Ingrate!

[HappyBuilder]

Playing a game on Emperor level as Rome...

Persia, which is at the antipode of the planet from me, is having a tough time. Russia sacked Pasagardae before most civs had settled a 3rd city and Greece has been beseiging them for much of the game. Interested in preventing either of those civs from gaining a second capital, I become Persia's benefactor. I gift them nearby units produced by friendly city-states; I provide them with horses and iron free of charge so they can defend themselves. We have a declaration of friendship and I give them everything they ask.

Then the effers denounce me.

A turn later Greece asks me if I'd like to destroy Persia and I give them the greenlight, DOWing them, thus stripping them of all of their strategic resources. Why did Persia betray me? Allegedly they feared I was a warmongering menace to the world. Even if I was (a debatable proposition), denouncing me was suicide for them. I don't care what other factors went into the decision, "Losing their support will mean my immediate death" should have been the overriding factor.

HB

 

[wcbarney]

I think too many of us assign human-like reasoning power to the AI civs. All they are -- bottom line -- are a few thousand lines of computer code.

 

[HappyBuilder]

I think too many of us assign human-like reasoning power to the AI civs. All they are -- bottom line -- are a few thousand lines of computer code.

True enough, but they can be programmed to prioritize. Survival ought to be among the highest priorities.

HB

 

[Atwork]

Same experience here.

I propped up the Aztecs for a thousand years. I have 28 cities -- they have 4. We declared our friendship, I denounced everyone that denounced them...whatever they wanted or needed, I provided. I built them an army. I was providing them enough gold per turn so they could run a deficit. I gave them strategic and luxury resources at no cost. I gave them open borders so that they could launch attacks from my lands. We've signed several RAs. Etc., etc.

Then, just as I was about to join them in a war against a common enemy, suddenly, they declare a suicide war against me. With their pikeman, they attack my musketman who were fortified on hills inside of forts. My ally city-state proceeds to decimate their lines from behind. Within one turn, they lost 75% of their military (half of which I provided). Sigh. A few turns later they beg for peace and I am merciful. Now I have set about builiding them back up.

The game logic really needs to be tweeked to not allow this sort of thing to happen. A dependent state should never declare war against its supporter IF there is no advantage that can possibly be gained. Here, the chances of the Aztecs gaining even a single military victory was nearly 0%.

The game logic should have triggered the Aztecs asking me into war against our common enemy, especially considering they were already using my lands to fight that other war. Instead, they ended up being surrounded on all sides by enemies and being involved in an impossible war.

When the Aztecs declared war, I think their leader said something like, "I can no longer help but to demonstrate my deceptive ways," or something like that. Yeah, well freakin brilliant. Should have said, "I can no longer help but to demonstrate my suicidal tendencies."

Diplomacy and AI strategic relations has to be the main focus to really get this game to the next level. It is slowly improving, but it has a very very long way to go still.

 

[chazzycat]

I like that the AI is willing to backstab, but yeah, it doesn't always make sense. Sometimes it seems like they are doing it just for the sake of back-stabbing (i.e. they have a low "loyalty" value). That shouldn't really be the case...ideally the AI should consider:

1) what are the goals of the backstab? Generally, this should include 1 or more city to capture.
2) is there any chance of success in attaining these goals, based on strength levels?

only THEN should the "loyalty" factor come into play. so if a non-loyal civ thinks they have a decent shot at meeting their goal, they should go for it. But a more loyal civ would have to be totally sure in the outcome in order to backstab.

Maybe thats the case already, I'm not sure, but this does seem like something that can be improved with some adjusting somewhere.

Another way to handle this situation would be to have some kind of diplomatic modifier recognizing a benevolent dependency, but that would be tricky.

 

[CoolLizy]

As I pointed out in another thread, I've gotten to the point where I trust "friendly" civs far less than "guarded" and sometimes even "hostile" ones. The AI seems to have chronic backstabing disorder, and it gets a bit tiring watching them put "friendly" up thinking they're being witty and fooling me. My past few games not one civ attacked me without becoming "friendly" with me first, and not one civ who acted as my ally downgraded to "guarded" or "hostile" first as relations began to sour. -_-

 

[Depravo]

I think too many of us assign human-like reasoning power to the AI civs. All they are -- bottom line -- are a few thousand lines of computer code.

A few thousand? More like

10 DERP
20 GOTO 10

 

[CoolLizy]

A few thousand? More like

10 DERP
20 GOTO 10

Except that Civ is programmed in C++, so it'd be more like...

AI_Player::takeTurn() {
   while(true) {
      Derp();
   }
   return;
}

 

[LegionSteve]

A few thousand? More like

10 DERP
20 GOTO 10

very good

 

[MisterBoomBoom]

My belief is that for as many "backstabbing factors" we have in the game there should also be a distinctive "loyalty" factor....which would mean for every Insane Monty or backstabber in the game should have a balancing Civ that is ultra loyal to the end.

Also I can help but notice many players hand over military units to other civs. I simply can't. They totally are inept in their use so why bother?

 

[chazzycat]

there is actually such a "loyalty" factor. I have noticed the Koreans & Monghols seem to be quite loyal. I don't like looking at the cheat sheet, that's just from experience...

 

[Smokeybear]

A turn later Greece asks me if I'd like to destroy Persia and I give them the greenlight, DOWing them, thus stripping them of all of their strategic resources. Why did Persia betray me? Allegedly they feared I was a warmongering menace to the world. Even if I was (a debatable proposition), denouncing me was suicide for them. I don't care what other factors went into the decision, "Losing their support will mean my immediate death" should have been the overriding factor.

The overriding factor seems to be their lemming-like programming requiring each and every civ, no matter how puny or irrelevant, to HAVE TO TRY TO WIN and beat the human player. Therefore, despite all common sense and logic to the contrary, they had no (programmed) choice but to launch a desperation war, even one with no hope. There is no such thing as a friend, in the game. Building up another civ in hopes they will somehow aid you in any significant way, is foolhardy. All you are doing is just building up another future enemy. Always.

 

[AxelRhodd]

I think that the AI should be able to remain loyal and submissive ally to much more powerful human player WITHOUT compromising the ultimate objective of BEATING the human.

The AI should use all the help it is getting from human ally to build itself up in attempt to backstab the humane player later on.

Consider Atwork's situation with Aztecs. Instead of getting all suicidal, the Aztec AI should realise that:

Human much more powerful - do not provoke
Human gives help - take help and remain loyal UNTIL strong enough to challenge the human.

Basically, the weak AI should try to use the help to build itself up.
It should also try and bribe other strong Civs to attack the human in attempt to weaken it.

 

[Martin Alvito]

I'd like to tell you that Persia was motivated by the famous dictum that (to paraphrase), "Nations don't have friends, they have interests." But the behavior belies that, and the AI does dumb stuff like that all the time.

Closer to the CiV AI's behavior is Stalin's famous quote: "Gratitude is a disease of dogs."

But at the end of the day, it's an AI that operates from a random number generator. Any AI with that characteristic is going to put it all on red and spin the wheel some of the time. The problem with the CiV AI is that it seems to do this much more often than other, similar AIs...even within the same series.

 

[Atwork]

Also I can['t] help but notice many players hand over military units to other civs. I simply can't. They totally are inept in their use so why bother?

To bump up their military rating so that other civs will be less likely to declare war or bully them. Therefore, hopefully I'm not forced to take sides quite as often.

And to serve as a bumper state so that I have the option of remaining neutral if a war does break out. Afterall, if it's AI versus AI, then the ineptitude is equal on both sides. Numbers should prevail.

Basically, the weak AI should try to use the help to build itself up.
It should also try and bribe other strong Civs to attack the human in attempt to weaken it.

In the case of my game, the best way that the AI could have used my help was to focus its attention on our common enemy, who was about the same size as the Aztecs. Had the Aztecs prevailed in that war, then it would have doubled its lands. Another conquest like that and suddenly the Aztecs might be contenders. But, the AI unfortunately doesn't behave so strategically. At least, not yet.

In its calculations, the AI should have the ultimate goal of, first, winning the game and beating the human second. But, strategically, the AI should focus on achievable results that, if achieved, will help the AI accomplishing its ultimate goals of (1) and (2).

The AI should recognize that a suicidal war against the human will not help it to achieve its ultimate goals, even if launched 'unexpectedly.' In my game's case, the AI should have recognized that focusing on a lesser target would, in the end, be more likely to help it achieve the ultimate goal of (1) and (2). By attacking me rather than the lesser enemy, the Aztecs ignored an achievable strategy that might have helped it achieve goals (1) and (2). Also, by ignoring the achievable strategy, it instead chose an impossible strategy, which significantly lessened the chances of it accomplishing (1) and (2). Even worse, by choosing an impossible strategy that could only fail, the Aztecs became vulnerable to all other civs, whereas previously, the Aztecs were fairly strong in relation to all other civs excluding me.

This is a significant failure of programming IMO. Predictable backstabbing by the AI eventually ceases to be unpredictable. Moreover, when the backstabbing is both predictable & bad strategy, then the programming has failed its purpose.

In short, the AI should focus on achievable victories and surprise attacks should be nothing more than a tool toward achieving victories. If the AI must focus its attention away from the human in order to obtain an achievable victory, which in turn will provide greater long-term leverage against the human, then the AI should choose that strategy. If attacking the human will not advance the AI's interest in winning the game, then the AI should not attack unless it is a part of a larger coordinated AI strategy to prevent the human from winning. If attacking the human does not at least serve a part of a larger coordinated AI strategy to prevent the human from winning, then the attack serves no purpose other than to increase the likelihood that the human will come away stronger. And that is just really bad AI strategy!

 

[Depravo]

This is a significant failure of programming IMO. Predictable backstabbing by the AI eventually ceases to be unpredictable. Moreover, when the backstabbing is both predictable & bad strategy, then the programming has failed its purpose.

QFT. Absolutely. I don't think it even qualifies as a backstab if the odds are so long against them. More of a bottom pinch.

 

[AxelRhodd]

I have just realized something: All AI Civs act together as one entity simply trying to produce the strongest AI opponent to the human player.
I suspect that the AI code does not treat the Civs as independent entities trying to win the game. I think the AI would be very different otherwise and allow for meaningful diplomatic relations and more rational but not necessarily predictable behaviour.

I dont even know what Im trying to say here lol... just thinking aloud...

 

[qalsip1977]

I have just realized something: All AI Civs act together as one entity simply trying to produce the strongest AI opponent to the human player.
I suspect that the AI code does not treat the Civs as independent entities trying to win the game. I think the AI would be very different otherwise and allow for meaningful diplomatic relations and more rational but not necessarily predictable behaviour.

I dont even know what Im trying to say here lol... just thinking aloud...

What you're saying is that, rather than the game having multiple entities (AIs) with multiple agendas and strategies, it's simply one collective entity controlling several Civs. Which basically means that all of the Civs in each game are working together to defeat the human player. This is an intriguing thought. I'm not sure if it's accurate...but intriguing nonetheless...

 

[Cicerosaurus]

I don't think it is correct however. Otherwise the AI's would not go to war with each other while I look on and build libraries etc...

 

[qalsip1977]

It would if it were programmed to. When you consider the 4-5 turn wars that happen between AIs, the times when one minute they are at war and the next they are DoF and denouncing you, there is some merit to this theory. Consider this scenario:

1. AI-A DoW AI-B
2. AI-A DoW human player.
3. 5 turns later AI-A and AI-B declare peace.
4. Human player attempts to negotiate peace with AI-A but the AI will not have it.
5. 5 turns later AI-A and AI-B DoF.
6. AI-B DoW human player.

I've seen this kind of behavior happen MANY times. It's clear that the AI is programmed to defeat the human player at all costs. It only seems logical that they'd all be in it together . I'm not saying I think this is certain...I'm just relaying evidence I've seen that it could be so.