AI attitude

[Yndy]

Bamspeedy I’m rather late to see this article but it is fabulous. It’s so good that I think it’s exploitative to know so much. For example the 100g donations in 10 g increments is a little unfair to know.

I infer one very, very important thing from here and it fits into my experience so far.

The breaking of an agreement is ignored by the civs fighting with the civ that you have broke the trade with.

Your ROP betrayal is disregarded by any current enemies of the civ that you betrayed.

This would mean a lot e.g. to allow gpt trading with your stained rep.

Example (not tested): I ROP betray the Germans but not finish them instead contain them in a small tundra-ville surrounded by my troops. My rep is stained and I suffer the consequences. But I really need to get Sci Method to get ToE. Japan has it so I declare war to Germany, bring Japan in with a lump sum payment (should be small as Germany is very little and weak. Bingo! I can trade gpt with Japan who now is glad that I betrayed the hatred Germans. I make sure that no Japanese sets foot on the German soil so that I can do the thing again later.

Anyone care to test this (I don’t have time)?

 

[henrsve]

Really nice work Bamspeedy! One question that I have is if the rep goes down even if the cities are auto-razed (they have the size of 1 and gets razed)? If that's the case you can't do a rush in the beginning without have a terrible rep., right?

Regards

 

[ControlFreak]

@henrsve

You must have missed Bamspeedys previous post on page 1.

 

[Gen]

One more thing which certainly have negative impact upon AI attitude is piracy (using privateers). Does somebody know the exact numbers and if this is a permanent penalty?

 

[Bamspeedy]

henrsve-Yes, early war would be hard to do and keep your rep at it's peak. So you must choose your target cities carefully and go for capitals or any other cities that have over 10 culture in them. I would also say go for cities that are at population 2+, but the AI frequently whips these cities back down to size 1.

The penalty that other civs (the non-victims) add towards you for razing is only +1, so you could raze a few and still have an OK rep, just don't raze too many cities.

Gen- I did not study privateers, but they have 'hidden nationality', so the AI isn't supposed to know where the ships came from or belong to. It would be suprising if using privateers did cause the AI to become mad at you.

 

[Gen]

Originally posted by Bamspeedy
Gen- I did not study privateers, but they have 'hidden nationality', so the AI isn't supposed to know where the ships came from or belong to. It would be suprising if using privateers did cause the AI to become mad at you.

Yes, it is indeed surprising, but it's true. At least that was the case in version 1.21f of vanilla Civ3 and before. I've played slightly modded games, lowering AI-AI trade rate (which greatly lenghtened the age of sail) and extensively using privateers. In short time I pissed two civilizations: America went from polite to furious and the second civ (Zulus?) form gracious to annoyed IIRC. Attitude started to drop just after I started sinking enemy ships and remained respectively in state of furious/annoyed for the rest of the game. I'm fairly sure that was the only reason that could annoy them - I did not fight them, did not break any deals etc.
I did post my observations in a thread about privateers:

Privateers

Mad Bomber concluded that AI did actually know nationality of my privateers but couldn't oficcially prove it. Maybe that's an AI cheat just like knowing whole map and every unit position? Do they know nationality of all units too or was it a coincidence?

 

[billindenver]

Hey Bamspeedy, does having an AI city flip to you affect attitude?

 

[Bamspeedy]

Ok Gen and billindenver, I'll study privateers and culture flips when I have the time. I first need to try to finish the GOTM, so you won't hear the answer from me until next week probably.

 

[Gen]

Sure Bamspeedy, no need to hurry

 

[Hygro]

i would assume everytume you sink a ship with a privateer, you lose as much attitude bonus as killing a unit during wartime.

 

[Cartouche Bee]

As I suspected, gifting is really an exploit now.

Many times early in the game the AI are strapped for cash and you have maps and techs to trade, now you can just gift them money, their attitude improves and then you sell them goods that they previously could not afford to recoup your gifts. Now you can jack every AI's attitude 10 points early in the game.

Another view point is about embassies, many times I used to make an embassy solely to improve the AI attitude for trade purposes cause I don't need ROP or an Alliance anyway, now it's cheaper to improve their attitude with gifts.

[Edit:] With gifts being of temp value, not exploit almost worthless.

 

[ControlFreak]

Originally posted by Cartouche Bee
As I suspected, gifting is really an exploit now.

Many times early in the game the AI are strapped for cash and you have maps and techs to trade, now you can just gift them money, their attitude improves and then you sell them goods that they previously could not afford to recoup your gifts.

This makes me think that the limit of +10 for gifts is not an error but rather a intentional fix to exploiting this gift/tech sale for more gold loop. Otherwise, you could rinse and repeat this throughout the middle ages and always have AI be gracious.

 

[zagnut]

In the latest GOTM game I decided to try for a diplimatic victory toward the end of the Industrial Age. I had finally caught up in techs and because of my wise plays, all of the other civs were involved in a World War on the other continent. Because of alliances that I made most of the other civs were polite to me. I figured this was going to be an easy diplomatic victory. As a result I decided not to invade the other continent.

Then your article on Attitude was published and I thought, hey, this will be even easier because all I have to do is give away lots of techs and luxuries, and pay 50 gold, and the other civs will love me. Well, it didn't turn out that easy.

As soon as the World War ended, the attitude of the other civs went right to furious and no matter how much I paid them off, they stuck at annoyed (the next level above furious). I kept paying and paying, giving and giving, but could never improve their attitude above annoyed. As a result, all of the elections were inconclusive and resulted in many abstentions.

At the time I was in Republic, ahead in techs, score and territory owned.

Here are some things I noticed:

1. I know you only tested in PTW because vanilla civ doesn't have a debug mode. I'm not sure all of the conclusions hold, but there appears to be no conclusive way to test this.

2. I paid lots of 50 gold bribes because I had lots of gold. I never noticed that any of them raised the attitude of the other civ.

3. Rights of Passage seemed to work just as you tested. They made a big difference. Usually the attitude went up one level with the agreement.

4. Alliances made the biggest difference. A civ that was annoyed with me before the alliance, became polite after it. I thought this was a little strange because your test showed a max benefit of -5. Donating a tech also gives a -5 benefit, but I didn't find it to be anywhere near the benefit of the alliance. I gave away lots of techs, 3 in one turn to the English, and never noticed an improvement in attitude.

5. It could be that I was so reviled that no matter how much I gave away it could never get me into the negative numbers. However, I don't think that is the case. Except for Ancient Times the only wars I fought were declared against me by the other civs. I did raze about 1/4 of the cities on my continent. I never broke an alliance or a ROP.

6. The max benefit from gifting is -10. Does that mean that after I give away ten gifts of 50 gold each to the same civ, I can never get a further benefit, even from giving techs, luxuries, etc.?

I don't want to sound as if I am criticizing your work. I am not. I just wonder whether I am interpreting some of your results correctly and whether there is a difference between PTW and vanilla civ. Any thoughts you have would be greatly appreciated.

By the way, I gave up on a diplomatic victory, invaded the other continent and achieved a space ship victory in 1758.

 

[SirPleb]

Bamspeedy, if you don't mind and know the answers, I have questions about one part of your article.

In the section titled "Permanent Attitude Changes" you note two factors caused by declaring war:
+4 "if you’ve declared war on that civ before"
+1 "You declare war on a friendly civ (all civs who are at peace with this civ will add on the point, others will add the point if they ever sign peace with this civ)"

Questions I have (if you know the answer, I wouldn't dream of asking you to go dig them out, you've already given us so much here! ):

1) Does the +4 for a second declaration of war apply to all Civs (who aren't at war with the target of course) or just to the target Civ?

2) If the +4 for second war only applies to the target Civ, then what about the +1 which applied to other Civs - is it applied to everyone again?
I.e. suppose Japan, China, and I are all at peace. And suppose I've declared war on Japan once in the past, and I now declare war on Japan again. As I understand it:
a) Japan will now have a permanent +5 added for the two wars I started with her. (+1 first time, +4 second time.)
b) China will have a permanent +1 added for the first time I declared war on Japan. Does China add another +4, or another +1, or nothing for this second time I'm hitting Japan?

3) Is the +4 cumulative if you repeatedly declare war? I.e. an additional +4 the third time you declare war?

 

[SirPleb]

zagnut, I've been playing with all this info from Bamspeedy and I think I can answer your questions:

Originally posted by zagnut
1. I know you only tested in PTW because vanilla civ doesn't have a debug mode. I'm not sure all of the conclusions hold, but there appears to be no conclusive way to test this.

I think it highly likely that most (maybe even all) of Bamspeedy's discoveries apply exactly the same way in 1.29. His discoveries perfectly explain some situations which have mystified me in the past. I've replayed from a few selected saves to test some of his discoveries and they've worked in 1.29.

2. I paid lots of 50 gold bribes because I had lots of gold. I never noticed that any of them raised the attitude of the other civ.

Could it be that before you started checking the results you had already gifted them something? After the first 100 gold value of gifts (i.e. gifts of tech, maps, or gold) you'll have the maximum -10 benefit. Further gifts won't make any difference. So if you give them one tech which they feel is worth 100 gold or more, that will get the maximum benefit already.

5. It could be that I was so reviled that no matter how much I gave away it could never get me into the negative numbers. However, I don't think that is the case. Except for Ancient Times the only wars I fought were declared against me by the other civs. I did raze about 1/4 of the cities on my continent. I never broke an alliance or a ROP.

Razing those cities was probably the problem. That's the most important thing I've learned from Bamspeedy's work. I don't know how big your continent was but suppose you razed 20 cities - that would be a permanent +20 applied by every other Civ! (Regardless of who each city belonged to.)

6. The max benefit from gifting is -10. Does that mean that after I give away ten gifts of 50 gold each to the same civ, I can never get a further benefit, even from giving techs, luxuries, etc.?

Yup, that's what it means. And the number is actually ten gifts of 10 gold each, or just two gifts of 50 gold each. Sure doesn't take much to reach the maximum!

 

[zagnut]

Thanks for the answers SirPleb. I didn't realize razing gave you such a hit. The fact that it is permanent and applied by all other civs will certainly make me think twice about razing cities. Usually I keep cities in the later game because my culture has caught up. But in the early game I am usually behind on the higher levels and don't want the flip problem so I raze. Guess I'll have to re-think that strategy.

On the issue of not getting an effect from my gold gifts, the reason probably was that I had gifted a few techs previously. I would imagine that it doesn't take many tech gifts to equal 100 gold.

It certainly brings home the point that if you want a diplomatic victory, you better start planning it from the beginning.

 

[Bamspeedy]

Originally posted by SirPleb

1) Does the +4 for a second declaration of war apply to all Civs (who aren't at war with the target of course) or just to the target Civ?

2) If the +4 for second war only applies to the target Civ, then what about the +1 which applied to other Civs - is it applied to everyone again?
I.e. suppose Japan, China, and I are all at peace. And suppose I've declared war on Japan once in the past, and I now declare war on Japan again. As I understand it:
a) Japan will now have a permanent +5 added for the two wars I started with her. (+1 first time, +4 second time.)
b) China will have a permanent +1 added for the first time I declared war on Japan. Does China add another +4, or another +1, or nothing for this second time I'm hitting Japan?

3) Is the +4 cumulative if you repeatedly declare war? I.e. an additional +4 the third time you declare war?

I admit the way I wrote it, may have confused people and I should have worded it differently. Or else you misunderstood what I wrote. If you know how I should word it to avoid confusion, please let me know.

The +4 is what the target civ adds on the first time you declare war. So the first time you declared war on Japan, Japan would add on +4. The +1 is what the non-target civs will add on when you declared war on the civ you had been at peace with. So in your case, China will add on +1, the first time you declared war on Japan. I used the words in past tense, but the penalties apply even when you haven't been at war with them in a long time.

I did forget to study if this was cumulitive if you have multiple wars with the same civ. I know I was thinking about that, but for some odd reason I never did get around to studying it. I must have had a brain-fart.

So, if you declare war on Japan a second time, the penalties will either be: Japan is now at +4, or +8 and China would be at +1 or +2, depending on whether the effect is cumulitive or not.

I'm not 100% positive, but I seem to remember that that it is cumulitive if you declare war on multiple civs. Like if you declare war on 2 civs, then the 3rd civ that hasn't been your victim yet, will add on +2 (+1 for each one of the civs). This would make playing a 16+AI civ game harder, as after you kill 15 civs, the last civ would have at least +15 towards you just from you declaring war on the others. So you'd have to trick civs into declaring war on you, instead of you being the one to declare war.

4. Alliances made the biggest difference. A civ that was annoyed with me before the alliance, became polite after it. I thought this was a little strange because your test showed a max benefit of -5. Donating a tech also gives a -5 benefit, but I didn't find it to be anywhere near the benefit of the alliance. I gave away lots of techs, 3 in one turn to the English, and never noticed an improvement in attitude.

-5 just for the alliance, but up to -15 total if you are doing damage to that common enemy (killing units, capturing cities, pillaging, etc.). Also, they could be at +2 (annoyed) and then at -3 (polite), to a -5 could cause them to go from annoyed to polite (the -15 max could also cause a civ to go from furious to polite, like Sir Pleb pointed out several posts ago).

Because of the range values for when civs change attitude levels (annoyed, polite, etc.), this explains why there has been so much confusion in the past about what hurts your rep. Some would insist ROP improves attitude, while others say it doesn't help at all. The ones who didn't see an improvement may have been at +9 (annoyed), so they would need -9 to see any improvement (cautious at 0). Another person could be at +1, so any little thing they do to improve attitude would cause them to see an improvement.

6. The max benefit from gifting is -10. Does that mean that after I give away ten gifts of 50 gold each to the same civ, I can never get a further benefit, even from giving techs, luxuries, etc.?

Sir Pleb was right about what he said, but there's just one little thing to clarify here. If the AI is receiving luxuries, then you get the -5 for trading luxuries. So if you aren't already selling them luxuries, then donating it to them will help (only during the 20 turns that the deal is going on). Whether luxuries count towards the 'gifting' limit, I'm not sure, but I don't think it counts towards that limit, because it does apply to another category (trading luxuries).

For gifting I only tested lump sum payments and techs (these were the easiest for me to put a value on and now how much money it took to get an attitude change). I didn't test if resources would count towards that, or if gold/turn does (like 5 gpt for 20 turns=100 gold max needed).

 

[Catt]

Firaxis has been pretty good, not great, about how reputation and attitude are handled -- i.e., for the most part, effects on player attitudinal and reputational standing are largely within control of the player. And, based on that perception and my own game experience, I was skeptical about an "auto-raze" carrying the same hit as a deliberate razing. So I tested it myself -- one very brief test, so I can't call it conclusive.

Sure looks to me like auto-razing results in an attitude hit just as if you affirmatively decided to raze. I tested this by comparing enemy and neutral civ attitudes in three diffeent circumstances: (1) 5-turn war with an autoraze; (2) 5-turn war with a city capture (no raze); and (3) 5-turn war with a deliberate raze (not an auto-raze). The auto-raze and deliberate raze games resulted in more or less identical attitude adjustments. The capture game resulted in significantly less attitude deterioration in the enemy (and the slight [-1] improvement in neutral civ attitude). Absent compelling evidence that reputation is treated differently (i.e., difference in degree of rep hit if deliberate raze or auto-raze), I'd guess auto-razing is treated just like razing for purposes of reputation as well (assuming that there is actually a discrete reputation effect wrt razing).

As I said above, I think Firaxis has generally done a good job with implementations of attitude and reputation, but not in this case and that is [punch] . At least the attitude hit among all but the "razee" civs is very minor . . . still, the player really shouldn't be so disadvantaged by a "feature" that requires an auto-raze without player input. [punch]

 

[Bamspeedy]

Well, I did find out a very important fact that somehow was incorrect in my original study. This fact is about the gifts and explains what I was seeing when I played the GOTM. This new information was tested in PTW 1.14f. 8 AI civs.

The attitude adjustment for donations for the most part is temporary. How long the AI remembers your donation is dependant on the size of your donation. I gave 2 civs 500 gold each at 3950 B.C., and around turn 48, their attitude started to drop by 1 pt/turn, but even 20+ turns after having 9 points lost, they still had a -1. Then I tested with donating 1,000 gold at 3950 B.C. and it took until turn 94 before I started losing points again, but they still had -1 more than if I had never donated them anything (and they didn't change governments, build any culture, nor did I trade any techs/maps with them, or anything else that would influence attitude). Then I tried the standard 100 gold amount, and after about 5-10 turns (sometimes immediately), the attitude started droping, but still had the -1. Then I tried just a 10 gold and 30 gold donations. These started disappearing right away, including going back to the status before making the donation. At any point, I could make another 100 gold donation and restore the points back to the maximum. So just before a UN vote, you wouldn't need to donate all your money, just 100 gold to each civ.

So this leads me to believe:
-1 is permanent if you make a donation of 100 gold or higher. You will get a temporary bonus that is maxed out at -10 (-1 for each 10 gold you donate). The larger the donation is, the longer you will get to keep the temporary bonus, but you will not get more than -10 for donations at any given time-and you can not give 100 gold one turn and 100 gold the next turn and expect -20, it is maxed out at -10. The temporary bonus seems to decay at about 10 gold/turn. So if you made a 1000 gold donation, it would take 'approximately' 90 turns before you start losing points, and 100 turns before it's completely gone. A 10 gold donation would be completely forgotten by the next turn.

I am sorry if the previous information I posted caused anyone to lose any games. I think in my previous tests I had donated much larger amounts and that is why I didn't see the bonus dissappearing despite playing 30-40+ turns. I will try to edit the first post, but it will be hard, because that post is at the limit for number of words allowed.

Gold per turn payments as donations:
These are figured out as what the donation would be at the end of the deal and are added on immediately. For example 2 gold/turn for 20 turns = 40 gold, so -4 (which will dissappear before the 20 turn deal is even done). For some reason giving them 5 gold/turn only equals -9, so you have to give them another 10 gold to reach the -10 max. So giving them a lump sum of 100 gold is better than 5 gold/turn. Perhaps this is reputation figured in here? Maybe they aren't sure if you will keep the deal for the full 20 turns? Or else it is a programming error. Even then, the bonus will decay while the deal is in progress since the gpt deal was figured in at the very start.

 

[Bamspeedy]

Privateers: Yes, you do get the AI mad at you for using them. These are temporary penalties, though. It seems the same penalty as if you were doing damage to them during war. +1 when you attack their ship (but your privateer dies), and +3 if you do sink their ship. You recover these points by just not using your privateers for awhile.