kwon Ekstrom
said, that mobs should have some group knowledge that could be said
to have been taught as they grew up, and then personal knowledge
that they learned on their own (really, once the game simulation
began).  This latter set of knowledge would have to be communicable,
and if evaluated highly, be added to the group's store of knowledge
rather than just passed from indivudal to individual.  I've got some
ideas, but I need to do a bit more reading to see if anyone has
thought along similar lines before I go wasting my time reinventing
the wheel.  It seems like there are fields of AI out there that have
gone in this direction, but I have yet to see whether they've
applied the concept to "chatterbots" or whether it was strictly for
solving difficult mathematical problems.

_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev