AI Understands Monopoly Value

[Catt]

I came across something interesting yesterday in my current game. I don't recall seeing this before, and also don't recall seeing a post or thread on the subject.

It seems that in my current game an AI civ understands that it controls a monopoly of one of the available luxuries, and seeks to charge a monopolist's price for the luxury. The civ in question is Russia. Catherine controls several Spices and all the Incense on the map. One other civ controls several Spices. In other words, the "Spice-less" of the world may trade with two different potential suppliers, while the "Incense-less" of the world must go to Catherine. The price differential is not huge, but it is evident, and seems to contradict in a small way the well-known and widely-believed "fact" that AI civ's value luxuries only on the basis of the number of happy faces they will produce in the buyer's empire.

Below is a screen shot showing Catherine ready to accept a supply of Silks and 8 gpt for a supply of her Spices (I have obscured the minimap because I may share this game beginning with others, and don't want to spoil the game).

[Catt]

Below is a screen shot showing Catherine unwilling to accept a supply of Silks and 9 gpt for a supply of her Incense. Both shots taken from the same turn, without any intervening changes (without ever leaving the diplo screen).

[Ethelred]

Interesting. I think someone mentioned this possibility recently but it was in passing and no evidence was presented.

Thank you.

[Theseus]

Interesting... one would think there would also be a modifier for the number of resources Russia has, in which case spices would be more expensive. That's obviously not the case.

Cathy should charge more!!

[alexman]

Good job, as usual, Catt!

I agree with Theseus. The value of the monopoly should be greater than what it appears to be.

In any case, monopolies are something to aim for. In addition to not having the AI trade with each other instead of with the human, the human can charge more for his own monopoly resources.

[Catt]

I didn't figure out to the gold piece what the difference was, but I could go down to 7 gpt and add 10 or 12 gold as a lump sum for the Spices and could add some number of gold as a lump to get her to accept 9 gpt for the Incense -- as you both have noted, not a big difference.

BUT - assuming Cathy survives, it will be interesting to see the price differential when hospitals (and larger cities) come into being. Also, it would be interesting to see the price differential if, say, 3 or 4 different civs had Spices -- i.e., plenty of suppliers in an efficient marketplace. My gut tells me that there is an algorithm that works something like the tech devaluation algorithm -> an equation involving number of suppliers, number of potential buyers, value in happy faces, etc.

Catt

[bongo]

Thats interesting. In my current game there are two suppliers of incense, myself and the french. If I conquer their city the rest of the world will have to buy incense from me only. Which mean I can charge more

Life is good!

[Alex]

This thread proves that we need a comprehensive and detailed commerce screen, not that almost useless screen that we have now.

[DeepO]

well, I've seen it before as well, but I'm unsure what is causing it. I think there are more factors at work, of which monopoly certainly is one. Other things might include how much is available of a certain luxury, how bad the other party needs it (I've seen quite a few times that the 8th lux is a lot more expensive than the 3rd or 4th.), and amount of happy faces or size of the recipient. Sometimes you see a price difference between luxuries even if there is no monopoly.

BTW, the same sometimes happens with resources: in my last game the Indians wanted to pay more for saltpeter then for rubber (oil was available, but no tanks yet, so saltpeter wasn't totally useless). I was the only one with an excess saltpeter, while rubber could be bought from multiple civs...

DeepO

[Arrian]

I have noticed an AI placing a higher value on one of its luxuries over another, but for some reason I didn't really think about it.

-Arrian

[Skeeve]

Nice analysis.
It makes sense to me that the algorithm dealing with resource value may include a modifier for a monopoly (or oligopoly).

In any case, from a miltary standpoint, the AI is programmed to recognize the value of resources (as is evidenced by bombers and battleships destroying improvements on them), so it would make sense that the AI would be programmed to take advantage of them peacefully (economically).

On the other hand, the AI sometimes spends a lot of time bombing (or pillaging) improvements on useless resources (like saltpeter after tanks are around).

[vmxa1]

It seems to me that the price they want is a function of how much you have. If you have lots of cash they will ask for large sums. It hardly ever seems to be a relationship with it true value.

[Catt]

Originally posted by DeepO
Other things might include how much is available of a certain luxury, how bad the other party needs it (I've seen quite a few times that the 8th lux is a lot more expensive than the 3rd or 4th.), and amount of happy faces or size of the recipient.

I have always operated on the belief that the value of a luxury to any given civ was directly related to the number fo happy faces it produces. To civ A, the luxury might be luxury number 8 in a sprawling, well-developed empire with numerous size 12+ metropolises. In that case the luxury will generate 4 happy faces in every metropolis that has a marketplace. However, the same luxury is worth a heck of a lot less to a 3 city civ that only has one existing luxury -- it will produce exactly 3 happy faces. This pricing methodology is why we so often see complaints from posters that %^**&^ wants furs, wines, gems, spices, my world map, combustion, and 425 gold all for one lousy supply of silks (i.e., the human player is way ahead of %^**&^ and the supply of silks will generate a ton of happiness for the human while the return value of the multiple luxuries doesn't even compare).

Sometimes you see a price difference between luxuries even if there is no monopoly.

I don't doubt it - but I'd just never seen it before. I have definitely seen a price difference for the same luxury from different civs, but had never seen a price difference for different luxuries from the same civ.

BTW, the same sometimes happens with resources: in my last game the Indians wanted to pay more for saltpeter then for rubber (oil was available, but no tanks yet, so saltpeter wasn't totally useless). I was the only one with an excess saltpeter, while rubber could be bought from multiple civs...

I don't have a good sense of how the AI values strategic resources. My best guess is it has something to do with availability, and with the productive power of the buyer civ. I played a recent OCC game in which a scarce oil resource cost me 10 - 15 gpt during the age of tanks, planes, and battleships, IIRC. But how many tanks can an OCC civ make in 20 turns? In a "normal" game with a lot of cities that can produce a lot of advanced units in 20 turns, I can't ever recall getting a deal on oil that was even close to 10 -15 gpt.

But the real difference is in the discussion between strategic and luxury resources. Strategic resources enable certain units and even certain improvements -- the value of the units obviously varies over the course of the game. But luxury resources are luxury resources - they all produce happy faces, and one's citizens do not care if one chooses to import furs, wines, gems, etc. - they just want the happy faces.

So, I honestly have no idea if there is a monopoly or "scarcity" factor coded into luxury resource valuation, but I was surprised to see pricing variations on different luxuries from the same civ -- if the only pricing methodology was happy faces in buyer civ, one luxury should always be equally valuable as the other luxury.

Originally posted by vmxa1
It seems to me that the price they want is a function of how much you have. If you have lots of cash they will ask for large sums. It hardly ever seems to be a relationship with it true value.

In one of the chats Soren indicated that this was coincidence and that the AI does not price based upon the human buyer's ability to pay.

Catt

[alexman]

Originally posted by Catt
I don't have a good sense of how the AI values strategic resources. My best guess is it has something to do with availability, and with the productive power of the buyer civ.

I believe it also depends on what the buyer can do with the strategic resource. For example, you can buy oil (available with Refining) dirt cheap before you discover Motorized Transportation (which allows you to use the oil for tanks).

[Catt]

Originally posted by alexman


I believe it also depends on what the buyer can do with the strategic resource. For example, you can buy oil (available with Refining) dirt cheap before you discover Motorized Transportation (which allows you to use the oil for tanks).

Yup - exactly. What will the oil allow me to produce?

If I haven't discovered Combustion, then nothing! If I've gotten all the way into the modern age but only have 1 or 2 cities, not much! If I'm one of the 3 or 4 leading civs with a decent and well-developed empire, then my lack of oil is a critical handicap, and its availability would allow me to crank out an awful lot of powerful units quickly. If I'm the biggest, baddest, meanest civ around . . . well, I might just take that oil rather than buy it

[vmxa1]

So Catt are they saying it will ask for 7455 gold even if I have 15? Not that I doubt them, but that is some coincidence as it happens to me every game. You can even run a test and turn of the RNG and then spend some cash and offer a deal, the price is now different.

[Catt]

Originally posted by vmxa1
So Catt are they saying it will ask for 7455 gold even if I have 15? Not that I doubt them, but that is some coincidence as it happens to me every game. You can even run a test and turn of the RNG and then spend some cash and offer a deal, the price is now different.

Well it was only a quick exchange so he didn't get into detail, IIRC. But I walked away from the chat log with the impression that the AI will determine an asset's value regardless of whether you have 10 gold or 10,000 gold in your treasury. It will then look at your tradeable assets and put together a deal that seems attractive (by whatever measure it uses) that equals or exceeds the "base" price it has determined is appropriate. I have not tested this myself.

In my experience, if you don't have enough gold to pay the price, they will ask for additional non-monetary consideration - maps, techs, resources, etc. (i.e., they won't ask for more gold than you have). If you don't have enough tradeable assets period, gold or in-kind, they will say something to the effect of "Regrettably, I don't see how any deal is possible at this time" (after you've selected "What may I trade you for _________")

BTW - vmxa1 knows this but for others who don't, the "What may I trade you for XXXX" is not a good indicator of true price as they will "overcharge" you - I can almost always negotiate their asking price in the rare instances when I use the "what'll ya give me for this" or "what'll this cost me" options.

Catt

[Catt]

vmxa1 - I found the snippet of chat I was remembering. Here it is with the intervening chat stuff cut out:

Bleyn: "Soren, is there some sort of exchange rate when it comes to money in trades? It drives me a little nuts that when I've got a nice full treasury and try to trade the AI wants most of it, but if I use up my money, they are content with getting a pittance..."

SorenJohnson_Firaxis: "bleyn: that is not a factor... coincidence?"

It was from the chat held on July 19 and appears on page 5. The link is here.

Catt

[vmxa1]

That nice, but what can i say, the more money I have the more they ask for plain and simple. It may not have anything to do with how they value it, as I am sure they sell it to the other civs for less. A coincidence is once or rarely, not constantly. It seems they ask for a very big chunk of what ever cash you have at all most all times, so I just can't buy the coincidence. Is that not what you see? Do you see ocassions where you have say 2000 and the AI says you can have it for 50?

[Catt]

This subject is off the topic of the thread, but, yes, I don't think they ask for more simply because I have more. I might not dispute that they ask for more gold if you have more gold (instead of asking for other items), but I'd say that I haven't seen any "patterns" or "coincidences" that they ask for more value in the aggregate than they otherwise would if I happen to be richer (all other factors being equal). My view is that the AI likes gold -- gold offers the maximum freedom of choice. As the game progresses, the AI still prefers gold so asking prices increase in actual gold terms but value is still calculated at true value bases. In my experince, buying techs or resources in the industrial age is significantly more expensive than in the middle ages -- but producing gold in the industrial age is significantly easier than it is in the middle age.

And yes, I have seen instances where my treasury sits at 2000 and a world map may be had for 10 gold; I have also seen intances where my treaury sits at 2000 and a world map mimay be had for several hundred gold -- this usually indicates a new landmass that the AI has discovered.

Finally, AIs may trade any number of things to other AI civs for less than they would trade it to you -- but I think that is a well-known result from an endlessly-discussed "AI-to-AI Trade Rate" editor factor. They won't offer it for less to another AI, but they will come to agreement with the other AI for far less than what they determine is "fair" value.

[vmxa1]

Maps are not much of a test as once the map is exposed they are really worth nothing. But anyway it does not matter, just curious. I rarely do trades as from my perspective they ask too much for anything, so I just prefer to take it.

[DeepO]

Originally posted by Catt
I have always operated on the belief that the value of a luxury to any given civ was directly related to the number fo happy faces it produces. To civ A, the luxury might be luxury number 8 in a sprawling, well-developed empire with numerous size 12+ metropolises. In that case the luxury will generate 4 happy faces in every metropolis that has a marketplace. However, the same luxury is worth a heck of a lot less to a 3 city civ that only has one existing luxury -- it will produce exactly 3 happy faces.

I'm not doubting that, but I think there is more to it. would the AI simply count how many faces it gives in an empire? I mean, some cities have marketplaces, others have not. Would the AI just count cities, or count the number of added happiness exactly?

Plus, there is another point to consider: I normally have all 8 luxuries in each game, at least half of them traded. As these 4 traded luxs deals don't end all at the same turn, they are, when they're up for renewal, always the 8th luxury... gaining it would mean 4 happy faces. This is different than what you experience in the game: when you have the 8 luxs, the 4 traded luxs are highly different in price. There has to be something else at work.

Maybe, but this would need testing, there is a fixed order in the luxs: let's say the spices are always first, while ivory is the last in the line on the city view. This means that spices always provide one happy face, while ivory can, if all the rest are present, provide 4 faces. This would make it possible to have a fixed price: if your spice deal expires, the cost would be related to 1 happy face, while if your ivory deal expires, the price would be for 4 faces.
I don't know the order by heart, but if this would be true, there would be a built in value of lux items which is different: ivory is far more worth then spices... a strategic insight that hasn't been made yet. I will certainly look into this in next game.

I don't doubt it - but I'd just never seen it before. I have definitely seen a price difference for the same luxury from different civs, but had never seen a price difference for different luxuries from the same civ.

Oh, I've seen it quite regularly. Before going shopping, I make it a point to first of all see all the civs, and see what they would offer me (selling maps). I do this with techs (as there will be differences between civs as well: some will ask for more for the same tech), but I certainly do it with luxs. It can mean a difference if you first buy wines from Civ A, and later spices from Civ B, but unfortunately that is very hard to tell in advance.

I don't have a good sense of how the AI values strategic resources. My best guess is it has something to do with availability, and with the productive power of the buyer civ. I played a recent OCC game in which a scarce oil resource cost me 10 - 15 gpt during the age of tanks, planes, and battleships, IIRC. But how many tanks can an OCC civ make in 20 turns? In a "normal" game with a lot of cities that can produce a lot of advanced units in 20 turns, I can't ever recall getting a deal on oil that was even close to 10 -15 gpt.

maybe... it does come into play (oil being extremely cheap before motorized transportation is one example), but this would not explain the difference between saltpeter and rubber. After all, if you only have 1 city, the productive power is the same: you both need infantry and cavalry. Iron is something else, as you will need it for RRs and factories, and coal is the odd one completely, as you don't use it for units but for improvements only. It is highly complex, that's for sure.

So, I honestly have no idea if there is a monopoly or "scarcity" factor coded into luxury resource valuation, but I was surprised to see pricing variations on different luxuries from the same civ -- if the only pricing methodology was happy faces in buyer civ, one luxury should always be equally valuable as the other luxury.

It isn't a rigid demand... see above. If that order is kept and used in prices used, it would mean that one luxury is better to get than another. If it would be so, it is an important thing to find out.

DeepO

[Catt]

Originally posted by vmxa1
. . . . I just prefer to take it.

I agree - this is often the most efficient method of acquiring items of interest .

Catt

[Catt]

Originally posted by DeepO

I mean, some cities have marketplaces, others have not. Would the AI just count cities, or count the number of added happiness exactly?

I have always believed that the AI counts happy faces, and hadn't seen anyhthing to dissuade me from that view (although I certainly haven't focused on this aspect of gameplay). If, while at the trading table with an AI civ, I could trade for either luxury item at the ocst of XX gold per turn, if I traded for one luxury and then went back for the second luxury at the same price -- forget it -- the second luxury's price increased dramatically.

. . . when you have the 8 luxs, the 4 traded luxs are highly different in price. There has to be something else at work.

First off - I agree that it seems to be more complicated than simple "happy face count." But, is it always different for you in terms of price if "price" is normalized into gold? I can sometimes trade a luxury and a map for a luxury from a large, well-developed AI civ, even though the AI's luxury will be my 8th luxury. If I try and trade for my 8th luxury from a small AI civ, it might very well come at a cost of a 3 luxuries, a tech, a RoP, a large lump sum payment, and significant gold-per-turn. Unless the payment made for the luxury is in a constant form (i.e., gold), it is hard to compare luxury prices across civs (assuming one believes in the "happy faces count" pricing model of AI civs) because my luxury or strategic resource is valued quite differently by the AI buying civ.

maybe... it does come into play (oil being extremely cheap before motorized transportation is one example), but this would not explain the difference between saltpeter and rubber.

But, if the AI recognizes that in an age of rubber, rubber itself is more valuable than saltpeter, then the hypothesis might get us there. If I can trade for rubber than I have discovered Replaceable Parts and can therefore build infantry -- saltpeter gets me Musketmen, Cavalry and Coastal Fortresses. Rubber - by virtue of its later appearence -- is therefore more valuable than saltpeter.

Catt

[DeepO]

Originally posted by Catt
I have always believed that the AI counts happy faces, and hadn't seen anyhthing to dissuade me from that view (although I certainly haven't focused on this aspect of gameplay). If, while at the trading table with an AI civ, I could trade for either luxury item at the ocst of XX gold per turn, if I traded for one luxury and then went back for the second luxury at the same price -- forget it -- the second luxury's price increased dramatically.

Well... that's true, and normal as well. It doesn't have to be a pure face count, though, a simple 'this is the 7 lux, that is the 8th lux' could work as well. But it does give problems when you are renegotiating: in that case each luxury is the 8th luxury, as you are renewing a deal when the 7 other luxs are still present. And I'm positive it does not count like that: some luxuries will cost more than others... I'm going to look out if it depends on the order in the city view.

First off - I agree that it seems to be more complicated than simple "happy face count." But, is it always different for you in terms of price if "price" is normalized into gold?

I always measure in gold... that's the only point in deciding where you're luxuries would be best sold (if you don't have a monopoly yourself), and where you buy luxuries you need... I'm fully aware of the difference in price, sometimes giving 3 luxuries in return is not enough for a civ.

But, if the AI recognizes that in an age of rubber, rubber itself is more valuable than saltpeter, then the hypothesis might get us there. If I can trade for rubber than I have discovered Replaceable Parts and can therefore build infantry -- saltpeter gets me Musketmen, Cavalry and Coastal Fortresses. Rubber - by virtue of its later appearence -- is therefore more valuable than saltpeter.

True.. logic would suggest that. But I had a case where the AI favored saltpeter over rubber in my last game... so also there there has to be a monopoly thing to it.

DeepO

[Catt]

Edit / Clarification to First Post:

At the time of the screenshots, three different civs had excess Spices (instead of the two civs I originally posted).

[Ekanata]

Originally posted by Catt
I have always believed that the AI counts happy faces, and hadn't seen anyhthing to dissuade me from that view (although I certainly haven't focused on this aspect of gameplay). If, while at the trading table with an AI civ, I could trade for either luxury item at the ocst of XX gold per turn, if I traded for one luxury and then went back for the second luxury at the same price -- forget it -- the second luxury's price increased dramatically.

Have you ever try pilaging your lux tiles before trading lux from AI, and later re-roading the tile (might have to be done in different turn, just in case)? I think (assuming your theory is correct) that could temporary reduce the value of AI's lux.

From my observation, it seems that the lux value is based on its worldwide scarcity instead of monopoly matter.

[Catt]

Originally posted by Ekanata


Have you ever try pilaging your lux tiles before trading lux from AI, and later re-roading the tile (might have to be done in different turn, just in case)? I think (assuming your theory is correct) that could temporary reduce the value of AI's lux.

It never even occured to me - but a very perceptive and creative idea. It should certainly offer some interesting test possibilities. As for gameplay, it feels a little bit like an exploit to me and I'm not sure I'd be comfortable using it, but we all have different feelings on that subject. Assuming this works in testing, this would be a great Dirty Trick! (a thread in the Strategy Forum regarding exploits, quasi-exploits, and sneaky tactics).

Great idea!

Catt

[badams52]

Originally posted by Catt


I have always believed that the AI counts happy faces, and hadn't seen anyhthing to dissuade me from that view (although I certainly haven't focused on this aspect of gameplay). If, while at the trading table with an AI civ, I could trade for either luxury item at the ocst of XX gold per turn, if I traded for one luxury and then went back for the second luxury at the same price -- forget it -- the second luxury's price increased dramatically.

Catt

I would have to agree with Catt on this point that happy faces are the most important when setting price for a trade. Here is an interesting test I had with the computer trading civs recently.

I had a modded game where I increased the number of civs in a standard map to 16. I wanted to live in a crowded world and see what happens. So the luxury resources were spread over the planet as such: 14 dies, 16 spices, 11 firs, 11 wines, 10 insense, 9 silks, 3 ivory, and 0 gems.

From the save game, here is a breakdown of who owns which luxury, how long before renegotiation happens, which civ currently gives me the luxury (*) and how many available at time of renegotiation:

Incense: Egypt (Me) 10 (6 extra)
Silks: Egypt 9 (5 extra)
Wines: Egypt 9 (6 extra), America 2 (0 extra)
Firs: renegotiate in 3 turn: Germany* 7 (3 extra), England 4 (1 extra)
Ivory: renegotiate in 10 turns: India* 3 (1 extra)
Dyes: renegotiate in 12 turns: Japan* 11 (8 extra), India 3 (0 extra)
Spices: renegotiate in 12 turns: France* 7 (5 extra), China 8 (5 extra)
Russia has no luxury resources in it's border and never offers anything when I offer to trade.

From the spread you can see that I have one scarce, monopolized luxury (ivory), one mostly monopolized luxury (Dyes - never could buy from India), and two easily avaible resources (spices and firs) with one country (England) having a short supply from which I need to trade. In this era, there are 9 countries left. For every trade, I always negotiated down to the lowest price in gpt.

Here is what happened as the trades came up:

3rd turn:
Egypt's status: 6 currently held luxuries.
Firs: Deals offered: England 96 gpt, Germany 100 gpt.
Took England's offer.
Egypt now has 7 luxuries.

10th turn:
Egypt's status: 6 currently held luxuries.
Ivory: India 100 gpt. Decided to wait and see what happens in two turns to negotiate all luxuries at the same time.

12th turn:
Egypt's status: 4 currently held luxuries.
Ivory: none availavble
Dyes: Japan 79 gpt.
Spices: China 76 gpt, France 75 gpt.
thought it was strange that prices dropped so much since after trying a different save, these countries were in the 100 gpt range for their luxuries.
Took Japan's 79 gpt
Egypt now has 5 luxuries.
Spices: China 76, France 75 gpt. note: The prices did not rise during this turn.
Took France's 75 gpt
Egypt now has 6 luxuries.
And increased luxuries to 30 percent to keep WLTK days.

Being curious, I went back and reloaded my 10th turn to see what would happen if I took India's deal, how would it affect the 2 next deals.

10th turn (replay):
Egypt's status: 6 currently held luxuries.
Ivory: India 100 gpt.
Took the offer. Egypt now holds 7 luxuries.

12th turn (replay):
Egypt's status: 5 currently held luxuries.
Dyes: Japan 79 gpt.
Spices: China 76 gpt, France 75 gpt.
No surprises on the price here, same as before.
Took France's deal.
Egypt's status: 6 luxuries.
Dyes: Japan 105 gpt.

Reload 12th turn again
Egypt's status: 5 currently held luxuries.
Dyes: Japan 79 gpt.
Spices: China 76 gpt, France 75 gpt.
Took Japan's deal.
Egypt's status: 6 luxuries.
Spices: China 101gpt, France 99 gpt.

There isn't too much I can draw for certain about luxury trading with the computer except that it's very interesting that difference between having the 6th v 7th luxury is a substancial financial difference whereas there is no difference between the 5th and 6th. I wonder if the difference between 7 and 8 are non existant as well.

Also what I think this shows, which we all probably already knew is that the difference between monopolies and scacities is negligable compared to the differences between which number luxury you're trying to aquire. Ivory was a very scarce luxury with only the owner and two other civs able to trade for it. Only 1 out of every 3 civs could own it, but it's price was very comparable to firs, spices, and dyes as a 7th luxury. If I had such a rare luxury, I would want to charge an arm and a leg for it but in fact, the price of dyes - a plentiful, monopolized resource - was more expensive than ivory as the 7th luxury. And the Germans, who had plenty of firs and plenty of extra to boot asked more than England who was trying to deal it's last one. This just goes to show me that a monopoly on a resource or it's scarcity isn't as important as how many luxuries the computer opponent has and how big his civ is. This is just a matter of the rich having to pay a higher price for a luxury not because they're richer, but because the benefit to them is much higher.