[MUD-Dev] World and History Creation

Ling ling at kanga.nu
Thu Mar 30 12:08:07 New Zealand Daylight Time 2000

Had to reformat the below significantly.  Hope it comes out okay... (and 
doesn't get rejected!)

Justin Rogers wrote:
> Ling wrote:
> When using fractals to generate your world, given it a
> sufficient size you wind up making bookshelves of cds full of
> data anyway.  Its not that

I would disagree here since seemingly large fractal landscapes can be 
generated without a corresponding increase in storage.  I'm not sure at the 
moment but I think Black & White, for instance, use what is essentially the 
same technique to give the illusion that the world is incredibly detailed.  
All that is needed to be stored is the difference of the original fractal 
landscape and the current landscape (now that a meteor just made a big gaping 
hole).  Pseudo-random seeding, I believe, in the FAQ.  The big drawback of 
fractals is that it doesn't do it's own time-ageing effects.
> the MUD would run on GIS data natively, its that the GIS data is
> used to make a more *realistic* world.  IE, this is the most
> realistic world a human can possibly describe or imagine.  So using GIS
> data which has several powers of complexity, simplifying it to down to 
> classes of terrain, height, etc..., and then finally using the data for
> your simplified world to create a *visible* world for the user makes for
> a very real experience.
> And guess what.  The user will never know that the world is
> better, worse, or event that it is the world.  Pick remote locations,
> stuff people in the Amazon.  Hell, put someone on Long Island, New York
> and they will	think it is a fabulous new world.
I subscribe to the "What they don't know won't hurt" school of thought. I 
argue that whether a mud uses real data, like real GIS data (or the satellite 
data in that bookshelf of CDs behind me) or faked data wouldn't really affect 
whether the user thinks it is realistic that much. I admit, I'm basing this 
opinion off self-evaluations after gaming.  When I play a game, I don't 
think: Why is there a mountain to the east and a river flowing north/south?  
In fact, for a good game, I just get absorbed, accept the limits of the game 
and just play - so long as everything else is consistent.  I'm the sort of 
person to read those Gamer's Murphy's Laws and only then realise (did you 
know that in Champions, a normal person can destroy a car in 5 minutes using 
his bare feet!).

Having said that, I do have a little program that accretes a sphere in a 
solar system (Population II type star), cool it to make a big spherical rice 
pudding skin (representing a supercontinent), land some comets on it, do some 
calculations on heat convection which puts heat under the supercontinent so 
it flys apart into continents, rain on it, etc.  Looks great and incredibly 
detailed but I still load up the terrain editor and write my initials in the 
format of a mountain range over the ice caps.

> > What I'd really like to do these days is take the 36 plots
> > thing and integrate that into an Alife engine so key characters are
> > created during world generation with motivations and stuff.  It would
> > then be possible to generate legends...  Although I suspect an
> > injection of illicit substances into the code might help (considering
> > how bizarre legends can seem now we have an established framework 
> > for "reality").
> [snipped bizarre behaviours]

It isn't the bizarre behaviours that throw me.  It's the bizarre nature of 
legends: like cracking your head open and a daughter pops out; some woman 
floating around in a lake holding not-your-average sword up; and a certain 
witch with a name like our Prime Minister running around in a certain 
I was (too busy commuting these days) trying to figure out how to incorporate 
a local lore/history type thing into the world generation. Simulating nations 
battling out or people milling around doing purely survival stuff is fine but 
a bit barren in terms of history.  What I'd like to do is add in some spice 
in the form of creating key characters at random points during the simulation 
and then taking one of the plots and wrapping it round this character.  
Whilst that is happening, it would be rather funky if the engine also create 
other characters, add them to the existing plot, create more characters and 
plots and draw them into this big intricate weave.
I suppose the best way to make a history is to take the Twilight 2000 
approach.  A bunch of people sat around playing a game of politics based on 
then-present world politics and extrapolated the future using that. After 
all, history only has to be made once.  I seem to have an obsession. :-)  My 
big concern with history is that to give a world depth, not only does it has 
to be internally realistic/consistent/whatever but it should also 
feel "evolved" for lack of a better world.  Many of the ALife stuff I've seen 
have concerned quite dry simulations (admittedly, my own first-hand 
experience of ALife equivalent sims tend to be wargames (of the synthetic 
environment/defence sort) which are dry by definition).

That reminds me, today's science snippet which I don't think has appeared on 
this list before is the "small-world network" (now made famous by the six 
degrees of Kevin Bacon).  All nodes (think npcs within a mud) have a 
connection to most of the neighbours but a few far thrown ones to nodes at 
the far end of the network.  So...  If you do have a mud where npcs "know" of 
other characters, it would make sense for some of these npcs to "know" of 
characters far, far away.  This list, for example, is an example of the small-
world network.  I realise it's common sense but it made the papers only a few 
years ago!  There are other applications as well but I'll let you folks 
guess.  That is, if you find it useful...
  |   Ling Lo

MUD-Dev mailing list
MUD-Dev at kanga.nu

More information about the MUD-Dev mailing list