Stupid AI makes me sick

[r_rolo1]

Probably he is also using a paper thin army

 

[Jastrow]

1. What does exarcebate mean? Didn't find any translation on leo.org.

In this context, probably "verschärfen" would be the most apt/common translation.

 

[Seraiel]

On B&S Mapscript there are nearly no Civs that don't have Sushi-Ressources, maybe one once in a while, be sure I screw them, but that's the exception to the rule, that's why I have to stay cautious with (almost) everyone. I am using a paperthin army, my SoD consists out of 10 Units + 10 Siege and from now 40 cities, 38 are doing research and only 2 focus on production. Otherwise, things like Medicine in the BCs would be impossible, and getting all the ressources from the AI when founding Sushi is enormously important, because one really needs all of them to get the +20-30 that really make the score explode. 2 pissed of AI often mean 20 ressources less, and that's already -5 that one could have.

 

[vranasm]

no expert on this, but a question... I think you use the maximum number of opponents for Huge map allowed by HOF. What about reverse situation? Try to use the least number of AI's allowed by HOF?

It should allow you more cities / land, since domination land limit is tied to number of AI's and thus you should be more aggressive.

From my very limited experience if I need longer game (for example space race) it usually is better to have lesser number of AI's and having more cities by myself.
Highest number of AI's is usually best for quick military games due to the rushes being much better and crowding limiting AI tech speed and rex.

I guess getting highest score should involve a lot of cities of yourself and thus the lesser land limit the better.
Maybe the game then is a bit tougher (the "checkers" probably won't rush 6 AI's but "only" 2-3), but I would maybe give it a try...

Really not sure now, but I vaguely remember the number for huge being something like 10 or 11 AI's and if you "checker" rush 3 of them you have almost the "standard" number of AI's you're used to.

 

[DanF5771]

Thanks for clarifying my mistake.

Although it may be AI_getHasMetCounter that is the main reason, the test game still bothers me how I didn't get with my tech partner any "Our trades have been fair and forthright!", but easily got the max "You have traded with our worst enemy!".

On noble it should be equal trading...

Unevenness of trades is only relevant for "Our trades have been fair and forthright!", but it is completely irrelevant for "You have traded with our worst enemy!" as long as it is a trade where both parties put something on the table.
What counts for the diplo demerit an AI gives you, is the value of the items that his enemies received from you. As is well known, the magnitude is inversely proportional to the time you've had contact with this AI. Probably not so well known: gifts have 3 times the impact that trades have.

In your test game, try taking various technologies (e.g. The Wheel, Monotheism, ... , Stealth + Fusion are too expensive to see differences) away from you and Gandhi with World Builder and then either 'trade' them to him again (for the known world map = uneven; or an equivalent tech = even) or gift them back and check out how Sury reacts ...

 

[BigBossPlayer]

@Seraiel

about our last conversation about hill archers

the number is in the green car ... trail of 80 squares ..

alliances with stef & mar ..... you know what i mean

please write email o p.m.

 

[Kaosprophet]

• AI demands monopoly techs and very expensive techs (both, the player will never succeed with, normally a player cannot even ask for such techs if it's not MM)
• AI demands the complete money (again, as a player it's already hard to get 50g from an AI) .
• AI can ask to join their war, when one already is at war (AI itself would answer with WHEOOHRN to that, the player can't) .

I'm very happy with Civ4, it's the best turn-based strategy game and one of the best games I've ever played, I do not even have problems with AI getting bonuses instead of Fireaxis programming a better AI, I see that as a challenge like in chess and as a lack of programming ressources. Again though, those points make AI look very stupid, and that's the odd-side of a really beautiful game. If Fireaxis had thought that through, they would give the player the same means the AI has, i. e., the player having an attitude towards the AI, then the player can choose if he is a type like MM, or more a type like any other AI, if he chooses to be an MM, he gets bonuses in Diplomacy and AI can ask for monopoly techs, if he chooses to be normal, monopoly techs are blocked for a certain number of turns, the AI could not ask, also the player would have something like a WHEOORHN with which he can block the join-war demands, a. s. o.

See, this is what I mean by bad fixes. This would only be advantageous for games where the player blatantly wants some kind of short-cut around the diplomacy side of the game; otherwise, it's a freedom-limiting parity for it's own sake that just opens up even more avenues to exploit the AI.

A better solution is for the AI to be reacting to each other according to the same criteria it reacts to players. That is, it'll run (behind the scenes) the same sequence of demands etc. that it would against a player; and when it runs into a "we demand x" when "x" is something the ai is hard-coded to refuse (monopoly tech, go to war when in wheoorhn,) it takes the same diplomacy hit as the player would (adjusted towards difficulty, of course.)

*Maybe* one could open up the opportunity to make requests that'll be flat-out refused regardless, but it would be a purely aesthetic change for the most part: either experience or code-delving will fairly quickly let you know what will be refused out of hand. Having them blocked out from the start is just saving you from going through the motions.
(And, to an extent, avoiding the 'provocation exploit' that was once common in Civ3 - where one goads the AI into declaring on him to reduce war-weariness.)

And getting negative diplomacy for giving in to a demand, aswell as AI going to war for denied demands, I see that as a bug or very sick feature.

It's a feature, based on the notion that trying to please (or even just not-annoy) *everyone* should be a difficult (if not impossible, depending on personalities involved) juggling act.

You're trying to do something very difficult, and even making it somewhat more difficult than normal on top of that, in hopes of achieving greater rewards for pulling it off. If it was any easier, it would have been done many times already. Any improvements to the AI on that front should keep that in mind, and be focused on making the difficulty more intuitive (or, even better: more scalable) rather than on erasing the difficulty of what should be a tricky endeavor.