Large scale scenarios against AI not recommended?

Discuss and post your mods and scenarios here for others to download.

Moderator: Vic

Post Reply
GaryChildress
Posts: 6761
Joined: Sun Jul 17, 2005 3:41 pm
Location: The Divided Nations of Earth

Large scale scenarios against AI not recommended?

Post by GaryChildress »

It appears that most of the grand strategy scale scenarios (ones with the entire European theater in them) are mostly Human (multiplayer) only scenarios. Is it not recommended to do a grand strategy scenario to play against the AI? I would like to do a grand strategy version WWII scenario of ATG but I see that most of them don't incorporate the AI.
User avatar
Ormand
Posts: 754
Joined: Tue Feb 17, 2009 1:31 am

RE: Large scale scenarios against AI not recommended?

Post by Ormand »

To tell the truth, I am not sure if it is a good or bad idea. I am going to guess that for historical scenarios the problem is that the AI will need to be controlled through events. For example, a WWII scenario going from 1936 to the end. Controlling a German AI to go for Anschluss, Sudetenland, a full invasion of Czechoslovakia, attacking Denmark and Norway, etc has to be done through events, which is tricky to get the AI to respond correctly. You can get the declaration of war alright, but getting the AI to focus and act is harder. This can be done through setting some RuleVars for AI behavior, and also making he AI value areas with virtual victory points. It can probably be done, but it will take work. I have started the core of more general politically based random game that has most of the pieces in place for human players, but have not bitten the bullet to make events for AI regimes. This will probably not be too hard since it meant for random games, thus there aren't expectations that certain events will happen and react accordingly.

Random games are more or less a free-for-all, so you start it, and just go. But, for historical games, people will expect a fairly reasonably accurate response from the AI, which I think will be difficult to model, and thus lead to some disappointment. That said, I do think that it can be done, but it will take work. The other thing is the length each turn is. As I discovered with the GoT idea, 14 regimes takes quite some time to finish each turn.
One man alone can be pretty dumb sometimes, but for real bona fide stupidity, there ain't nothin' can beat teamwork -- Edward Abbey
User avatar
Tac2i
Posts: 2078
Joined: Tue Apr 12, 2005 5:57 pm
Location: WV USA

RE: Large scale scenarios against AI not recommended?

Post by Tac2i »

Perhaps in 2015 or 2016 we will see Vic find the time to create Advanced Tactics Platinum. He has two games under construction at this time: Decisive Campaigns 3 and Shadow Empire. I know Vic looks to improve the AI in AT Platinum.
Tac2i (formerly webizen)
GrumpyMel
Posts: 864
Joined: Fri Dec 28, 2007 8:37 pm

RE: Large scale scenarios against AI not recommended?

Post by GrumpyMel »

Hey Gary,

Basicaly what Ormand said. The AI is pretty decent at handling random games and even some of the set piece scenarios with standard ATG units and rules, etc (it's still nowhere near as intelligent as a human opponent which is why it's generaly given a boost in order to make a decent game). However, if you look at alot of the larger WW2 scenario's, like my ETO (which was actualy made for AT and ported into ATG), the designers start straying away from standard ATG with custom SFT's custom terrains, different tech's, political events, pre-programed events, etc. The further away you stray from standard vanilla ATG rules, events and dynamics the more the AI struggles with handling those in a semi-sensible fashion to the point where it becomes sort of dysfunctional.

It's why many of us who like to do those sort of custom, non-standard things in our scenarios don't even bother to attempt to make the A.I. to work with our scenarios of that nature.

I think you could do a reasonable job of getting the A.I. to play as a "scripted opponent" in a WWII Grand Strategy game if you stick very close to the standard ATG rules, SFT's and Terrain, etc but you would need to do alot of pre-scripted events (as well as playing with VP's and priorties) to force the A.I. down a specific path that would sorta make sense for the scenario. However, that would sort of take the essence of the "Grand Strategy" out of the Grand Strategy game and it would somewhat kill the replay value. I think what you would have a really difficult time with is to get the A.I. to "think" and react dynamicaly to the different things the player was doing and the different decisions he was making on a Grand Strategy Level and make a decent challenge out of it. Honestly, I never even thought of trying it with any of my scenarios... there was more then enough work in getting them to play well with only human opponents. YMMV.



User avatar
cpdeyoung
Posts: 5378
Joined: Tue Jul 17, 2007 3:26 pm
Location: South Carolina, USA

RE: Large scale scenarios against AI not recommended?

Post by cpdeyoung »

Playing GD1938 (the Global Domination 1938 scenario) as the Axis against the computer controlled Allies is quite fun. You must help the AI out with declarations of war. I usually throw a few coins, perhaps starting with four, and if they come up heads I declare war on (for) the Soviets. If the Axis are declaring war because they want to then the DoW comes at the beginning of their turns, if the DoW is dictated by the coin flip it comes after the end of the Axis turns.

If you feel there is a better chance for Stalin to attack remove a coin requiring less heads, and so on. The same procedure applies to USA entry.

Playing the Allies against a computer Axis is less fun as the AI is passive diplomatically. This could be handled by events.

Chuck
User avatar
Jeffrey H.
Posts: 3154
Joined: Fri Apr 13, 2007 6:39 pm
Location: San Diego, Ca.

RE: Large scale scenarios against AI not recommended?

Post by Jeffrey H. »

ORIGINAL: Gary Childress

It appears that most of the grand strategy scale scenarios (ones with the entire European theater in them) are mostly Human (multiplayer) only scenarios. Is it not recommended to do a grand strategy scenario to play against the AI? I would like to do a grand strategy version WWII scenario of ATG but I see that most of them don't incorporate the AI.

It used to be that the "World at War" scenario by Tom Weber was AI capable and people did play it that way, but that was a long time ago and not particularly rewarding IMO for the human.

I have myself played a "Global Domination" AI capable scenario that eventually blew itself up, (no longer ran) due to too many nations and ram limitations. The scenario was revised with only 5 or so nations with build capablity and I have yet to give that one a go. The other one was fun, but turns took like 30 minutes to complete but it was really cool in it's own way watching the AI do it's thing, declaring war randomnly on each other.

History began July 4th, 1776. Anything before that was a mistake.

Ron Swanson
User avatar
cpdeyoung
Posts: 5378
Joined: Tue Jul 17, 2007 3:26 pm
Location: South Carolina, USA

RE: Large scale scenarios against AI not recommended?

Post by cpdeyoung »

@Jeffrey H.

I would like to give that a try. I do not expect perfection, and I will give feedback.

If it is just not ready for "release" I'll understand.

Thanks,

Chuck
User avatar
Jeffrey H.
Posts: 3154
Joined: Fri Apr 13, 2007 6:39 pm
Location: San Diego, Ca.

RE: Large scale scenarios against AI not recommended?

Post by Jeffrey H. »

ORIGINAL: cpdeyoung

@Jeffrey H.

I would like to give that a try. I do not expect perfection, and I will give feedback.

If it is just not ready for "release" I'll understand.

Thanks,

Chuck

It's a Bombur mod, in the scenario bank. I forget exactly what it's called. Fish around in there and I'm sure you'll turn it up.
History began July 4th, 1776. Anything before that was a mistake.

Ron Swanson
Post Reply

Return to “Mods and Scenarios”