Can the AI take a hint?

[mrt1212]

In my latest adventure I've had a close rival declare war twice against me and both times give up a couple planets after their offensive war became a fiasco. Maybe the algo for declaring war should factor in catostrophic previous outcomes before being so eager to repeat the whole thing again.

To be fair I did set AI to aggressive but I didn't take that to mean that they'd haplessly throw themselves against the rock of my empire in spite of the history of me slapping them down mightily.

 

[Emraldis]

They'll definitely get you this time!

...

...

*smack*

 

[mrt1212]

It feels like a little brother starting some crap and then me taking over their room as a consequence.

 

[Emraldis]

It feels like a little brother starting some crap and then me taking over their room as a consequence.

hehe. They never learn do they?

 

[smjjames]

I once had a OPM (One Planet Minor) keep asking me repeatedly and repeatedly and repeatedly to start a war with a rival neighbor, only time it stopped was when they got pissed off enough to hate me. So, no, they can't 'get a clue' sometimes.

 

[Slynx]

i wish there would be some more diplomacy.
like "how much should i pay you so you'd stop protecting those fools, cuz i really like you and don't wanna kill you yet" (literaly happened to me in my last game. unfortunately i had to declare war on my only galaxy friend and cleanse all his planets(and purge his capital). that even made my fungoid race a bit sad (for a day or two)
or "can i pay you to pretend to be my friend?"
or "let me buy few planets from you"
and so on...

or at least the same thing AI can say to you "hey your outpost\colony is too near. can you please go away?"

 

[Deep Space Explorer]

I just want the ability to tell AI "If you attack me one more time i will purge your entire species".

 

[scaper12123]

The AI doesn't think quite how we do. Usually to us it's a matter of "Crush their fleet and we win." To the AI, it's a matter of "try literally everything we can and see what sticks." What the AI doesn't seem to realize is that, while they can get a strong fleet by doomstacking, you already have a significantly stronger fleet and also will doomstack. They also aren't too smart about being evasive and tricky. How can they after all, when they're just a little beep-boop soulless drone? So the end conclusion is the AI vastly overestimates its chances against you

 

[mrt1212]

I just want the ability to tell AI "If you attack me one more time i will purge your entire species".

The ultimate ultimatum! So in my current game we're on round 4 of offensive wars that will end disastrously and yield another 3 planets in my pocket. They're making my job too easy.

 

[Syr]

i wish there would be some more diplomacy.
like "how much should i pay you so you'd stop protecting those fools, cuz i really like you and don't wanna kill you yet" (literaly happened to me in my last game. unfortunately i had to declare war on my only galaxy friend and cleanse all his planets(and purge his capital). that even made my fungoid race a bit sad (for a day or two)
or "can i pay you to pretend to be my friend?"
or "let me buy few planets from you"
and so on...

or at least the same thing AI can say to you "hey your outpost\colony is too near. can you please go away?"

Exactly this!
This happens to me pretty much every game where someone I like and dont want to off ends up protecting someone that I hate. When its not in a federation I can at least usually find some way to drag only one into a war and not the other through other alliances they have made (ie, attacking someone i dont care about in order to actually attack someone entirely different) - when it is a federation.. well..

That also brings up the AI's issue of joining 'Loser Leages' - ie, federations that have been continually on the losing ends of wars and have really weak piddly members that can't stand up to aggressors (or even one of my one-system vassals), but haven't been completely eaten up yet due to the wargoals system. Joining up with them when their combined fleet power is less than your vassals' is stupid, yet the AI will do this if someone has been causing enough threat... (The threat system in and of itself also doesn't make that much sense either...) - It doesn't take into account existing friendliness, trust, etc. Also threat shouldn't be generated for defensive wars (unless you're cleansing planets).

If a federation isn't able to protect its members, they should lose faith in their federation after each defeat (and people shouldnt want to join it because the participants would have some sort of 'weak ally' modifier towards any other AI players. Eventually faith in that federation will be so low that it falls apart, and the members will likely have to become vassals if they want protection because nobody sees them as worthy equals. (Or you could maybe calculate weakness based on existing fleet + gross economy, but that doesnt provide an easy way to calculate loss of faith in a federation, unless you check against all the unfriendly/hostile/domineering/belligerent/angry empires (towards any member) powers as well)


...


Also an option to say 'Thats a pretty empire you have there, it would be a shame if something were to happen to it... how about we reach an amiable agreement instead?'
Being able to do a show of force would also be nice. The AI should take a good long look at a fleet thats (literally) over 100x bigger than its own and decide if it reaaaaaallly wants to risk angering said neighbor.
Me w/ 300k fleet parked in their sensor range, 1/3rd of the galaxy in my direct control, and 2/3rds in my combined federation + vassals (which include my pet 'liberated FEs', one of which is ranked 3rd in the galaxy due to a decent 20k fleet & tech), highest kill counts vs. both the unbidden and the prethoryn (I consoled in one when the other started for extra challenge), and the one to defeat both, killer of every guardian in the galaxy except the infinity sphere which didnt get killed, and a strong enough economy to continuously churn out enough ships in a month to defeat most other empires in the galaxy - even while my fleet is deployed: 'Hey, I demand you be my vassal'
Empire w/ 2.5k fleet, 3 planets, no allies, severe and chronic civil unrest after multiple defeats & humiliations & other issues w/ a neighbor I just ate up via liberate & vassalize, and a barely functioning economy: 'Nah'
... refused subjugation modifier should also scale based on strength too...
Sure, my next step is to simply 'liberate and vassalize', for they are but a bump in the road, but thats still painfully unrealistic behavior.

Fallen empires being condescending when I could completely steamroll them in a war also gets at me.. There should be points where they begin to.. at least respect your relative power, even if they dont actually respect you due to ethos. But since they dont respect me, I steamroll them, take their 100% habitability goodies, and liberate the remains into my pet vassals for research agreements. And they still do it after I've done the same to their peers (including any who tried to awaken but got steamrolled in the process anyways).

 

[Sheriff Godwin Law]

No, there is no way for an AI to receive your refusal and know it means "no never" instead of "No, not this instance." And there is no way for an AI to realize that calculating his war goal calculations around your total fleet strength versus his is woefully inaccurate to how the war will actually go.

I suppose they could add an extra "fear" modifier to the AI that applies to aggressive actions and refusing demands. Something like a -20 to it's ability to calculate whether it wants to go to war, and a +20 to whether it will accept demands from you, losing one point every year or so. But that seems like it would be exploitable.

 

[mrt1212]

No, there is no way for an AI to receive your refusal and know it means "no never" instead of "No, not this instance." And there is no way for an AI to realize that calculating his war goal calculations around your total fleet strength versus his is woefully inaccurate to how the war will actually go.

I suppose they could add an extra "fear" modifier to the AI that applies to aggressive actions and refusing demands. Something like a -20 to it's ability to calculate whether it wants to go to war, and a +20 to whether it will accept demands from you, losing one point every year or so. But that seems like it would be exploitable.

It doesn't even need to be that complex - there should just be a standard penalty toward the "should we fight these guys" check for each failed offensive war that results in planet loss. The way war demands work, they can pick a fight with me and I can pick off a chunk of empire over and over again. By the 3rd failed offensive war they should know better especially if the consequences have been shedding over half their empire to me.

 

[Sheriff Godwin Law]

Well, it does work like that in a roundabout way. You bite off a chunk of empire, they have less fleet support and so when they rebuild their fleet it is smaller than it was before. They calculate their chances of winning based on the relative size of their fleet compared to yours and somehow, after losing a sixth of their empire, they still think they can win, then a third, then apparently a half.

Part of that is that you are playing with aggressive AI, aggressive AI over value their desires and undervalue the negatives that would urge caution. You may actually be in a situation where the aggressive AI actually wants your territory more and more after each war precisely because it was taken from him. And if so, your problem may be resolvable by simply tweaking the AI values.

 

[mrt1212]

Well, it does work like that in a
roundabout way. You bite off a chunk of empire, they have less fleet support and so when they rebuild their fleet it is smaller than it was before. They calculate their chances of winning based on the relative size of their fleet compared to yours and somehow, after losing a sixth of their empire, they still think they can win, then a third, then apparently a half.

Part of that is that you are playing with aggressive AI, aggressive AI over value their desires and undervalue the negatives that would urge caution. You may actually be in a situation where the aggressive AI actually wants your territory more and more after each war precisely because it was taken from him. And if so, your problem may be resolvable by simply tweaking the AI values.

Will have to give it a shot. I enabled it because I wanted the AI to get at me and make life tough by forcing me to actually plan around a defensive war but it's like flinging a kindergartner at an adult.

Edit: I guess I didn't realize at the start of the game when I enabled aggressive AI that I could suplex the hell out of them with defensive wars.

 

[The Founder]

In my latest adventure I've had a close rival declare war twice against me and both times give up a couple planets after their offensive war became a fiasco. Maybe the algo for declaring war should factor in catostrophic previous outcomes before being so eager to repeat the whole thing again.

To be fair I did set AI to aggressive but I didn't take that to mean that they'd haplessly throw themselves against the rock of my empire in spite of the history of me slapping them down mightily.

Could you be a Militarist by chance?
I had two Wars by the AI declare on the Tzynn AI backfire horribly*. That +10% happinies while in defensive wars is a serious bonus to production capacity. Armsrace Edict and Weapon damage to thier part too.

*Once was a random empire. Then the Ix'Idar declared. The original empire white-peaced. Then the Ix'Idar got mostly conquered.

 

[mrt1212]

Could you be a Militarist by chance?
I had two Wars by the AI declare on the Tzynn AI backfire horribly*. That +10% happinies while in defensive wars is a serious bonus to production capacity. Armsrace Edict and Weapon damage to thier part too.

*Once was a random empire. Then the Ix'Idar declared. The original empire white-peaced. Then the Ix'Idar got mostly conquered.

Actually I am. I've taken a real liking to the human militarists after a few starts as the UN human. The gameplay really fits me like a glove and my role playing vision which is "humans uber alles, except those subjects that happen to live on planets I can't inhabit"

 

[The Founder]

Actually I am. I've taken a real liking to the human militarists after a few starts as the UN human. The gameplay really fits me like a glove and my role playing vision which is "humans uber alles, except those subjects that happen to live on planets I can't inhabit"

The AI seems to have real issues evaluating Militarists Strenght.
The +10% Happiness really buffs thier economy while defending. While the bonus weapon damage means every ship produced is worth more. And with Armsrace they got much cheaper port upgardes, wich churn out ships even faster.

I actually once thought about a approach to militarism where one would Rival a lot of people, then feign weakness by not building ships so the other side attacks. Only to be quickly out-produced and conquered.

 

[Slynx]

interesting idea..but how quickly you will be able to produce a fleet that will hold agains something like 10-50-100k fleet (or multiple attackers?)
also fleet capacity and tech level is already in strength comparison. and 2 out of 3 may still lead to equivalent strength result for the Ai

 

[The Founder]

interesting idea..but how quickly you will be able to produce a fleet that will hold agains something like 10-50-100k fleet (or multiple attackers?)
also fleet capacity and tech level is already in strength comparison. and 2 out of 3 may still lead to equivalent strength result for the Ai

Unless they have a Federation, they will not attack in multiples.
The AI hates to share thier spoils just as much as the player does. If they think they can defeat you on thier own, they will declare without allies.

Obviously it no longer works in lategame. More of a early game feint.

 

[Syr]

When the economy gets brought into play, the AI's threat evaluation breaks apart in general - doesn't even have to be the militarist specific bonuses.

I like to grab as much territory as possible early on, but using mostly outposts and strategically placed colonies to avoid causing research costs to skyrocket, and then proceed to econ-tech rush my way to being the most theoretically powerful. The AI in general will look at this sprawling, advanced, resource-rich empire but with a minimal fleet and think its easy pickings, only to get stomped into submission by suddenly mass-produced ships equipped with much high tier tech less than half a century into the game. They usually lose so much in wargoals that i wipe them out with a followup liberation or conquest on their remaining handful of worlds right after the truce expires.