Here is an idea I had for two games, let''s call them "SimAnimal" and "Predator": SimAnimal The player develops a fictional animal society. He starts off with a few relatively dumb creatures which he must teach/engineer/evolve to survive and thrive. He does not directly control the animals'' behavior; he just influences their AI. He is initially given a plot of land, which he can configure to some degree. Presumably, the animals would eventually learn to move around in groups/herds/packs. As the player''s society''s population grows, he is granted additional plots of land. The state of his society is periodically sent to a server maintained by me, the game developer. This server also tracks the allocation of land. Predator The player''s prey are the animal societies developed by SimAnimal players. The gameplay would go something like this: A society (along with its land) is downloaded from the server and simulated on the Predator player''s local machine. The Predator player controls a squad of human hunters armed with fictional weapons and traps, and he goes on a 5-minute killing spree. Afterwards, the entire episode is uploaded to the server, and the damage done to the society would be permanent. The affected SimAnimal player could download and watch a replay of the killing spree to see how his animals fared against the Predator player. Of course there would be some limits on how often a SimAnimal player''s society could be downloaded and hunted. That''s about it. The two games would use the same engine and art, but SimAnimal would be completely free -- after all, the SimAnimal players are really just providing content for the Predator players. Any thoughts, ideas, suggestions?

Intriguing.

Judging by the info you provided, it looks as though it''s already more effort to raise and develop the animals than to be the predators. I don''t think I''d take the time to raise a good stable if it could be wiped out in minutes by an inexperienced predator team. I want to know what kind of chance my animals would stand against the hunters.

I dig your idea. I see something here very Rampart-ish, but that might not be what you had in mind.

-The Well-Suited Punk-

If the SimAnimals that are not easy to kill like:
- large animals like dinosaurs, elaphants, etc, or
- dangerous animals like lions, tigers, bears, sabertooths tigers, etc,
- something small like a swarm of mosquitoes, a pack of rats, etc
- live in a completly different enviroment, ocean, plains, etc.

and the preaditors don''t know what they are going after, this could be a fun game. Then you could have a diverse team that would not be able to do as much dammage in every enviroment, or you could have a specialized team (large game hunters with elaphant guns) and hope you get an area with large animals. As long as there is some ability for the sim animals to either fight back, defend or flee this would probably make a decent game.

Kars

Neat idea. Kind of like how Silent Hunter 2 and Destroyer Command dovetail with one another. Although in that example, I''m not sure who''s the predator and who''s the victim.

_________________________
The Idea Foundry

i really like the idea, i''ve never seen something like this. surely, give some thought and beingh well planned out, this could actually get somewhere.

---
shurcool
my project

This is one of the most interesting ideas I''ve seen here.

In order for it to work, I believe a few things are required:
1 - Animals must be able to conclusively "defeat" a predator
2 - There must be something to gain by defeating predators
3 - AI design system must be powerful yet accessible

One may "defeat" a predator by evading it for a given amount of time. Or maybe the animal will be able to fight back against the predator. Whataver you do, make sure there are many ways to "defeat" a predator. This "animal victory" is essential, because otherwise there is no real goal to the game; there is no competition.

In order to foster competition, predators and prey may be ranked. Prey may ally with one another to defeat strong predators (and visa versa), but at the same time both animals and hunters are competing with one another for higher rankings on their respective lists. I see this as a highly competitive online game.

Now, about AI design: It must be possible for the average gamer to build decent animals. However, advanced gamer/coders should have the ability to put their skills to work. One way to do this would be to provide a user-friendly graphical interface to a much more powerful underlying script engine directly accessible to those who want to harness it. The script language may even be a slightly-modified C, C++, or Java, run through a VM (as with Quake III mods). In such a situation, I could see the game becoming a useful testing ground for AI researchers.

Beyond AI, the physical bodies of animals would have to be designed as well, by connecting many premade modular parts. You can probably get more creative if you change the "animals" to robots. Then, "defeating" a predator could give the player cash to buy more parts!

Tell me what you think of these ideas.

TerranFury mentioned the importance of fostering competition, and the necessity of there being a way for the prey to defeat the predators. The only thing in his post I really am not in tune with is the idea of buying parts. I think there are two derivations of that idea that are more consistent with the original premise (one more so than the other).

The first idea is selective breeding. The SimAnimal player is given random species with different strengths/weaknesses and biological compatibilities. The player then selects a prime specimen and crossbreeds with another prime animal, hopefully inheriting both animals'' strengths and none of their weaknesses (ha! I finally found a non-AI use for GAs! Using them we could have a complete crossbread and mutation model). Hopefully over time, the player can even develop species sufficiently similar to crossbreed to previously incompatible species (A and B are compatible as are B and C, but A and C are incompatible; hopefully the offspring AB is at least partly compatible with C).

The second idea is genetic engineering. In this scenario the player would purchase (or re eive as bonuses, or retrieve from the site of the carnage) genetic material which can be stripped for DNA strains which can then be spliced into the next generation (Superman''s Doomsday, anyone?)

Throw in prestige so that there''s competitin between predators and incentive for them to attempt to tackle fearsome species.

I feel tingly all over.

Thanks for your feedback, everyone. Here are a few more thoughts, along with some questions for you guys:

First, regarding animal AI and breeding:
- SimAnimal should appeal to casual and hardcore gamers.
- For casual gamers, I like Oluseyi''s first suggestion. Players simply watch their animals for a while, then select two "prime specimens" to breed. Perhaps a player replays a recent Predator attack to see which animals were most alert, most cautious, or most aggressive. The offspring inherit physical and behavioral attributes, but there is some randomness and the occasional mutation, too.
- The hardcore gamer should be able to write AI scripts, as TerranFury mentioned. The player''s script may work in conjunction with triggers which he''s placed in the terrain (danger spots, escape routes, etc.). The question is, how would an animal''s scripted behavior mix with its inherited behavior? Should a player be able to program his animals to attack hunters, even if he hasn''t bred them to be aggressive/brave?
- As you say, TerranFury, all of this could be done with robots, which would certainly give me more latitude. I like the idea, but my main objection is that robots are not "cute" . I think that animals have a more universal appeal, and animal breeding is something that people can instantly understand.

Global stats/ranking:
- This is a great idea, and it would be easy to implement through the server.
- SimAnimal societies are ranked by population size and lethality (number of hunter kills).
- Predator players are ranked by kills.
- Predator kills are weighted by the lethality of the animal; this should make "newbie" societies less attractive to Predators, which is good.
- Top-ranking SimAnimal societies will inevitably be the most attractive to Predators, which is good.
- There is no similar built-in balancing for Top-ranking Predators, though, which is bad. What could be done about this?
- TerranFury mentioned alliances... any thoughts on this?

Ok, one last idea:
- Rather than killing specimens, the Predators use non-lethal traps and tranquilizers to catch them.
- Predators could sell specimens back to their owner, or to the highest bidder. A SimAnimal player might pay a large sum for a prime specimen. Money would be another stat to rank, I suppose.
- This would lessen the damage a Predator player could do to a SimAnimal player''s society; if the SimAnimal player lost a "key" specimen, he could just buy it back.
- SimAnimal players could sell their animals, too.
- Would this complicate things too much? I wouldn''t want a SimAnimal player to have to worry about money on top of everything else, unless he''s interested in that.
- I''m tempted at this point to just merge the two games... everyone raises a society, and everyone hunts in order to obtain new specimens.

First the economy aspect: Predators get money from breeders. How do breeders get money? We would make the system symmetrical if breeders got their money from predators. One solution is simple: Breeders keep the weapons left behind by defeated predators. Predators buy their weapons from breeders. So that predators don't put everything on the line when they go out hunting, they can perhaps have a small "weapons locker" where they can safely store a few guns. Now we have a global weapons market and a global species market.

These economic relationships could lead to interesting alliances between hunter and breeder :
Breeder: I'll give you $1,000 if you can capture me a Great Plated Mastadon.
Hunter: I won't be able to do it with this Daisy BB-gun.
Breeder: You can borrow this übergun I picked off the guy who didn't see my Dark Panther in the shadows.
Hunter: Nice, that'll do the trick. You've got a deal.



Next, the physical representation of "prey:" I agree that animals are preferable to robots. From a programmer's perspective, however, how do you represent an animal? What are its fundamental building blocks? I mentioned robots because there can simply be a heirarchy of parts and connectors. But what about animals? We can get a wide variety of behaviors from different genotypes, but how do we get a wide variety of bodies from different genotypes? Right now I'm thinking Wolfram-style about cellular automata and metaballs, but it seems kind of far-fetched. This is the problem that I'm trying to think of a solution to now, and I'd like to know what your thoughts are on the subject.

Now, about AI scripts and "nature versus nurture." I think it would be fair to allow advanced users to completely override the behaviors defined in the genotype with a script. The genotype still defines the physical characteristics, which put a practical limit on what the AI can decide. However, I would prefer a more elegant solution, and right now I can't think of one.

About alliances: I don't think an "alliance" should be anything official. I do think, however, that partnerships - both temporary and long term - would evolve naturally from such an MMO game world.

[edited by - TerranFury on July 3, 2002 2:23:21 PM]

In my last post, I mentioned what I called "symmetry" in the economy. Now here''s something else to make the game even more symmetrical: Much as breeders develop better species, hunters design better weapons! Weapon design is easy: As with robots, you use a heirarchy of parts.

The main problem we''re left with is on the breeder''s side: How do we use a genotype to define an animal''s body? How can we have a single genetic system describe either a blue whale or an antelope or a bird of prey?