AI Attitude and Trade

I never saw an article on the interaction between AI attitude (Gracious <-> Furious and beyond) and the cost of goods purchased from the given civ, so I did a few simple tests* on some existing save files (Civ3 1.29).

In the most detailed test, I asked Persia what they would need for (monopoly priced) Radio at decreasing attitude levels in terms of gold per turn (Deity level, Persia has a higher Power rank):


-------------------------------
Attitude - Asking Price (Negotiated Price)

Gracious** - 681 gpt (620 gpt)
Polite (1 demand down from Gracious) - 699 gpt (635 gpt)
Cautious - 705 gpt (641 gpt)
Furious (1 demand down from Annoyed) - 728 (662)
Furious (+ 11 demands) - 775 (705)
Furious (+ 21 demands) - 822 (747)
Furious (+ 31 demands) - 869 (790)
Furious (+ 41 demands) - 916 (833)
Furious (unlimited demands) - 939 (854)
--------------------------------

For each attitude point drop below Cautious, Persia charged about 0.32% more for the tech. Also, they consistently asked for exactly 10% more gold per turn than the minimum they were willing to take. Therefore, the actual price of a tech = (Asking Price)/(1.1).

In further testing it appears that the % increase in price per drop in attitude is related to the AI aggression settings. France (aggression 1) increased their asking price by about 0.08% above neutral for each point drop in attitude, while the Iroquois (aggression 2?) had ~0.16% correlation, and the Americans (aggression 3) had ~0.24% correlation.

However, the price increase does not seem to be strictly determined by aggression as Germany (aggression 5) showed the same 0.32% price hike per attitude drop as Persia and there were other discrepancies which might have to do with cultural linking or something similar. I did not collect sufficient data at attitude levels better than Cautious, but the positive discounting effect for good attitude appears to be less (half?) than the negative effect of bad attitudes.

Overall, it seems like the trade bonus from having good relations with the AI is relatively insignificant until the expensive Industrial Age techs. Razing too many cities and declaring too many offensive wars will make purchases from the AI more expensive, but typical trading partners will not end up in a +100 furious state.

--------------------------------------------
*To get easily obtained values for testing, I gave away all hard goods and resources so that the AI would only ask for gold per turn (gpt). Then, I recorded the AI asking price and negotiated price at various attitude levels. To control attitude level, I first gave gifts to the target AI until they were at least Cautious (neutral attitude), then demanded tribute until they were at each desired attitude level. Testing with mixed lump sum and gpt indicates that the attitude effect is directly influencing the price of tech rather than the valuation of gpt verses lump sum.

**(undetermined level which included 100g gift plus alliance against a common enemy)

Edit- Of course, I should acknowledge the article on AI Attitude by Bamspeedy from which I designed these tests.

However, the price increase does not seem to be strictly determined by aggression as Germany (aggression 5) showed the same 0.32% price hike per attitude drop as Persia and there were other discrepancies which might have to do with cultural linking or something similar.

Click to expand...

Persia has agression of 4, right?
Well, then this breakdown might show why there wasn't a difference between agression levels of 4 and 5, but there was a difference at the other aggression levels.

Attitude adjustments for aggression levels:

Emperor-Deity
5 = +3
4 = +3
3 = +2
2 = +1
1 = 0

I asked them wanted for a peace treaty they wanted a city
I know ita almost wont take anything

As for the aggression level dependence, it seems to be a doom loop: To get a non-monopolized tech cheaper, you make a deal with a polite/gracious civ, thus even add to the attitude counter, but the cautious/furious civ is kind of "embargoed" and attitude gets even worse by the time.
Of course, this applies to games with balanced tech spreadings.
Anyway, I assume that asking for numerous techs (well, 2, that is) in one deal is more expensive than making two seperate deals,
if attitude actually changes after the first agreement. Also, it might be a good advice to deal techs always at the end of a "multible negotiations" run.

I wonder if other tradings/agreements would apply to the same rules - such as selling techs, ROP, MA, resource deals etc.

I *suspect* loan rates to be attitude-independent.

PS: How come that (most recently) some threads are placed in "Strategy&Tips" while they fit well to "Strategy Articles" (and vice versa)? I'd suggest to move it before it falls into the disregarded cellar of S&T.

Originally posted by Grille


PS: How come that (most recently) some threads are placed in "Strategy&Tips" while they fit well to "Strategy Articles" (and vice versa)? I'd suggest to move it before it falls into the disregarded cellar of S&T.

Click to expand...

Take the compliment and move this thread, Zwingli!

Based on Bamspeedy's "AI Attitude" article

Regent/Monarch
Aggression Setting - (Base Attitude)
5 - (+3)
4 - (+2)
3 - (+1)
2 - ( 0)
1 - (-1)
-----------------------------------------------
It appears that the base attitude of a civ determines how much diplomatic relations effect the value of trade deals. Poor diplomatic relations reduce the sales price of technology in the same way that they increase the cost of tech (although to a much lesser extent).

When selling Nationalism on the Regent level:

-1 Base Attitude: Diplomatic relations have no effect on the sales price of technology whether they are Gracious or +100 Furious (262 gpt @ all levels).

0 Base Attitude: Each point in attitude drop reduces tech sales price by ~0.04% of neutral price (Nationalism sold at 257 gpt @ Cautious and 247 gpt @ maximum Furious)

+1 Base Attitude: Each point in attitude drop reduces tech sales price by ~0.07% of neutral price (Nationalism sold at 252 gpt @ Cautious and 234 gpt @ maximum Furious)

+2 Base Attitude: Each point in attitude drop reduces tech sales price by ~0.10% of neutral price (Nationalism sold at 247 gpt @ Cautious and 222 gpt @ maximum Furious)

+3 Base Attitude: Unfortunately, I used Germany as my aggression 5 civ, and they got Nationalism as their free tech. Given the trend I would guess that they would have reduced sales price by 0.13 - 0.16%.
--------------------------------------------

Combined with the earlier data, it seems that the extent of diplomatic effects on trade valuation is proportional to the starting attitude (based on aggression). When selling technology, it is an almost negligible effect for commonly seen attitude levels. The higher attitude effect seen when purchasing tech for gold per turn (as opposed to selling it) indicates that the AI charges more interest due to poor diplomatic relations. Another useful fact is that the AI offering price on technology was always 81% of the maximum negotiated price (assuming the AI civ had enough gold).

The standard 81% AI offering price on tech and 10% ask aren't a surprise, given how the experienced player can often predict the bottm line within a coin or two

nice article!

very valuable information. This should be archived someplace.

You could ask a mod to move this to the strategy articles forum. If it's in the Strategy articles forum it may even end up in the War Academy.

Has anybody checked what percentages of full price the "insulted", "doubtful", and "close" levels are at? This is useful when you need to know just how short of making a deal you are.

What about the effects of culture and favored gov? Would it make a difference if they were impressed or in awe of your culture?

Originally posted by budweiser
What about the effects of culture and favored gov? Would it make a difference if they were impressed or in awe of your culture?

Click to expand...

It was because of complicating factors like this that I eventually gave up on trying to find an exact formula for how much attitude influences the balance of trade. Culture, favored government, shunned government, cultural linking, reputation, and relative power all could potentially have some type of impact on the cost of trading.

I did find out in the process of testing that the AI use of nukes is also attitude dependant. If their attitude is better than around +50 (extremely furious), then they will not launch any nukes on the outbreak of war, but if it is worse than around +50 they will launch the entire ICBM arsenal at you on the first turn of war. I suppose it wouldn't be too difficult to have relations deteriorate to +50 during a late modern war with cities being razed all over the place. It might be a good idea to send your enemy a modest gift before engaging in modern war .