[MUD-Dev] RE: Realistic Ecological Models

Sasha Hart Sasha.Hart at directory.reed.edu
Mon Apr 29 18:52:30 New Zealand Standard Time 2002


[David Held]

> Why should I simulate food shortages?  If everything needs to eat,
> either there will be a food shortage or there won't be.

If this occurs in your game, then you are implicitly simulating food
shortages, aren't you (even if you aren't explicitly aiming at it)?

What I would say (with the utmost cowardice and willingness to
repent if an effective argument is presented) is that most of the
factors I discussed were sufficiently general and robust to apply to
vaguely real-like populations simulated at many levels - from "i
have a number for each population's size only" to "i have a thousand
autonomous neural-net driven antelopes spinning around a voxel
landscape."

How well they match with your model depends at least in part on how
lifelike your ecology is. It might be more correct (if deflationary)
to say that those observations on ecology are just observations on
the math which typically results when people try to duplicate
population size fluctuations.

>>  -- Dealing with this problem may entail speciation or other
>>  abstractions on genetics or at least variation in phenotype
 
> I think this is a separate aspect that should be studied in its
> own right.

I think you are right. But I am afraid that this kind of thing is
necessary in instances where it is not specifically of interest.
Fortunately I only have an awful hunch that this is the case. The
hunch, as I mentioned, comes in part from the problem of needing a
large amount of diversity - but also from the observation that
because natural selection is an engine for producing stable
populations, that an ecosystem without it might have to settle for
second-class hacks, interventions, etc. to be run properly. I hope
not, even if it would be no catastrophe if this were the case.

[Mike Shaver]

> Have you collected a bibliography on this sort of ecological
> modelling?  I hate to reinvent the wheel. =)

hehe. The "reinventing the wheel" comment applies painfully well to
me, as I have spent months in the past making important discoveries
on the order of "the world is not flat" after writing fairly large
amounts of code.

I haven't scraped together a bibliography, but it is a great
idea. In the meantime, if you (or anyone else) wants a start on what
to look at, send me a note describing what you are doing or looking
for, and if I have read anything that bears on it I will pass it
along.

You can get a lot of mileage out of otherwise ridiculously naive web
searches. There are a million biology students taking this more
seriously than I possibly could, even though it's a big part of my
hobby.

[Lee]

> This can be deceptively hard.  Since mobs will be prey, and your
> player base are the predators, you need an accurate way to model
> the changes in your player density, or a good system for changing
> the birth rate/death rate based on it.
 
> Say, for example, your game server is knocked off the net for an
> hour.  Not actually crashed, but knocked off-line.  With a fixed
> birth rate, mobs will consistently outnumber players for a good
> period of time after they can log back in.
 
> Say you make birth rate proportional to number of players, but
> leave death rate alone.  If you have any group of players that
> logs on at relatively the same time, but only stays connected for
> a short period, then you have another spike in mob population, and
> nothing to put it back down.

You are completely right that it could be hard, particularly in the
situation you give (player base as predators, moreover as very
substantial predators).

First, I have a concern that this system will keep up with demand in
the way you seem to be asking for (I'm not sure). Just because a
player is on doesn't mean he's hunting - so what assumption do you
make, that he'll hunt or he won't? Is player hunting rate that
dependable? Probably depending on the game. I can imagine collecting
kill rate stats and using these to make even finer-grained
adjustments.

Second, the problem you describe should be roughly as bad as how
much stock you put in player predation (through your design) as
something to depend on. Personally, I would prefer not to depend on
players killing OR not killing things - and use other kinds of
population control (non-player predators, disease, carrying
capacity) to regulate population levels which are critical to keep
down.  If nothing else it is possible to put a simple cap on
population.

Still, it is a troublesome problem, because we want players to
interact meaningfully with the ecology. My tentative answer to this
is A) that players can interact in ways which may have zero effect
on population sizes, or B) that they can interact selectively - as
"tiebreakers" or just another species in a largish ecology.  I think
both are still acceptable uses of the ecology, but they do require
some careful design... perhaps too much, but that's yet to be proven
:)

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



More information about the MUD-Dev mailing list