I've been a long-time lurker, and I thought it would be time to come out of
the shadows and start [trying] contributing ;)
> Thus spake Damion Schubert...
>> This depends on whether or not you figure out how to make your randomly
>> created
>> content both engaging and interesting. Is it possible? Maybe. It's
>> worth
>> noting,
>> though, that the randomly created content in both Diablo (map) and
>> Daggerfall
>> (maps and quests) were pretty miserably bad.
>
> Naturally - how good the system is depends on...how good the system is :P
> I have high hopes, that, even if it isn't an immediate solution, it'll
> help identify what is the solution, however.
Does the quality of procedurally-generated missions depend upon them being a
natural outcome of the environment in which the mission-producing NPC finds
itself? Procedural algorithms produce results only as good as the assumed
inputs, so if the algorithm is running on what amounts to approximations of
the factors that would elicit said NPC to issue a mission or quest, then if
those approximations are off, the mission or quest will be, well, "off".
The approximations I am most thinking of are those assumptions on the
motivation for the quest. If the motivation is not an immediate, actual
need of that NPC, then there will be a noticeable disconnect between the
mission requirements and reality, such as it is in an MMO. "I need you to
kill 10 swamp rats because they're eating grain stores," make sense only if
upon completing said quest there is an actual noticeable improvement in the
lot of the NPC issuing the quest. "I need you to go take out the Orc Bandit
outpost that has popped up at the edge of the city limits," only really
means anything of true substance if that outpost's presence is actually
hindering the NPC town's situation- implying that once gone, the NPC town
will actually perform better (grow, pay more for goods, etc). The
implication, then, is that the creation of the procedurally-generated
mission must be elicited by environmental changes that have actually
influenced the state of the mission-issuing entity. Completion of that
mission must actually allow for the issuing entity to experience a net
benefit, and this benefit must be [potentially] observable by the player
base, though it does not necessarily have to be immediate.
It seems that a virtual ecology of sorts is what is required, with the
mission-generating algorithm a natural part of that ecology rather than a
"contrived" system placed upon it. I think the missions in a game world
are, simply put, the most advanced forms of entity interaction; like the
combat scripts but at the more abstract level of spawn spots and their
interaction with other spawn spots.
I realize this implies a rather significant revisiting of how the game
environment is run, but I feel like we have a toehold on how to really
capture it. Besides, how exciting will it be when we see individual spawn
spots making calculated decisions on courses of action to maximize their own
benefit? It reminds me of the Southpark episode where Cartman raises sea
monkeys who actually think, and begin to create statues in his honor and
worship him, eventually going to war over their Cartman-centric religion and
blowing up their fish tank.