A closer look discussion of tech trading since V1.29

I am not 100% sure that we did not get a back door screw up in the tech trading when V1.29 was released. Since these changes were undocumented in the release notes it indicates to me that the internal Firaxis view of the changes may have been sort of "hands-off", "that's under the hood", and "that's Soren's baby". I cannot be sure that this was tested and understood before release to us.

Many players may not experience any real impact of these factors because I get a strong impression that 80% to 90% of all games are played at difficulty levels that ARE NOT emperor or deity level. Monarch difficulty levels might see a slight influence, but games at regent and below will be just equalized or even distorted in favor of the human player. This could just mask the problem without indicating it does not exist.

This discussion will be a bit technical because it requires that any comments include: the difficulty level of the example as well as some details about the number of civs you have contact with and what techs are available in the trade matrix.

I want to start the discussion with a set of examples to see if I can establish a range of expectations of what we should see in the game if things are working properly.

--------------------------

In the first example in a pure unmodified emperor game; I am India and meet Persia as my first contact in about 3150BC. India has Alphabet*, Pottery, and Ceremonial Burial* while Persia already has Bronze Working*, Masonry*, and Warrior Code. The techs that the two civs started with are marked with asterisks. So each civ has already gained another tech and both civs have no other contacts with other civs. This is a one to one relationship. Xerxes is at 87 points and India is at 84 points when we meet. the power graph ratios are at about 3/5 for Persia and 2/5 for India. Our military guy syas we are weak because it is early and this is emperor so Xerxes still has his 7 additional free units as an advantage.

On emperor level the tech cost has been hard coded to relate to the inverse of the AI cost factor. Persia has a tech cost multiplier of 10 while the human player has a tech cost multiplier of ( 10*10/8 ) = 12.5 where 8 is the AI cost factor for the emperor difficulty level.

The values of the techs held by these two civs are:

India - Alphabet - (5 *12.5) = 63 - AI values at 5 * 10 = 50
India - Pottery - (2 *12.5) = 25 - AI values at 2 * 10 = 20
India - Ceremonial Burial - (2 *12.5) = 25 - AI values at 2 * 10 = 20
Persia- Bronze - (3 *12.5) = 38 - AI values at 3 * 10 = 30
Persia - Masonry - (4 *12.5) = 50 - AI values at 4 * 10 = 40
Persia - Warrior Code - (3 *12.5) = 38 - AI values at 3 * 10 = 30

From the human perspective, India has 113 gold worth of techs while Persia has 126 gold worth of techs. The most valuable tech in the potential trade pool is the alphabet which should be worth some combination of techs suchs as Pottery+ Bronze Working etc.

The AI is able to research the techs cheaper, so Xerxes values his techs potentially at 100 gold while he sees the India techs as being worth 90 gold.

If we miss-match the values an compare the human value of Xerxes techs (what they would be worth to the human if we had to research them) to the AI value of the India techs (what Xerxes would have to pay to research them) then we get a comparison of Persia having 126 gold worth of techs vesus only 90 gold in the India tech pool.

Just focusing on one tech - Bronze working - India could research bronze working for 38 gold but if we offer to buy that tech from Jerkses for 76 gold or 2x what it would be valued at from the human perspective, we get the message "They will be probably be insulted by this deal".

Probing to see what the exact combination of techs and and cash would be to but the bronze working tech, I find that we could trade Alphabet to Jerkses for Bronze Working and 10 gold (all his cash). So this indicates that he values the Alphabet tech more highly than bronze working but he has insufficient cash for us to figure out how much more highly that value may be.

Alphabet is a 5 tech while Masonry is a 4 tech so that trade should be closer in value. We we offer that trade we can again get the tech plus 10 gold.

Again, probing for a trade, I offer one of the cheap (2 factor) techs to Jerkses for one of his 3 factor techs and find he is insulted. I obviously need to sweeten the deal with some cash to balance the deal. If the comparison in value was his 3 factor tech valued at 38 versus our 2 factor tech valued at 20 then we would need 18 extra gold to make the trade equal plus 1 gold to push that trade into his favor, but that just gets us the "I doubt they will accept this proposal" look from our foriegn advisor monkey.

Pushing the gold plus tech offer all the way to a 2tech + 76 gold still only gets us to the "close to a deal" status. So we could research it for 38 gold but cannot buy it for 2x that cost plus a tech that sould be wothe at least another 50%.

Probing again, I try to trade the alphabet (5 multiplier) for xerxes two 3 value techs so we should have a comparison of 5 versus 6 at some level. I can't get him off the "doubt they will accept this" level even when its alphabet+76gold for bronze working and warrior code. The switch point where he will trade the two 3 value techs seems to be alphabet+one of the 2 value techs+ between 37 and 38 gold. Clearly Xerxes values Pottery and Ceremonila Burial equally and this is based on the 2 factor times some sort of formula.

I try one final probe, to see if we can get a 3 factor tech for two of the 2 factor techs plus some cash. That trade crossover point is somewhere between 3 and 4 gold plus the two cheap techs to get bronze working but the answer is different for Warrior code where we can get the tech plus 8 gold from Xerxes for our two cheap techs. Xerxes has no possible contact with another civ at this point but he is valuing Warrior Code and Bronzeworking differently even though they have the same cost factor (also India is researching Writing, so that cannot be influencing the relative cost values of Bronze Working vs Warrior Code.

There seem to be some other invisible factors that influence the value of techs in trade deals beyond the value of the techs and the number of civs that the trading partners have contact with who have the techs in question.?

Why is this trading table so imbalanced compared to the tech values we should expect to see?

Are we seeing the double impact of cost factors where it is more expensive for us to research techs and then the AI devalues these techs in trade deals even further by using the AI cost advantage to depreciate any potential trade deal?

Why would the AI be valuing two equal value technologies differently in the absense of any possible contact with outside influences from other civs or any potential resource influences??

Well, since 1.29, all I know is that I now actually do research in Ancient times (on higher levels). The valuation of techs has tilted so far that it's easier to beat them into giving techs to you than try and trade rationally (luxuries included). Seems like the tech pendulum, swung patch to patch, to extremes and we have not yet found equilibrium.

Maybe a few patches into PTW this will get sorted out but I guess we will have to 'wait and see'.

cracker: i do remember it being vaguely mentioned that ALL TRADE deals have a mark-up multiplier and the number 30 was thrown around. this was a long time ago so it may not apply.
so in addition to any transaction being first and foremost a fair deal from the AI perspective, it also has to bring in some profit - cashwise or otherwise.

Muck luck in deciphering the mysteries of AI tech pricing. I don't have a lot to offer, but thought I'd mention it just to make sure you hadn't overlooked it.

First, from the July 19 Apolyton chat where the revised tech valuation changes were discussed -- two quotes from Soren that are vague, but that may affect trading value beyond what we know via the editor values and the previous work done by mydisease on trading forumlas.

Quote 1: ". . . in 1.21 the AI devalued optional techs, but I decreased that effect in 1.29."

This factor doesn't appear to weigh against your results because none of the techs are optional -- but, unless the editor values were simply lowered / raised from version to version (I doubt), it does seem to highlight for me that AI tech valuation methodolgies are not entirely exposed through the editor values nor through previous work on the devaluation curve based on tech knowledge, map size, etc.

Quote 2A: ". . .if you are the first civ to research a tech, the value goes up in trading..."

followed by

Quote 2B: ". . . well, you will only get a big boost the first time you sell it as you are the 'sole possessor' . . ."

This quote seems to imply to me that the trading value of a tech may actually be higher than its "editor value" when only one civ has discovered the tech. Open to me whther a value of 100%+ depends on being the "sole possessor" period, or the "sole possessor" among known civs as seems to be the default rule for basic tech devaluation.

Originally posted by cracker
Why would the AI be valuing two equal value technologies differently in the absense of any possible contact with outside influences from other civs or any potential resource influences??

Click to expand...

Could the AI be researching one of the two techs -- at least under previous versions didn't the trade value of techs decline as homegrown research progressed? (i.e., could Xerxes have been part of the way to discovery of one of the two equally valuable techs?

Again, good luck -- I think this will require a lot of testing on all sorts of map sizes, etc.

cracker: good post - you have put what I suspected and felt into exact words.... i was hoping Firaxis would answer to my post asking just how the AIs arrive at a tech price, but no luck yet......

I think your Cultural value and Power value compared with the one you are trading with matters.

I think some Firaxian said that if you have a better culture they will lower their prices.

EDIT: And Yes, the more civs you know that have the tech, the lower the price will be. That's why I wait to buy techs until I know almost every civ.

Yep and additional it depends with whom you trade; France and India give usually far better prizes than other.

The BW / WC discrepancie could be that they value the attack unit which came available with WC higher than the defense unit from BW; Or they take into account that they had to do research for WC but not for BW;

That there are more influences than the numbers in the Editor is true; Best example are the high prizes for the government-techs(+nationalism);

Rowain

I just thought of another thing. If you try to trade 2 techs for 1 of theirs. Like 2 value 3 techs for a value 6. Maybe they are researching one of those techs already? Then they might value it less. Or should anyways...

Another such example: I trade Babylon Witing and get 30 Gold (all) for it. He was NOT willing to trade me Polytheism, despite me offering Writing plus 200 Gold. Rome also knows Writing, so i can understand that.

Now, i can sell him literature (of which i am the sole owner). Poly --- Lit???? No way, he won't sell it for whatever I offer.


This is unmodded Regent level - what the ****? I know he'll do a titfortat with Rome next turn since Rome has Map Making!

even better: he researched Map making himself, then offered Literature for Poly the next turn. Doh! I simply cannot get how the Ais evaluate deals......

and even more absurd: i offer Philosophy, Currency and Construction and 455 Gold and Wm for Monarchy - and Hammu doesn't take it........ Well!

Originally posted by Lt. 'Killer' M.
and even more absurd: i offer Philosophy, Currency and Construction and 455 Gold and Wm for Monarchy - and Hammu doesn't take it........ Well!

Click to expand...

I smell a war coming.

Yep, he attacked me a about 25 turns later. Well, Chivalry was 9 off and I am Japan.......... and since I was culture-pressing him his four biggest towns are in horsemen/Samurai reach of my towns

Well, the AI SHOULD do unexpected things, right? If you get it down to knowing EXACTLY what the AI will do when, there isn't much fun in playing, is there?

Anyway, if you play vs other humans you have NO IDEA what they'll give or take in a trade. They could be a newbie who doesn't understand the value of anything or it could be a seasoned pro. Heck, if everyone playing against Sid Meier KNOWS it's him he'll lose if all the other players are leery of trading with him--they don't even have to officially collude to give him bad deals because individually they'll be afraid they're giving him too good a deal until he sweetens it a whole lot!

Originally posted by pterrok
Well, the AI SHOULD do unexpected things, right? If you get it down to knowing EXACTLY what the AI will do when, there isn't much fun in playing, is there?

Anyway, if you play vs other humans you have NO IDEA what they'll give or take in a trade. They could be a newbie who doesn't understand the value of anything or it could be a seasoned pro. Heck, if everyone playing against Sid Meier KNOWS it's him he'll lose if all the other players are leery of trading with him--they don't even have to officially collude to give him bad deals because individually they'll be afraid they're giving him too good a deal until he sweetens it a whole lot!

Click to expand...

Ever play 'Monopoly'? People don't make trades in that game till it's way too late, if ever. Will be the same in PTW.

Catt,

To refocus on the discussion of the two techs that the AI is valuing differently:

These two techs - Bronze Working and Warrior Code -- are bothe 3 factor techs that the AI already holds. I am not researching either tech and neither the AI or my civ have any contacts with other players.

Xerxes is valuing Bronze working at Pottery+Burial+4 gold
while he is valuing Warrior Code at Pottery+Burial+8 gold

my sense is that this difference is a mathematical 10% that results from some unknown calculation factor. Not sure if this is related to the fact that Xerxes had Bronze working for free at the beginning of the game or not.

The 2A factor that you quote may be the significant element in distorting the trade value of Warrior Code up by 10%.

Is it possible it has anything to do with how many other techs have this tech as a prerequisite? I mean, with Bronze Working, you can only get to Iron Working. With Warrior Code, you can get Horseback Riding and Monarchy.

Originally posted by cracker
The 2A factor that you quote may be the significant element in distorting the trade value of Warrior Code up by 10%.

Click to expand...

It's really all just speculation, of course, trying to divine an algorithm with multiple variables based on limited observations and the parsing of off-hand explanations of a programmer in a chat room -- I hesitate to add to the drumbeat of observations coming in from folks simply because they are all anecdotal (I have done no testing). But, perhaps like you, I have seen enough seemingly bizarre trading deals (or offers) that I am convinced there are several additional variables no one has accounted for and that are exposed nowhere in the editor or readmes (seems more pronounced to me under 1.29f, but I also thought it was there in 1.21f). That's why I think it is a daunting task without an explanation from Soren.

Despite my own warning about offering anecdotal evidence, it would be nice to understand why Nationalism, even when widely known, costs an arm and a leg. (Sure, I can guess that its new unit, its new ability to mobilize and draft, and the new diplomatic option of MPPs makes the tech exceedingly "valuable" -- but only for trading apparently, not for editor research values). It raises the (painstakingly difficult to answer) question - what roles do new units, new governements or abilities, etc. play in tech trading value (versus editor value)?

Monarchy, Polytheism and Nationalism all have a high trade value because they HAVE A HIGH COST in comparison with the techs before and around them. For example, Nationalism costs 120, compared with 64 the two most expensive techs up to the point (Magnetism and Theory Of Gravity). Therefore I think it is reasonable to expect them to be valued more by the AI. And now back to the original topic...

Originally posted by cracker
Why is this trading table so imbalanced compared to the tech values we should expect to see?

Are we seeing the double impact of cost factors where it is more expensive for us to research techs and then the AI devalues these techs in trade deals even further by using the AI cost advantage to depreciate any potential trade deal?

Click to expand...

What values do you expect to see? You assume that the AI has a multiplier of 10 for tech valuation and the player uses a product of the inverse cost factor. Try changing the AI cost factor to 1 in the editor and see if this holds true. I tried this on an otherwise unmodified tiny map with 4 civs on chieftain:

I play India starting with Alphabet and Ceremonial Burial. Research to 0 and I soon meet China, with which I can make the following deals:

Offer Ceremonial Burial(2) and 41 gold for Warrior Code(3). China is insulted and refuses.
Offer Ceremonial Burial(2), Alphabet(5) and 23 gold for Masonry(4) and Warrior Code(3). China would happily accept.

Costs to research techs:
Ceremonial Burial : India = 320 gold ; China = 27 gold
Alphabet : India = 800 gold ; China = 67 gold
Warrior Code : India = 400 gold ; China = 48 gold
Masonry : India = 544 gold ; China = 64 gold

Now as you can see in the second trade, I can get a roughly fair deal, so tech valuation is obviously based on one of the research cost sets (the player's or the AI's), but not a combination. The first trade shows that the valuation used must be the human's as if it were China's, 41 gold would more than account for the difference in cost(48-27=21) between ceremonial burial and warrior code. So the AI's valuation of its and your techs is related to the research cost of the human player and hence the AI cost advantage is only counted once. But this still gives no idea as to how the many variables not stated in the editor affect tech valuation or how it devalues as techs become more common.

Originally posted by mydisease
Monarchy, Polytheism and Nationalism all have a high trade value because they HAVE A HIGH COST in comparison with the techs before and around them. For example, Nationalism costs 120, compared with 64 the two most expensive techs up to the point (Magnetism and Theory Of Gravity). Therefore I think it is reasonable to expect them to be valued more by the AI. And now back to the original topic...

Click to expand...

I don't disagree that it is hearsay - or better yet, merely anecdotal evidence subject to the vagaries of human memory.

But comparing Nationalism to Theory of Gravity is a red herring - there is a big step up in cost from Middle Age to Industrial Age technologies. Nationalism has the same editor cost as both Steam Power and Industrialization (120) -- but it is seemingly (remember my caveat about vagaries of human memory ) far, far cheaper to trade for Steam Power or Industrialization than it is for Nationalism -- this despite the fact that with several scientific civs in the game you can generally be assured of not having to pay first civ prices for Nationalism.