Re: [glob2-devel] Creature AI

[Gabriel Walt]

Sorry for the late answer, I wrote that email 2 days ago, but I had some 
problems with Savannah...

----------  Forwarded Message  ----------

Subject: Re: [glob2-devel] Creature AI
Date: Friday 20 May 2005 12:30
From: Gabriel Walt <address@hidden>
To: address@hidden

Hello,

Wow what a mail! I like your idea a lot! It would make glob2 a quite different 
game... But for me it's totally ok if we consider this as glob3! It's also 
not a big problem to me if your idea takes some time to be done: Globulation 
is the result of 5 years of work for Nuage and Nct at part time. :)

By the way have a look at that, it might interest you (Nct sent it on that
mailing list two weeks ago): http://dev.ic2.org/nero_public/

As Martin I'm a bit afraid about the boot strap problem: Units must already
have reached quite a good level to simply survive. They must have learned to
pick up food, to bring that food to the inn and to finally go to the inn when
hungry. But worse: Starting globules must imperatively know how to bring
resources to the inn to be able to produce the next generation of globules!
Maybe globules should start with a pre-recorded intelligence that allows them
to at least survive and produce new globules.

As a simple glob2 player I want some gameplay. ;) For instance I'm a bit
afraid about how long it will take for the colony to evolve. A game should
last about 1 or 2 hours on average. If it's much more than we should think
about a way to make a game playable in this amount of time. For instance I
could imagine that a player could breed his race, save it, and then play with
it against another player... That's just an idea. But the beginning of a game
shouldn't be made too annoying if you wait a long time until a few globules
are somehow able to do what you want them to do.

To add some gameplay, I suggest that it's the player that chooses what
globules are chosen for replication and crossing-over when the swarm produces
a new unit. The player should have some kind of breeding laboratory, which
would be a stat screen, where he can see the summary of the scores of all
globules and can choose the specimens that will be selected for reproduction.
Maybe this might also be the main occupation of the player at the beginning
of the game, while globules don't work efficiently enough yet.

But I like your idea a lot, and hope that it is doable, and I wish you the
energy to do it. I like a the idea that a globule could be able to walk, and
suddenly fly away for instance because there was an obstacle, then land,
gather some resource, fly away again, and land on a nearby island and plant
the resource there - for instance because the player has set on that island
an area that rewards units planting something there. I really hope that a
genetic algorithm can reach such a level of complexity.

> Note: walking into a building wouldn't cause a unit to heal/lunch/etc..
> For example, a unit could simply walk through a friendly building rather
> than around it.

I'm not sure about that, but it's a detail. I think it's visually much clearer 
if a unit that goes into a building gets upgraded by that building. It's 
already a complex game, this would imho add more confusion.

> I would like some system of global incentives - so if you want, you
> can bump the "aggression" value up, rewarding your warriors more for
> random acts of violence, and watch them go on the rampage.  Then
> later, put it back down again so they have a chance to recover and
> train.

Yes this would be cool too!

> What other new features could I crowbar in?
> What features should I take out?

As usual: Start easy, take a subset of your ideas and test if it works. :)

> Do any of you have GA experience you can share with me?

Nct is doing his PhD on GA and stuff like that, he know a lot about it.

Gabriel

-------------------------------------------------------