[MUD-Dev] Game Economies

[J C Lawrence]

On Wed, 9 Jun 1999 21:43:59 -0700 (PDT) 
Matthew Mihaly <diablo at best.com> wrote:

> Defining what the outcome is going to be is sort of the opposite
> of this sort of thing. 

I should note that I'm not really a purist here.  I set up some
systems and then fiddle with the feedback rates until the resultant
behaviour fits as something I think is "interesting".  What's
important is that I don't really predefine the objective before I
start.  I start with what I think is a neat system, implement it,
observe the behaviours, and then pick one or more than I think is
"neat" and try and "bring it out" by tweaking the implementation and
surrounding systems.

To take the Trask Collector example I've discussed here previously
the goal was fairly simple:

  Trash Collectors would have a significantly oscillating
populations.  The period, rate, and amplitude of oscillation would
apparently vary randomly over time, but be subject to short range
accurate prediction and manipulation.

ie I wanted lemming-like population explosion/collapse mechanics
where the base population pattern was a fairly regularly wave whose
amplitude seeming randomly blew off the scale in one direction or
the other.  

Getting that was another matter.  I installed a variety of feedback
loops which I thought would create the required instability and
oscillating patterns and succeeded beyond my requirements.  The
initial population and oscillation rate was fine, but then I hit an
unconstrained population explosion, and got several million TC's in
the world.  They soon all died of starvation resulting in several
million more TC spores, and another semi-stable population
oscillation at a new higher level population count (based on the
fact that TC corpses are edible), mass starvation, and a further
repeat at a yeat higher level.  ie I got a stair stepping
progression where the graph wobbled about a bit, exploded,
collapsed, wobbled about a bit at a new higher level, exploded,
collapsed, wobbled about a bit at a yet new higher level, exploded,
collapsed, etc.

Not what I wanted.  Every percentage feedback tweak or intersection
with an other in-game feedback system either ran towards TC
exctinction, on further unconstrained growth.  Neither is of course
acceptable.  I spent a *LONG* time diddling values and adding more
tweaks and twiddles to try and smooth out the lmit points, all
without good effect.  <ugliness> 

I'm not pleased, but I also haven't given up even if I have to
artificially constrain the population model.  I'm toying now with
the idea of TC predators who have two survival modes, one based on
reasonable TC populations, and another massively TC consumptive one
based on run-away TC or TC-spore populations.  So far all the
attempts have headed towards TC extinction, or long-term apparent
extinction with periodic, rapid (eg over about an hour and a half
RL), uncontrollable/manipulable, and highly unpredictable massive
population explosions (ie the normal population is roughly one TC or
TC-spore per 100 locations, and then every month or three it
explodes over a period of 90 minutes to around 4,000 TC's per
location, and then collapses back to the prior state within the next
15 minutes).

Some more smoothing to do.

> What JC is doing sounds really interesting, though I have no idea
> how much fun it is for the players. 

Past a certain point players become volunteer (even paying) torture
victims.  It helps of course that I'm not really in this for the
players but the intellectual delight of it.

> I doubt I will ever take the risk of putting a system in with the
> intention of allowing emergent properties, simply because they
> could be bad for business.  

Oddly enough that would be precisely why I would use them, but then
I guess that explains what I do for a living.

--
J C Lawrence                                   Home: claw at kanga.nu
---------(*)                Linux/IA64 - Work: claw at varesearch.com
 ... Beware of cromagnons wearing chewing gum and palm pilots ...


_______________________________________________
MUD-Dev maillist  -  MUD-Dev at kanga.nu
http://www.kanga.nu/lists/listinfo/mud-dev