Explotive gpt trade with AI [Archive] - Civilization Fanatics' Forums

PaperBeetle

Dec 06, 2007, 12:06 PM

Ah, I could write essays on this strategy/exploit. Indeed I have used it in several GOTMs, and always taken care to note that fact in my spoilers.
It has also had its own threads before. Here (http://forums.civfanatics.com/showthread.php?t=124404) is one. You were there yourself Steve (before you were a mod I think).
There are lots of ways to abuse the AI in civ; this sounds pretty extreme in theory, and really trivializes the game. :p
But later,
I tested out Offa's suggested exploit a bit in GOTM45. Without being spoilerish, I'll just say that in this test, at this point in the game (late AA, early Medieval, everybody still in Despotism), the AI didn't care to give me much at all in gpt deals, even when just given a gpt deal. I only attempted a few deals, since it requires me to give gpt before seeing if the AI would give me any gpt back, and the results were so poor that I cut it off very quickly. I may test a bit more later on in the game.
Of course you were right the first time. If one has the stomach to use this to its fullest extent (e.g. pulling some backwards Easter Island-esque civ all the way from the AA to the modern) it is a virtually infinite supply of cash.

I remember at least one other discussion of it in the GOTM fora, several pages long, and Offa using the phrase "egregious exploit". But I couldn't find that one :hmm:.

Anyway, I think we certainly need a definite ruling before the new SG starts. Team Regular used the tech trick in the last SG. Of course, playing in Feudalism it was of little use, but it probably saved us a couple of turns due to paying for our military while we kept research cranked up. If we had been assiduous about building only cheap units and upgrading them, it could have been a different story.
But the last SG also demonstrated why this stratploit is so hard to rule against. Every time you come up with a way to ban it, or set a limit to it (e.g. no giving the AI uneccessary gpt - I tried that as a self-imposed limit but it just rewards you for ruining your gpt rep), there is also a perfectly common situation where it happens without you even notice. In SG12 we had a lot of small AI who we wanted to invade. A natural strategy is to buy e.g. the Aztecs to attack e.g. the Japs several turns before you land your own troops. And if you don't have a tech, you would use gpt to buy that alliance. Some turns later you have a new tech, so you want to sell it to the Aztecs. They give you gpt for it, thank you. Some turns later, the Japanese are gone, and the Aztecs are still stumping up that gpt. Was it an exploit or was it fair diplomacy? :dunno:

Anyway, back to the present. I would absolutely support a ban if you can find some formulation that still allows us more diplomatic felxibility than, well, Civ4.
I have used it my current game. Continue / abort / kill my victim and finish cleanly?

PaperBeetle

Dec 06, 2007, 12:29 PM

I've always been intrigued by how a set of fairly reasonable rules could give rise to such a potentially powerful exploit. After all, the trick does not actually cheat the AI. It still gets what it pays for, according to its own beaker-cost valuation. Some problems are:
(a) the AI's willingness to buy whatever you offer it.
(b) the simplicity of tech valuation. Perhaps a tech should be cheaper depending on the use a civ can get from it, a bit like resource valuations. Probably that creates a whole different set of imbalances.
(c) the "world bank's" readiness to underwrite civs for a relatively small fee. Would the system work better if negative treasuries were allowed?
(d) the invariant value of the gp relative to the shield and the beaker. If the world bank's generosity started causing inflation, excessive tech tricking could trash the world economy. :)

I. Larkin

Dec 06, 2007, 01:45 PM

How's this: You can't force the AI into double negative. That should be clear. If you're getting more than they have, then it's banned in the HoF.

I don't pretend to speak for the Gotm/Sgotm staff, but call my name an I'll appear, eventually. :devil:

Now, how might you know what they can afford? There's this 999 Trick (http://forums.civfanatics.com/showthread.php?s=&threadid=82464) If you can see what they have beforehand, then ask for more, well there's no ambiguity there.

If you can hurt the AI within the rules, please do, I hate the AIs, I'm not trying to defend them, but if someone does this, everyone better because it's too powerful not to. People compain about the game being too formulaic now, if everyone has to do this, then what?
In first approximation "yes" if we speak about Ernsworth.
But there are many situations when
1) it is more risky
2) Extendet in time.
3) more sophysticated conditions.
999 gpt may be varied. Say after GA.
What about following formula: (please, polish my Englis)
No gpt deal with AI must not include gpt gift that exceed 2 (my bet) or (1/4)(Steve bet) 999 information gpt?
Or may be we will find compromise as 1?

civ_steve

Dec 06, 2007, 02:56 PM

If I understand the 999 trick thread, the 999 result is the maximum gpt the AI can produce, 0 Science and 0 Luxuries, 100% commerce to taxes.

If I understand your proposal, I.Larkin, 2*(999 result) is completely in the negative gpt category. 1*(999) would leave the AI with no Science and no Luxury tax ability.

I figure that the AI might have 50% or higher for Science, and this gpt is the target, so I suggested 25% as being a somewhat safe amount to assure that the Luxury Tax is available to the AI, and some Science, too. After all, this trade takes advantage of the AI's programming not to cancel it.

It would be great to uncover a setting that would make the AI declare war on you if they were paying a big gpt and running negative. I might test that a bit as well. :D

PaperBeetle - thanks for finding that link! :goodjob: I was thinking of that discussion and I couldn't find it.

I. Larkin

Dec 06, 2007, 03:13 PM

Devil's advocate!!
Minimum I can agree 75%. (after tough negotiation 50%). After all we should have a chanse to ruin economy of strong AIs. Also, it is a risk like "settler on Galley". Niklas at SGOTM make "alternative possibility". I don't like it because it has "deliberate action" is unclear formulation. Nobody can prove that you knew that or not. But my formulation is clear. I think anybody understand what "gift" mean.
(the gpt you paied above after tough negotiation). IF MA costs a lot it self economy will be ruined anyway.

Niklas

Dec 06, 2007, 04:31 PM

Reposting from the SGOTM13 thread:
The player is not allowed to actively cancel a deal that gives gpt to an AI if the player is at the same time receiving gpt from that same AI in another deal that would not be cancelled.
I still prefer my formulation to Ivan's, because 2x or 1/4x 999 gpt is still a lot of money you could get for free. And it only deals with the Emsworth double-negative gpt issues, the deliberate tricking of an AI, and leaves all other potentially legal gpt deals scot free.

I really don't see the issue. Is it exploitish to get something for nothing? Of course it is. And that's exactly what you're doing here. And it's not getting better by the fact that you can effectively cripple the AI at the same time, making them easy targets for later invasion. That's two exploits in one gift wrapper, thanks a lot! In other words,
After all we should have a chanse to ruin economy of strong AIs.
No. Nonononono. This is no less an exploit than the free gold is. Systematically crippling an AI by tricking him into paying more than he can sustain, by abusing the way the AIs are programmed, is definitely an exploit.

There are of course corner cases, like the one in SGOTM12 that PaperBeetle pointed out. Which is why the question of intention is the pressing point. And it's not like there isn't precedence for using intention, the palace rank exploit is one specific exploit that is very much a grey zone but which we seem to be able to handle just fine. I don't see why this should be any different.

Don't do it if it's for the sole purpose of getting free money and/or a crippled AI. But if it happens due to unforseen circumstances, most likely it won't have a very profound effect on the game.

To summarize:

Is this an exploit? Yes. Definitely yes. And doubly yes, both the free gold and the crippling of AIs are exploits.
Can we ban exploits retroactively? Of course, that's the whole point of that grey-area clause in the rules.
Should we ban this? Of course, since it's an exploit. It doesn't help that it has been up for discussion before and deemed ok, since obviously the potential impact wasn't realized in those discussions. Lord Emsworth has shown us the extent of the problem, and so we must act.
Should games using this be accepted for this COTM? I'm very tempted to say h##l no, I don't want to compete against such games, but I realize the trouble in making such a harsh judgement. So I guess it comes down to a matter of size. How much did the exploit help in this particular game? If the effects were limited at best, then allow it. If it made the game much much easier, then exclude it. It might be un-fun to have your game excluded, but it's very much un-fun not being able to compete because someone else used an exploit.
Can we use intention as a premise in the ruling? Yes we can, it has been done before to good effect.

I. Larkin

Dec 06, 2007, 04:55 PM

Niklas, I disagree with this absolutely and even don't want to put detaild analysis.
Why for example make gpt and ask "remove or declare" allowed?
Or take "everething for gpt with MA" and kill?
It is more unfair.
At discussed exploit human player at least give something.

Ivan

Lanzelot

Dec 06, 2007, 05:07 PM

Here's my feedback on that topic.

I used this trick twice so far:

in GOTM71 (Deity level), where it gave me ca. 5000gpt. (I described it in the spoiler.) However, it had only been possible at a very late point, when I was already winning the game anyway. The trick requires to have a big lead in technology, so you have something to sell to all (or many) other civs.
in GOTM72, where it gave me a few hundred gpt. (See my spoiler again.) Even though this game was Warlord, where it should be much easier to pull this trick, I again only managed to set it up at a time, when the game was won anyway.

So first of all I'm not sure, how important this exploit/strategy is after all: at least at the higher difficulty levels it will probably be possible only after you are on the road to victory anyway. So it may speed up the victory a bit, but it won't let you win a badly played game.

And second, I don't feel it's an exploit: basically there are two ways of making it happen:

by interrupting a connected resource. This will give you a big reputation hit, so you will only be able to do it once, and you won't be able to trade for any resources for the rest of the game. So you are punished for doing it, therefore it's not really an exploit.
By killing another AI, thus ending the millitary alliances you have made against that AI. But setting up something like this on the higher difficulty levels requires quite a bit of skillful play: you need to be able to destroy an AI (if you end the military alliance by just signing peace, you'll get the rep hit again) and you need to get the other civs to signing an alliance with you.

In any case the trick only works well, if you are well ahead in tech, as you need something that you can sell. And the AIs can't complain that they didn't get anything for their money. For example in GOTM71 I sold half the industrial age to 5 or 6 civs! So as PaperBeetle already said, they got what they paid for! If some time later they can't afford it anymore and need to sell buildings and units in order to be able to pay their debts, is that my problem? This could as well happen in the "normal course of the game". Just consider this: an AI has a decent amount of gpt just from their own "natural" income. You sell them a couple of techs for this gpt. Then a few turns later an enemy invades that AI, takes a couple of their cities which had banks/marketplaces and additionally deprives them of two or three luxuries. As a consequence they loose a substantial part of their "natural" income and in addition have to increase their luxury slider because of the lost lux resources. And after that they are no longer able to pay the gpt they owe me. You can't possibly ban/prevent something like that...?!

So my opinion is: I wouldn't call it an exploit. Rather "skillful diplomacy"... :D
And it's not always possible. For example during COTM42 the opportunity never arose, even though I kept looking for it.

Cheers, Lanzelot

Chamnix

Dec 06, 2007, 05:31 PM

To me, the bigger problem of the Emsworth Agreements is not that you cripple the AI, but that you are manufacturing free gold out of nothing. You can legally cripple the AI to varying extents through RoP abuse, declaring war on a Sid AI that is isolated, or even buying a worker 10 turns into the game.

The big problem with the Emsworth Agreements to me is that they create free money out of nothing. The money that the player makes on a single turn can be far greater than the total amount of commerce produced by all the citizens of the world combined. None of the other potential exploits named by anyone so far in this thread have that effect - in several of them you can take the AI's money and essentially give nothing in return, by in no legal instance are you adding money to the world's economy from nowhere.

And second, I don't feel it's an exploit: basically there are two ways of making it happen:

...
By killing another AI, thus ending the millitary alliances you have made against that AI. But setting up something like this on the higher difficulty levels requires quite a bit of skillful play...

Not really - in almost every game you are going to destroy an AI at some point.

Niklas

Dec 06, 2007, 06:09 PM

Niklas, I disagree with this absolutely and even don't want to put detaild analysis.
If you can't bother to do an analysis, how could we discuss it? It's not like I'm going to accept "because I say so" as a decent argument in favor.

Why for example make gpt and ask "remove or declare" allowed?
Or take "everething for gpt with MA" and kill?
It is more unfair.
It is unfair in your treatment of the AI perhaps, but not unfair in that you fabricate gold out of nothing, just like Chamnix said. And I should point out that I have at least once made a substantial gpt deal with an AI and asked him to "remove or declare", only to have him choose the former option. As for taking "everything for gpt with MA and kill", isn't that an exploit too? At least I know I would never use it. It's not listed under acceptable exploits at least, so maybe we should bring that up to discussion as well?

At discussed exploit human player at least give something.
You are not giving anything that is even remotely proportional to what you gain. To say that they get what they pay for is quite frankly bollocks, it's like a door salesman selling things "for free" (but of course with huge loans tied to them) to a senile old woman. Technically she gets what she pays for, but the point is she couldn't pay for it in the first place.

This could as well happen in the "normal course of the game". Just consider this: an AI has a decent amount of gpt just from their own "natural" income. You sell them a couple of techs for this gpt. Then a few turns later an enemy invades that AI, takes a couple of their cities which had banks/marketplaces and additionally deprives them of two or three luxuries. As a consequence they loose a substantial part of their "natural" income and in addition have to increase their luxury slider because of the lost lux resources. And after that they are no longer able to pay the gpt they owe me. You can't possibly ban/prevent something like that...?!
Of course not. This is not an exploit, since it is impossible to exploit it. You can't know in advance that some other AI is going to invade your trade partner, and you wouldn't have traded differently if you knew. And in particular, you the player don't gain more money than you would have if that other AI hadn't invaded! There's no way you could exploit it, because it isn't an exploit. The analogy in the case of Emsworth Agreements is that you of course gain much much more if you cancel the gpt deal going to the AI than if you don't. And it's all in the hands of the player, so it is fully exploitable. I don't see the problem differentiating the two.

civ_steve

Dec 06, 2007, 06:38 PM

Lanzelot - I had seen your post in the GOTM71 spoiler, and that initially sent me off looking for the thread with the earlier negative gpt discussion (that PaperBeetle found!) With a Deity level game in the Industrial Era, you can get a lot of natural gpt from the AI doing Tech trading, so I'm not sure if you forced a negative gpt situation on them. (I'll have to reread your spoiler and perhaps do some more investigation).

One thing I'm clear about is intentionally forcing the AI into negative gpt will not be allowed. This is a situation where the AI ends up paying you more than the 999 amount. Things happen in the course of the game that may unintentionally exploit the AI to some degree, and that's part of the uncertainty of the game and is allowable.

The issue that's open is if a limited form of manipulation by gpt trading can or should be allowed to give the player another tool in controlling the AI (like other tools that are allowed.) I don't think it's correct to allow complete crippling of the AI economy, because that's potentially so powerful it would become required to be competitive. Is there an acceptable amount of gpt trading manipulation that benefits a player through skillful use without becoming massively abusive?

Another reason to limit or eliminate the manipulative gpt trading is because the reputation hit you so justly deserve can be avoided.

Niklas

Dec 06, 2007, 06:51 PM

The issue that's open is if a limited form of manipulation by gpt trading can or should be allowed to give the player another tool in controlling the AI (like other tools that are allowed.) I don't think it's correct to allow complete crippling of the AI economy, because that's potentially so powerful it would become required to be competitive. Is there an acceptable amount of gpt trading manipulation that benefits a player through skillful use without becoming massively abusive?
This completely misses the other point, which Chamnix pointed out as the more important one, the free gold that you invariably generate through the use of this exploit. I don't see how putting a limit to the amount of gold you get from the AI could ever solve that.

I recall games where one AI has fully legally given me over 400 gpt, so clearly a limit cannot be lower than that. But 400 gpt gotten through this exploit is still such a huge benefit that it would be impossible to compete against such a game without using the same exploit.

To tell the truth I'm quite baffled how some could see this not an exploit. I have yet to see a reasonable argument for why it isn't. There have been reasonable arguments regarding corner cases, but I've already argued why I don't think those are a big deal. Please, those who want to see this allowed, tell me why!?

AutomatedTeller

Dec 06, 2007, 10:50 PM

I think the 999 gives the total pre-corruption, not the total available.

I have never used this - but it is quite possible to get into a state where you are making just silly amounts. The game that Lord Emsworth talked about was a 150 AD 100K victory, where he ended up with something like 4M gold, making 590K gpt.

Elear

Dec 07, 2007, 12:36 AM

civ_steve

Dec 07, 2007, 01:02 AM

This completely misses the other point, which Chamnix pointed out as the more important one, the free gold that you invariably generate through the use of this exploit. ...
Chamnix refers to the free gold generated out of thin air; in other words, the gold due to forcing the AI into negative gpt. I'm referring to the total amount of gpt available to the AI. I have gained this gold many times by giving the AI the tech it's currently researching; once that gpt has been freed up, the AI will trade the full amount of available gpt, up to the total amount it's willing to trade for the Tech. In the case we're discussing, the player gains some of that gpt by giving the AI gpt through a trade, then breaking the deal afterwards. As long as the first gift to the AI is less than the amount the AI is spending on research, these two approaches are somewhat similar, except in my trade the AI gets another Tech, I get 100% of the available AI gpt, and I get no hit to my reputation.

I recall games where one AI has fully legally given me over 400 gpt, so clearly a limit cannot be lower than that. ...
It has to be relative to the AI's current gpt situation, which is why the 999 trick is interesting. I'd like to tie down exactly the amount it shows.

To tell the truth I'm quite baffled how some could see this not an exploit. I have yet to see a reasonable argument for why it isn't. There have been reasonable arguments regarding corner cases, but I've already argued why I don't think those are a big deal. Please, those who want to see this allowed, tell me why!?There are many ways to get additional gpt from the AI - I've mentioned the Tech trade/gift earlier. Getting the AI's gpt is probably OK; pushing them to negative gpt is definitely not. I have no problems with the wording you proposed in the SGOTM13 thread; that would be very clear cut. I'm just keeping the thread open to see what the community feeling is.

Paul#42

Dec 07, 2007, 03:46 AM

The player is not allowed to actively cancel a deal that gives gpt to an AI if the player is at the same time receiving gpt from that same AI in another deal that would not be cancelled.
I like this rule very much and think it covers all important aspects. :thumbsup:

Paul#42

Dec 07, 2007, 03:52 AM

At discussed exploit human player at least give something.
But it generates money that is not in the game.
The AI probably runs huge deficits and gets away with losing a single building or unit every turn.
That's an indirect exploition of the banned running deficit at insufficient cash exploit. :old:

Lanzelot

Dec 07, 2007, 05:00 AM

Not really - in almost every game you are going to destroy an AI at some point.

Well, desroying an AI is not enough. As I said, during the entire COTM42 I was not able to set up an Emsworth Deal, even though I destroyed a couple of AIs. Because two more prerequisites need to be fullfilled:

You need to have something valuable that you can sell to the other AIs (e.g. a tech lead)
The other AIs must be willing to sign a millitary alliance with you (which they won't do, if they are at war with you...)

In COTM42 it never happened that all three prerequisites were fullfilled at the same time, so I insist: the Emsworth trick requires a bit of careful preparation.

The player is not allowed to actively cancel a deal that gives gpt to an AI if the player is at the same time receiving gpt from that same AI in another deal that would not be cancelled.

Actually I don't like this rule too much, because in my opinion it has two defects:

It also would rule out deals, which would not push the AI into negative income. And if I understood the essence of this thread correctly, then most people agree this to be the exploit: generating gold out of thin air by pushing the AI into negative. But if I can rip the AI off their "normal" income by canceling a deal, why should that be prohibited?
It would not cover the way I exploited the AI in my GOTM72 game: in that game I had noticed that America, Aztekia and China had joined up against poor Arabia. Arabia was loosing like two cities per turn, and when they were down to one or two cities, I expected them to be gone pretty soon... At that point I signed alliances with America, Aztekia and China for all the gold I had and then sold my tech lead to them. Indeed about two turns later the Aztecs took the last Arabian city and from then on I was making about 500 extra gpt... :D This would not be covered by Niklas' rule, because I did not cancel any deal! The Aztecs did it, so it's their own fault. If they wanted to have my money keep coming in, they shouldn't have taken the last Arabian city...!

Instead I would propose a rule like this:
"A player is not allowed to pay an AI more than the AI demands."

So for example if the AI is willing to sign a military alliance with you for 10gpt, then you should not be allowed to pay them 1000gpt for that military alliance.
If you really want to ban the Emsworth Agreements without ruling out other diplomatic tricks that might be perfectly non-exploitative, then it would probably require a rule along these lines (perhaps combined with the 999-feature to determine the point when an AI goes negative?!).

However, one more thought about whether the "Emsworth Agreement" is really exploitative or not: if I understood Lord Emsworth's writeup correctly, then it was not the "plain Emsworth Agreements" which gave him such an absurd income, it was the fact that he applied these agreements "iteratively".
What I mean is: he used the gpt that he gained by such an agreement to setup a second Emsworth Agreement with even higher sums! For example, if he starts with a natural income of 100gpt and gifts these to 4 AIs, he will end up with 500gpt. Now if he still has enough stuff to sell (and another victim that can be destroyed), he can set up a second round using the 500gpt as starting point, and that will give him 2500 gpt. This is what will lead to exponential growth, which I readily admit is exploitative.

So perhaps a compromise would be: "Plain Emsworth Agreements" are allowed, but "Iterative Emsworth Agreements" are not, so you need to wait the full 20 turns before you are allowed to set it up again?!

This would eliminate the absurd sums you get from the exponential growth, and it would eliminate the problem that "free money is generated from thin air", because with one such deal the AI may not yet be in negative gpt, or perhaps only in like minus 50-100gpt, and that's not yet really generating money from thin air, because the AI pays for that by dissolving units and city improvements. (I can see that minus 1000gpt would be exploitative, because one unit/building is never worth that much. But to me it seems that loosing a unit/building per turn is an appropriate punishment for running minus 50gpt, so I have no problem with that.)

Cheers, Lanzelot

archphoenix

Dec 07, 2007, 05:47 AM

Instead I would propose a rule like this:
"A player is not allowed to pay an AI more than the AI demands."

I think this way should be the greatly solution, simple and clear.

WackenOpenAir

Dec 07, 2007, 06:22 AM

"The player is not allowed to actively cancel a deal that gives gpt to an AI if the player is at the same time receiving gpt from that same AI in another deal that would not be cancelled."
I like this rule very much and think it covers all important aspects. :thumbsup:

Yes, just keep it at that. (of course, the word actively includes any method that is planned by the player)
And if anyone finds a way around the rules, it should still count as abuse and is banned. We are gamers, not politicians.

Lanzelot

Dec 07, 2007, 06:25 AM

By the way: what is so bad about generating money from thin air? That's how the "real world" works...! :crazyeye: Most of the money in this world only exists as a "virtual" number in some bank's accounting books or as a virtual number on the scoreboard of some stock market... If you'd sum up all these virtual numbers and then compare it to the number of "physically existing" coins and bills, you'd be quite surprised!

Only joking,
Lanzelot

Niklas

Dec 07, 2007, 06:42 AM

It also would rule out deals, which would not push the AI into negative income. And if I understood the essence of this thread correctly, then most people agree this to be the exploit: generating gold out of thin air by pushing the AI into negative. But if I can rip the AI off their "normal" income by canceling a deal, why should that be prohibited?
I would still consider it an exploit even if the AI can afford it, but definitely not on the same scale.
It would not cover the way I exploited the AI in my GOTM72 game: in that game I had noticed that America, Aztekia and China had joined up against poor Arabia. Arabia was loosing like two cities per turn, and when they were down to one or two cities, I expected them to be gone pretty soon... At that point I signed alliances with America, Aztekia and China for all the gold I had and then sold my tech lead to them. Indeed about two turns later the Aztecs took the last Arabian city and from then on I was making about 500 extra gpt... :D This would not be covered by Niklas' rule, because I did not cancel any deal! The Aztecs did it, so it's their own fault. If they wanted to have my money keep coming in, they shouldn't have taken the last Arabian city...!
In some sense this could be argued to be perfectly acceptable, though I don't think it should be. But it is not as bad an exploit as the Emsworth Agreement simply because it is (almost) impossble for the player to engineer such a situation. If Emsworth Agreements are allowed, you would be forced to play that way in order to be competitive. You don't have that problem here. In some sense it is comparable to signing gpt deals and then asking for a "remove or declare", since it requires the AI to give you the opportunity in the first place. But yes, it happens often enough for it to be exploitish, and you are right that we should probably find a definition that covers this case as well.

Instead I would propose a rule like this:
"A player is not allowed to pay an AI more than the AI demands."
I like this idea, good thinking. Though it still isn't perfect. If the AI has a (monopoly) tech that you don't have, and you have a (monopoly) tech that AI lacks, you could easily trade for his tech using gpt and a luxury, then sell him your tech for gpt, and finally disconnect the luxury. Sure you destroy your trade reputation, but the gain would far outweigh that cost. And if you can do it with an MA (harder to set up of course) and kill, you don't even get that rep hit. And you are not breaking the rule you propose.

So how about a combination? If we put a big fat AND between our two rules we get something that captures all cases proposed, but it could also potentially generate a few "false positives", as you point out in your first point. I'm not sure what a "legally" ripping the AI of their "normal" income would be like though. Could you give an example that you don't consider exploitish? If it still involves making him pay more than he would normally have for something, I don't see how it could not be an exploit.

I can see two ways of composition. Either we simply concatenate the rules, like so:
The player is not allowed to actively cancel a deal that gives gpt to an AI if the player is at the same time receiving gpt from that same AI in another deal that would not be cancelled. Also, a player is not allowed to pay an AI more than the AI demands.
Or we tie it all to the deliberate act of actually taking the gpt from the AI, which would be my preference. It might be possible to turn the original deal around, like so:
The player is not allowed to trade for gpt from an AI if he there is an active gpt deal from the player to that AI that the player knows will be cancelled very shortly.
But clearly this would yield too many false positives, there are certainly very legal situations that would break the above. So to temper it a bit further, how about:
The player is not allowed to trade for gpt from an AI if he there is an active gpt deal from the player to that AI that the player knows will be cancelled very shortly, unless the player leaves the AI with at least as much free gpt as the value of the deal that will be cancelled.

How about it? Very convoluted sentence I know, but we can worry about the formulation after we decide on the actual contents of the rule. I'm not sure I'm perfectly happy with either proposed rule here (first or third), but both of them are IMO better than my first proposal or Lanzelot's proposal. And any proposal is of course much better than nothing.

On a side note, I see that the exploits listed on the GOTM site (http://gotm.civfanatics.net/games/rules.php) are not formulated like this at all, but rather as a (loose) description of a situation that would be an exploit. Using such a formulation here, we might have something like:

Emsworth Agreements
It is possible to make an AI pay much more gold per turn than he normally has available for something by
giving him gold per turn through a deal that includes a luxury or a military alliance
trading back that gold per turn for techs
break the luxury route or the alliance, either deliberately or through circumstances known to be about to happen.
This forces the AI to pay you gold per turn that he wouldn't have otherwise had, sometimes even forcing him into deficit spending. This is not allowed.
To me that would be the best formulation of all.

However, one more thought about whether the "Emsworth Agreement" is really exploitative or not: if I understood Lord Emsworth's writeup correctly, then it was not the "plain Emsworth Agreements" which gave him such an absurd income, it was the fact that he applied these agreements "iteratively".
What I mean is: he used the gpt that he gained by such an agreement to setup a second Emsworth Agreement with even higher sums! For example, if he starts with a natural income of 100gpt and gifts these to 4 AIs, he will end up with 500gpt. Now if he still has enough stuff to sell (and another victim that can be destroyed), he can set up a second round using the 500gpt as starting point, and that will give him 2500 gpt. This is what will lead to exponential growth, which I readily admit is exploitative.

So perhaps a compromise would be: "Plain Emsworth Agreements" are allowed, but "Iterative Emsworth Agreements" are not, so you need to wait the full 20 turns before you are allowed to set it up again?!

This would eliminate the absurd sums you get from the exponential growth, and it would eliminate the problem that "free money is generated from thin air", because with one such deal the AI may not yet be in negative gpt, or perhaps only in like minus 50-100gpt, and that's not yet really generating money from thin air, because the AI pays for that by dissolving units and city improvements.
No, I don't like this reasoning at all. 500 gpt is still an incredible amount of money "for free", you are still crippling the AI beyond the point he would normally have been willing to go. Likely you are still forcing him into negative spending, but even if you don't you are still crippling him and making him a much easier target for yourself.

(I can see that minus 1000gpt would be exploitative, because one unit/building is never worth that much. But to me it seems that loosing a unit/building per turn is an appropriate punishment for running minus 50gpt, so I have no problem with that.)
No way. We have banned it as an exploit if the player runs a large deficit and "only" loses a unit/building per turn. In this situation we're forcing it onto the AI so we don't even take the hit ourselves, even though we are the once benefitting from it. If the AI chose on its own volition to go into negative spending, then by our own rule he would be using an exploit. In this situation he doesn't even get a choice.

Niklas

Dec 07, 2007, 06:44 AM

Yes, just keep it at that. (of course, the word actively includes any method that is planned by the player)
And if anyone finds a way around the rules, it should still count as abuse and is banned. We are gamers, not politicians.
Put it that way and it certainly makes sense, and would cover all kinds of tricks you might pull. Good thinking WackenOpenAir!

I. Larkin

Dec 07, 2007, 07:32 AM

This completely misses the other point, which Chamnix pointed out as the more important one, the free gold that you invariably generate through the use of this exploit. I don't see how putting a limit to the amount of gold you get from the AI could ever solve that.

So what? It is so many things in civ that not happend in real word...

I. Larkin

Dec 07, 2007, 07:41 AM

But it generates money that is not in the game.
The AI probably runs huge deficits and gets away with losing a single building or unit every turn.
That's an indirect exploition of the banned running deficit at insufficient cash exploit. :old:
1) see post above.
2) Cashe defecite banned for human, not for AI

PaperBeetle

Dec 07, 2007, 07:50 AM

In principle I support Lanze's suggestion, "A player is not allowed to pay an AI more than the AI demands," and oppose Niklas's, for the reasons Lanze gives. To put it another way, the 'corner cases' that Niklas didn't mind would nevertheless be banned by the rule "The player is not allowed to actively cancel a deal that gives gpt to an AI if the player is at the same time receiving gpt from that same AI in another deal that would not be cancelled."
However I think "A player is not allowed to pay an AI more than the AI demands" also needs to be backed up by an explicit ban on buying hard goods for gpt, followed by a premature termination of that gpt deal. This is because once you have ruined your per-turn rep, buying an alliance + world map / chump change really can cost you hundreds of gpt.
But then that would extend to banning tech/cash for gpt followed by a dow on the recipient of the gpt. Which strategy has a long and illustrious history as an accepted GOTM practice. Not easy is it? :(

Chamnix

Dec 07, 2007, 08:13 AM

Emsworth Agreements
It is possible to make an AI pay much more gold per turn than he normally has available for something by

giving him gold per turn through a deal that includes a luxury or a military alliance
trading back that gold per turn for techs
break the luxury route or the alliance, either deliberately or through circumstances known to be about to happen.
This forces the AI to pay you gold per turn that he wouldn't have otherwise had, sometimes even forcing him into deficit spending. This is not allowed.

And if anyone finds a way around the rules, it should still count as abuse and is banned. We are gamers, not politicians.

:thumbsup::thumbsup:

PaperBeetle

Dec 07, 2007, 08:36 AM

This one sounded good at first:
The player is not allowed to trade for gpt from an AI if he there is an active gpt deal from the player to that AI that the player knows will be cancelled very shortly, unless the player leaves the AI with at least as much free gpt as the value of the deal that will be cancelled.
But actually I'm not sure it says much more than the original version. The AI will almost always have less spare gpt than the you gave for the alliance, because when you give it gpt it will just crank up its science slider to take advantage. Perhaps the 999 trick is of interest here?
Ultimately I am hoping for a formulation which allows for the Aztec-Japan-Byzantine deal I described from SG12 in the first page of this thread.

I guess this practice originally arose out of frustration about the lopsided nature of tech trading with the AI. When we buy a tech from them, they require payment for its full value, and we are expected to raise our tax slider if we need the gpt to make the deal. When they buy a tech from us, they pay only the minimum of (1) its full value and (2) their spare cash / gpt / tech / whatever else they have lying around. In its least exploitative form, this is a method for getting the AI to change its tax slider to meet a fair market price.

Marsden

Dec 07, 2007, 09:56 AM

I think the 999 gives the total pre-corruption, not the total available.

Yes, the big green number you get on your own domestic advisor screen. I do this all the time just to see how bad my economy is compared to an AI that's beating me to techs.

http://i77.photobucket.com/albums/j61/dac18643/composite.jpg

The income after taxmen, interest, and from other civs. (Does the ai ever use taxmen?) So taking that whole number away puts them in a serious crunch, but at least they have it. Taking more makes money out of nowhere. Do it enough and it could accelerate the tech rate because 100% research shouldn't be a problem getting tons of cash.

It's important to check it before you give them any because whatever you are giving will be added in(unless, of course, you aren't planning to cut the previous, then this whole discusion is unnecessary)

No this isn't easy. It's very skillful and tricky to do. No one said it's easy. But because it's hard to pull off doesn't mean it's not exploitive.

I. Larkin

Dec 07, 2007, 10:33 AM

It's important to check it before you give them any because whatever you are giving will be added in(unless, of course, you aren't planning to cut the previous, then this whole discusion is unnecessary)

So do you mean that we are not allowed to cut previous??

Paul#42

Dec 07, 2007, 10:51 AM

So do you mean that we are not allowed to cut previous??

Well, you are - and indirectly could be forced if resources are involved and cut - but you should not take more gpt than initially available.

However I still see a problem with the A player is not allowed to pay an AI more than the AI demands.
Sometimes I want to give the AI gold for free - to make them happy, to increase their research...

We will not find a rule that is completely idiot proof.
We have to find a simple rule and judge all connected cases in the spirit of the game...

If we fail, we could still make the human player gift the gpt back every turn... :groucho: :hide:

I. Larkin

Dec 07, 2007, 11:25 AM

In first approximation "yes" if we speak about Ernsworth.
But there are many situations when
1) it is more risky
2) Extendet in time.
3) more sophysticated conditions.
999 gpt may be varied. Say after GA.
What about following formula: (please, polish my Englis)
No gpt deal with AI must not include gpt gift that exceed 100-200% (my bet) or 25% (Steve bet) 999 information gpt or 0% (Lanzelot bet)?
Or may be we will find compromise as 1?
Nobody reply for this, that I edited something. Acording Lancelot 0%, we may say that plain gifts are allowed (non gpt) or plain gpt that does not bound with anything else.

Othniel

Dec 07, 2007, 02:04 PM

Some of the main "rule propositions" floated out there in the last two pages of this thread.
The player is not allowed to trade for gpt from an AI if he there is an active gpt deal from the player to that AI that the player knows will be cancelled very shortly, unless the player leaves the AI with at least as much free gpt as the value of the deal that will be cancelled.

"The player is not allowed to actively cancel a deal that gives gpt to an AI if the player is at the same time receiving gpt from that same AI in another deal that would not be cancelled." [emphasis mine]

Another slight variation on the scenarios thrown out so far is this:

We pull of an "Emsworth Agreement" with an AI by doing the following.
- Give an AI some gpt in exchange for an MA against another strong civ
- Trade back for your original gpt by giving the AI a tech or some other hard good

Let's say also that we didn't pay the AI an excessive amount of gpt for that deal, but rather simply paid the normal market price in gpt to get the MA.

In this case, we're not "expecting" the MA deal to end prematurely because the strong civ we're allied against is not in danger of being eliminated within 20 turns. However, we're all familiar with the AI passion for signing peace deals before their MA deals expire. In the very possible event that the AI does this--and trashes their own rep--we've just gotten some free gpt for no rep hit ourselves.

I don't consider the scenario I described to be exploitative. It's up to the AI to cancel or not--we're passive--and the AI very possibly will not cancel the deal. The power is in their hands. Plus, we've given them a tech so we didn't get something for nothing. Of course, a similar argument could be made for MA deals against a dying civ when the AI does the killing stroke. Again, we'd be "passive", although I think we'd know that deal is more likely to be canceled than in my scenario. Really, it's a slippery slope of "passive cancel" or "more passive cancel" and there is no distinct line.

Anyhow, I don't think my scenario is prohibited by the first statement I quoted above, but it might be prohibited by the second quote. This scenario is also another time-honored strat, although that doesn't preempt it from charges of being an exploit. Would / should this be banned too?

Summary:
This whole game loophole is like the English language; for every rule there are 10 notable exceptions :crazyeye:. I hope I don't confuse the matter by bringing in another corner scenario.

Basically, I agree with the rules I quoted above as long as we spell out some common exceptions that are allowed, such as the one I described above (assuming it is allowed ;)).

Niklas

Dec 07, 2007, 02:42 PM

I don't see the scenario presented by Othniel as a problem, since it is not the player cancelling the deal, or even expecting the deal to end. And it would not be ruled out by either of the statements above.

But really, I think focusing on words is a waste of time. For comparison, here's the rule for the rank palace exploit:
Rank Corruption Exploit
It is possible to move your Palace far away from the main core of your civilization, leaving your Forbidden Palace near its center. The result is that the rank component of the corruption calculation for cities nearer to the Forbidden Palace than to the Palace reduces to rank 1. The Conquests version of CivIII has been modified to eliminate this exploit. This thread discussed the ruling, which states that "You must not rebuild the palace in a location remote from the majority of your empire in order to gain a significant corruption advantage".
What is the majority of your empire? What is a significant advantage? Or even, what is "in order to", which denotes intention? These are impossible to define in a way that puts every single case either completely in the red or completely in the green. There have been situations where it is questionable whether the rule has been violated as formulated, but without the intention part we would rather accept than exclude. I recall there was some discussion regarding a GOTM as Carthage some time back where we started at a peninsula separated from everyone by a massive (!) mountain area. You would naturally capture Mecca on the other side and jump your palace there. Most everyone did, but some did the jump earlier than others, clearly reducing corruption in their original cores. But since the main reason for the palace jump was indeed to set up a new core, and not "in order to reduce corruption", no complaints were made against anyone and all games were accepted.

My point is we don't need, or indeed want, a rule that exactly ties down the cases into the green and red areas. As long as a trade is done within the spirit of the rules, there is no need for draconic rulings. And if we leave the rule similarly vague like the rank corruption exploit, the admins can make informed rulings from case to case. And you should of course always PM the admins to ask whether a particular trade is ok or not, just like the rules currently state.

So, the more I think about it the more I favor a formulation like:
Emsworth Agreements
It is possible to make an AI pay much more gold per turn than he normally has available for something by
giving him gold per turn through a deal that includes a luxury or a military alliance
trading back that gold per turn for techs
break the luxury route or the alliance, either deliberately or through circumstances known to be about to happen.
This forces the AI to pay you gold per turn that he wouldn't have otherwise had, sometimes even forcing him into deficit spending. This is not allowed.
Chamnix has already +1'ed it, what do the rest think? In particular those who would be inclined to allow certain scenarios, what's your take on this?

I guess this practice originally arose out of frustration about the lopsided nature of tech trading with the AI. When we buy a tech from them, they require payment for its full value, and we are expected to raise our tax slider if we need the gpt to make the deal. When they buy a tech from us, they pay only the minimum of (1) its full value and (2) their spare cash / gpt / tech / whatever else they have lying around. In its least exploitative form, this is a method for getting the AI to change its tax slider to meet a fair market price.

There is of course a very good reason why the AI is programmed this way, namely it cannot possibly make an informed judgement on whether reducing your tax slider in order to afford something is a good idea or not, which of course the player can. It will happily pay for whatever it can afford, and if we "force" it to afford more it will pay more. That's to me the whole nature of the exploit, to circumvent this "safety check" in the programming, and thus I see your musing as even further "proof" that this is indeed an exploit. We cannot expect the AI to play by the same rules as the player does, it's an AI!

Nobody reply for this, that I edited something. Acording Lancelot 0%, we may say that plain gifts are allowed (non gpt) or plain gpt that does not bound with anything else.
I replied to it in the broad sense, stating that I don't like any ruling that includes a reference to 999 gpt. This is not an issue about amounts, it is an issue about a particular practice. It's the practice we should rule out, and then it won't matter if it's 1 gpt or 100k gpt.

I. Larkin

Dec 07, 2007, 03:05 PM

Emsworth Agreements
It is possible to make an AI pay much more gold per turn than he normally has available for something by
giving him excessive gold per turn through a deal that includes a luxury or a military alliance
trading back that gold per turn for techs
break the luxury route or the alliance, either deliberately or through circumstances known to be about to happen.
This forces the AI to pay you gold per turn that he wouldn't have otherwise had, sometimes even forcing him into deficit spending. This is not allowed.
essential part of Ernswort is excessive gold. I never thought that it is exploit to pay gpt for MA and then trade them tech. I insist on "excessive" that may be in Lantzelot interpritation >0. Also "about to be happend". Yo may finish in 1, 2, 3 turns... In last GOTM all war with Babylon took 4 turns, as Deals lasts 20 I could gain something anyway.
Why tech is forbiden and any thinlike oter Lux, MPP, MAs, not??

Othniel

Dec 07, 2007, 03:42 PM

Emsworth Agreements
It is possible to make an AI pay much more gold per turn than he normally has available for something by

* giving him gold per turn through a deal that includes a luxury or a military alliance
* trading back that gold per turn for techs
* break the luxury route or the alliance, either deliberately or through circumstances known to be about to happen.

This forces the AI to pay you gold per turn that he wouldn't have otherwise had, sometimes even forcing him into deficit spending. This is not allowed.

Niklas, I like this...for the most part. ;)

The part I don't really like is the "break the luxury route". IMHO, if you can get gpt for FREE e.g. use the MA tricks, that practice is an exploit and should be banned. But, if you have to suffer an irreparable rep hit e.g. break a lux route, then the game has penalized you for being a deal-breaker and this practice should be allowed.

Grayer areas for me include using boot orders or spies to try and force a war declaration from the civ your trading with. IMO, they're grayer than the MA trick because the AI has some measure of control on deciding to declare or not. I'm probably fine either way on those, banned or not.

The major problem, of course, with allowing the breaking of luxury deals, etc, is that we introduce the possibility of forcing the AI into a double negative situation if we break a "really big" deal. If we don't want to allow a possible double-negative, then we have to use the 999 trick...and I know you, Niklas, and others are not if favor of having to use that trick.

So, where does that leave us? Well, again, I generally like Niklas' proposal. :thumbsup:

I think it still needs a little modification, or at least a general acceptance of the weaknesses it will contain.
Two Options I see:
- We ban certain kinds of trade breaking (that the game already penalizes with a rep hit) to avoid any possible double-negative scenarios
- We allow these rep-hit deals but have to introduce a more tedious rule and process through using the 999 trick

Option 2 is a little more tedious, but I favor it because it gives the player more control over their trade reputation.

There, that is my opinion. (I heartily reserve the right to change said opinion once it has been picked apart by people much smarter than me. :p)

AutomatedTeller

Dec 07, 2007, 04:50 PM

I'm unclear as to what is being banned here:

Are we banning putting the AI into a double-negative situation? Or are we banning making deals which screw the AI?

For example:

I am Persia in the late MA. I trade dyes and 100 GPT to Germany for a military alliance vs. Russia, who has 1 city left.

I then trade astronomy to germany for 100 GPT.

I then destroy Russia.

Which, if any, of these situations are legal?

A) Germany shows 22 gpt using 999 (chieftain)
B) Germany shows 300 gpt using 999 (SID)
C) Germany shows 175 gpt using 999 (SID)
D) Germany shows 125 gpt using 999 (SID)

I think we are outlawing A) and allowing B) Various proposals would outlaw C and D, too.

Niklas

Dec 07, 2007, 05:07 PM

AT, in my opinion we are definitely outlawing A through D. I see the exploit not necessarily tied to forcing the AI into negative spending, but to exploiting the way the AI is programmed to make deals that screw it over. Of course, either of B through D would be allowed if Germany had 100 gpt to trade you before you signed the MA against Russia.

For what it's worth I don't know what the 999 trick is. I've seen various references to it and even links but I haven't bothered looking. It's not something I want to use, and definitely not something I want to be forced to use if I am to be competitive.

Othniel, I don't think breaking your reputation is in any way proportional to the potential gain. In some games, notably archipelago games like COTM43, it will take a very long time for the AIs to meet each other and spill the word on your broken rep, so you could do the dirty deal over and over again. Yes, in the majority of all games that is not the case. That doesn't make it any less an exploit in the cases where it is. COTM43 is a perfect example of this.

For the grey areas you are talking about, I don't think they have anything to do with the case we are discussing. If you force a war declaration from the AI you would cancel all gpt deals, so you would no longer be taking his gold. Taking all you can get upfront from an AI using gpt as payment, and then declaring, that's a different thing IMO. You might be able to gain almost as much in that case, but you would not be crippling the AI.

Chamnix

Dec 07, 2007, 06:04 PM

I like Niklas' method of generally describing the exploit as opposed to trying to give details for every situation, but I think Lanzelot's rule is a good thing to use as a "baseline".

A player is not allowed to pay an AI more than the AI demands.

This is not to say that every time you give more than the AI demands it is an exploit (for example, you may want to increase their research or discourage an aggressive neighbor from attacking). It also doesn't mean that giving no more than what the AI demands is never an exploit (for example, if your rep is "partially blown" and demands huge gpt that you know you will be able to get back, or in Niklas' example of swapping monopoly techs for gpt but severing your gpt deal), but I think it is a very good guideline to start with if you are trying to determine what is exploitive.

Any time you are giving the AI more than it demands, you should look at your motives. Why are you giving away cash? If the intent is to get your gpt back and then sever the first deal, it should be banned.

In AutomatedTeller's example, A through D would be banned if you could get Germany to sign an alliance for less than 100 gpt. Why are you giving them more than you have to? If it is not to try to get more money back than Germany can afford, why are you doing it?

In PaperBeetle's example from SGOTM 12, I don't see a problem as long as you didn't pay the Aztecs more than necessary to get the alliance.

Othniel

Dec 07, 2007, 06:13 PM

I see the exploit not necessarily tied to forcing the AI into negative spending, but to exploiting the way the AI is programmed to make deals that screw it over. [.....]

For the grey areas you are talking about, I don't think they have anything to do with the case we are discussing. If you force a war declaration from the AI you would cancel all gpt deals, so you would no longer be taking his gold. Taking all you can get upfront from an AI using gpt as payment, and then declaring, that's a different thing IMO. You might be able to gain almost as much in that case, but you would not be crippling the AI.

What you said in response to AT is exactly why I included those grey areas.

If our goal is to prevent deals that screw over the AI, I think we have to look at tactics like boot orders and spy declarations. For instance, I can make a gpt deal for a tech, perhaps a valuable monoploy tech, and then try and force the AI to declare through boot orders. If I succeed, I just got a tech for free and ripped off the AI.

Again, I personally have no problem with using these boot orders and such. But IMHO we can't deny that we're not hosing the AI when we succeed with them.

Othniel, I don't think breaking your reputation is in any way proportional to the potential gain. In some games, notably archipelago games like COTM43, it will take a very long time for the AIs to meet each other and spill the word on your broken rep, so you could do the dirty deal over and over again. Yes, in the majority of all games that is not the case. That doesn't make it any less an exploit in the cases where it is. COTM43 is a perfect example of this.

I know that I'm a decent Deity player, not a world-elite player. As such, especially on SID, I have a hard time imagining me being able to substantially rip off the AI and then killing that civ before my dirty rep becomes known to the world at large. Perhaps better players can truly do that...

I know we have an advantage in COTM 43 with our super-curraghs able to travel on ocean, but it's still going to be hard to kill civs very quickly (unless it's a minor AI).

I just think that 95%+ of the time you never have a chance to substantially rip off an isolated AI before you rep is broadcast to the world at large. I have a hard time stomaching a rule for just those few instances where the stars line up and you can go a on a dirty deal spree.

If I'm wrong on how often that can be accomplished, then I would gladly second your opinion, Niklas. :)

Othniel

Dec 07, 2007, 07:04 PM

I like Niklas' method of generally describing the exploit as opposed to trying to give details for every situation...

I do too. I agree with keeping the rule as a general statement because that way it's so much easier to handle.

The two things I'm trying to accomplish without muddying the waters too much:
- refine the general proposal to say what we want it to say
- outline some various scenarios that may or may not be banned by our general proposal

The scenarios and grey areas I have described in previous posts are some of the strategies that might be banned by this rule, but it's not clear-cut. My belief is that these strategies are quite common, and if that's the case, other people will also have questions about their legality if we adopt the general rule. If that's the case, using PM to decide each case looks like an administrative nightmare for our volunteer admins.

I wonder if we might save time and headaches by making some corollaries to our general rule, outlining the legality of some common strategies as they relate to our general rule.

For instance:

Emsworth Agreements
It is possible to make an AI pay much more gold per turn than he normally has available for something by

* giving him gold per turn through a deal that includes a luxury or a military alliance
* trading back that gold per turn for techs
* break the luxury route or the alliance, either deliberately or through circumstances known to be about to happen.

This forces the AI to pay you gold per turn that he wouldn't have otherwise had, sometimes even forcing him into deficit spending. This is not allowed.

Corollaries:
Legal: Using boot orders to attempt to get the AI to cancel a gpt deal
Illegal: Signing an MA, using the process outlined above, against a civ about to be destroyed.

P.S. sorry for the double-post

PaperBeetle

Dec 07, 2007, 08:32 PM

With respect to breaking your rep, I have to agree with Niklas that the potential gain is much greater than that particular penalty. I believe it is possible to e.g. gift your neighbour saltpeter + 1000gpt, sell him a load of medieval techs for 1000gpt, cut the 'peter, rinse and repeat 20 turns later. After all, your broken per-turn rep doesn't worry the AI when it is recieving only hard goods in a deal, and nor does is bother them when you are giving them a per-turn gift. So that's certainly the same exploit, just a much easier version to execute.

In response to Ivan's question about selling stuff other than techs, yes certainly the ban should apply to anything which can be sold without the human losing its use, such as ROPs and resources, but not hard cash or workers. We just tend to discuss it in relation to techs because they are by far the most valuable tradeable items in the game.
I understand that a very similar exploit exists in Civ4, but is restricted to resource sales because of that game's ban on hire purchase.

civ_steve

Dec 08, 2007, 04:11 AM

Lots of good discussion. I think my key concern (when abusing the AI) is that the player not force a negative gpt situation on the AI. We've talked a lot about the gifting of gpt then taking the gift away. There's also a concern about MA's, and even if the player accepts the AI's demand, the demand could be so large that a potential negative gpt situation could occur. As I was thinking about this, and how we've been looking at the player's actions, perhaps we can look at the end result for a more general rule.

I'm thinking, perhaps we allow the player to make any trade desired. Freely. :) However, if any deal where the player is providing gpt to the AI ends prematurely, the player must immediately compare the amount of gpt he/she is receiving to the current 999 value of the AI. If that ratio is above a given amount, the trade is considered exploitive and the player must immediately restore the ratio, either through gpt gift or through gifting enough gold to make up the difference over the remaining term that they are getting the gold.

If that ratio were 1:1, this would allow the player to effectively make trades that snagged up to half of the total gpt available to that AI. If it were 1:2, the player could snag up to one third of the total gpt available to that AI. The advantage of this approach is that even MA's that break down (or any other agreement made for whatever reason) would cause an adjustment by the player to avoid totally crippling the AI.

Looking at Marsden's F1 screen, and treating it as if it were the AI, an allowed ratio of 1:1 would allow the player to set up whatever trade they wanted. If they gifted/traded gpt to the Marsden AI, then traded them Nationalism for gpt, then broke the original trade, they'd have to check to make sure they were getting no more than 389 gpt, leaving 389 gpt to the Marsden AI to cover their normal costs. The AI's expenses are 339 gpt (Corr + Maint + Unit ), so they are not forced into a negative gpt situation. Having gotten Nationalism 11 turns earlier, this may or may not be a good trade for the AI. And the AI would probably not offer 389 gpt at any rate.

If the ratio were 1:2, the player, assuming they broke an earlier trade/gift of gpt, could only be taking in 259 gpt; anything more would require an adjustment. This would leave 519 gpt, enough to cover expenses and still have nearly 200 gpt available to the AI. In this case, the AI gets Nationalism 11 turns early, and again 259 gpt is likely still more than the AI would offer.

One other feature of this approach is it would take the cumulative effect of whatever earlier gpt deals the player may have in place with the AI.

If this sounds reasonable, I think the 1:2 ratio is about what the AI might trade anyway. 1:1 might allow the player to gain some advantage, and probably not force negative gpt.

Niklas

Dec 08, 2007, 06:31 AM

I can only state again that I would be unhappy if we came up with any kind of rule that involves the 999 in any way. It is very possible to exploit the AI and gain plenty of gpt "for free" without getting close to the 999. It wouldn't be creating any gold "out of the air", but it would still be powerful enough that it would be hard to be competitive without using the strategy. I want the practice itself outlawed, and I don't see why we couldn't. And I definitely don't want the 999 trick to be required.

If our goal is to prevent deals that screw over the AI, I think we have to look at tactics like boot orders and spy declarations. For instance, I can make a gpt deal for a tech, perhaps a valuable monoploy tech, and then try and force the AI to declare through boot orders. If I succeed, I just got a tech for free and ripped off the AI.
I think we should be very clear on the difference between screwing the AI in a deal and screwing the "system". We have trade reps for a reason, and it should be totally allowed to "screw" the AI by for instance signing a gpt deal for tech and then declaring. I also don't see it as a problem to force a boot order to retain your trade rep, that's the AI's fault for declaring on you. In none of these cases does the AI really lose anything (other than cash or possibly workers), you gain a tech but you haven't locked the AI into a gpt payment that it cannot sustain.

By screwing the system, I mean tricking the AI into giving up more than it normally would, or even more than it has. By giving or trading the AI gpt, you are "by speculation" increasing the gpt available, bloating the market. It's the speculation that should be outlawed, nothing else.

That said, I think your idea of naming corollaries is a good one. :)

I know that I'm a decent Deity player, not a world-elite player. As such, especially on SID, I have a hard time imagining me being able to substantially rip off the AI and then killing that civ before my dirty rep becomes known to the world at large. Perhaps better players can truly do that...
Like PaperBeetle said, your broken rep isn't really a big problem, you can do it again and again anyway. But what I meant was not that you would kill off civs before your broken rep is known, rather that you can do the trade with each AI, since one AI doesn't know that you screwed another.

Say you have 20 gpt and a luxury available. Give that to AI A who doesn't know anyone else, and sell them a tech they don't have to get that gpt back. Cut the luxury, then reconnect it. Now you have 40 gpt. Give that to AI B and sell them a tech they don't have. Cut the luxury. Etc etc. Actually, like PaperBeetle says, you can do this even if your rep is known. What you couldn't do is to use your gpt to buy a tech you don't have, but that's really a minor problem.

AutomatedTeller

Dec 08, 2007, 10:42 AM

Well, I don't see any reason to think that you have to use the 999 trick if you never try to screw the AI.

I mean, all these deals are ways to exploit the AI - if all you ever do is make trades and not try to screw the AI, you never have to check the 999.

Conceptually, you could run into a problem like this for a turn or two with the AI - say you buy steam power for 600 gpt then get electricity for free but screw up and don't sell it back to the AI until the next turn, for 580 gpt. Then there will be a turn where the AI might be running a big deficit, but I don't think anyone would see that as an exploit.

civ_steve

Dec 08, 2007, 11:23 AM

It is very possible to exploit the AI and gain plenty of gpt "for free" without getting close to the 999. It wouldn't be creating any gold "out of the air", but it would still be powerful enough that it would be hard to be competitive without using the strategy.
I think we should be very clear on the difference between screwing the AI in a deal and screwing the "system". We have trade reps for a reason, and it should be totally allowed to "screw" the AI by for instance signing a gpt deal for tech and then declaring.I see very little difference between these two trades. In some ways the first trade, if limited, is LESS exploitive than the second. In the first, if limited, the AI gets a Tech and since I don't think they'll ever trade more for a Tech then they think its worth, they'll pay what they think is equal value, or less. And the player, in exchange for a possible trade rep hit, has gotten more of what the AI values the Tech for in exchange. In the second the AI gets NOTHING, and the player gets a free tech. This just manufactured an amount of free gold equal to the player's cost to research the Tech. Since we have always allowed the 'buy a Tech from the AI for gpt, then cut the payment' tactic, getting 'free gold' is not a good enough argument to ban the first trade.
By screwing the system, I mean tricking the AI into giving up more than it normally would, or even more than it has. By giving or trading the AI gpt, you are "by speculation" increasing the gpt available, bloating the market. It's the speculation that should be outlawed, nothing else. That's the first step of the manipulative gpt trade, and I thought everyone feels that part was OK. It's really the 3rd step, when the gift/trade of gpt is broken, where the exploit can occur. The problem is there are lots of ways to get from 1 to 3 without being exploitive, or that aren't covered in any of the rules discussions; even the spirit of the law may be clear to most but vague :) to others. So I'm proposing a comparison that looks just at Step 3, and if we can come to a general agreement on how much of the total gpt available to the AI can be snagged, we can avoid causing a negative gpt and crippling of the AI.
Say you have 20 gpt and a luxury available. Give that to AI A who doesn't know anyone else, and sell them a tech they don't have to get that gpt back. Cut the luxury, then reconnect it. Now you have 40 gpt. Give that to AI B and sell them a tech they don't have. Cut the luxury. Etc etc. I think this is fine as long as the AI doesn't pay more than it thinks the Tech is worth to it (and I don't believe it ever does that), and you don't take more gpt than the AI can afford. Checking the 999 number and applying a generic ratio is not a sure thing, but I believe it can generally prevent the negative gpt, AI crippling situation, and give the player a guideline on how far they can go.
I mean, all these deals are ways to exploit the AI - if all you ever do is make trades and not try to screw the AI, you never have to check the 999.
Yes! You only check 999 if a trade where you provide gpt to the AI is ended prematurely (through whatever means.) If this never occurs, you never have to check 999.

Elear

Dec 08, 2007, 11:39 AM

Yes! You only check 999 if a trade where you provide gpt to the AI is ended prematurely (through whatever means.) If this never occurs, you never have to check 999.

Precisely. :) The only acceptable way to terminate a GPT deal early (in my opinion) is to declare war, or get them to declare war. War has its own consequences.

-Elear

Lanzelot

Dec 08, 2007, 11:50 AM

At the moment it looks like there are two parties here: the ones who want to ban Emsworth Agreements altogether (as described by Niklas) and the ones who just want to ban those Emsworth Agreements that criple the AI "too much" (as described by civ_steve). I would like to add a few more ideas to the current discussion (let me use the abbreviation "EA" for Emsworth Agreement in the following):

Some have said, that if we allowed EAs, then everybody, who wants to be a competitive player, will need to use them. Well, what is so bad about that? These days everybody, who wants to be competitive, needs to know about the "tech broker strategy", about 4-turn settler factories or palace jumps and many more strategems. EAs would enrich our arsenal of strategic weapons and would make the trading and diplomacy aspect of the game more interesting/colorful. If EAs are well documented in the strategy forum, everybody can use them. I don't have a problem with that. This would be a point in favor of civ_steve's idea.
However I don't like the idea of 999-checking, calculating ratios and determining how much I have to pay back to keep a clean consciousness... :) But if that can't be avoided, so be it.
Please consider the following scenario: It is perfectly legal to buy an expensive tech for gtp and then declare war on the AI (or provoke it into declaring on you). Let's assume you are planning to do this, and you need a certain tech very badly, but you are not yet in a position to declare immediately, because your defensive troops are not yet ready. So you buy the tech now, pay 200 gpt for the first two turns, move the last remaining troops into position and then declare at the end of turn 2 (so you can at least save the remaining 18 payments...)
That would be perfectly legal as well, wouldn't it? But in the meantime your victim might have made a couple of deals with other AIs, for example spending your 200gpt for a few other techs. So when you declare, you might drive him into negative spending!

What I'm trying to point out here: I don't see a big difference between a) canceling a deal in which you give gpt to the AI while another deal is active in which that AI gives gpt to you and b) canceling a deal in which you give gpt to the AI while no other deal with that AI is active. In both cases you are stealing the same amount of money from the AI! And in my opinion buying an expensive tech and then not giving them the money which is rightfully theirs, is more "morally reprehensible" than giving them money for a MA and then no longer paying it when the MA is no longer active...

I'm even sure, that a world-class player would be able to plan for such a scheme. For example he checks CivAssistII and sees that civ A is very poor and that civs B, C and D all have an expensive tech, which civ A does not yet have. He buys another expensive tech from civ A for a large gpt amount and then the next turn he checks CivAssistII again and sees that civ A now has both techs from civs C and D. Most certainly civ A has now spent the money it got the previous turn, and now the player declares on civ A... :D
Something like this may be very powerful: you get a tech almost for free, civ A will be substantially cripled and you may even be able to capture a couple of towns very easily, because they will not be able to put up a decent fight because of their negative income and because of loosing a unit per turn...
The above can even be achieved without risking a war against civ A. For example just include a resource or a trade embargo against civ E in the tech deal and then disconnect the resource or make a trade with civ E... (I believe trade embargos can be broken that way, and it will probably have the same effect of canceling the entire deal?!)

There may be many more variations on this theme, and how can we decide for each of them, which one is legal and which one an exploit? I would say: "The situation is hopeless, but not serious..."

Cheers, Lanzelot

PS: I think the root of the entire dilemma is the fact that deals are "bound together" to a single unit. This looks like a bug to me. For example what I don't see is: if you buy a tech from the AI and pay with a resource and an amount of gpt and then cancel the resource part of the deal, why the gpt part of the deal should be canceled too. If we could fix that and have every element of a deal been treated as a single independent item, the problem would be gone. Do you think we could get the source code from Fireaxis? After all they are no longer fixing and selling it. And from what I've seen on this forum here, it appears that some of you have really close contacts to some Fireaxis developers...?! We could make a patch and require it as prerequisite for the GOTMs.

I. Larkin

Dec 08, 2007, 01:33 PM

Yes! You only check 999 if a trade where you provide gpt to the AI is ended prematurely (through whatever means.*) If this never occurs, you never have to check 999.
I like this idea. (Just thought to offer myself.) And, if it have happend human player may (or must(??)), (for his choise)
a) Restore gpt payment as it was before "bound deal" prematurely ended
or
b) gift AI gpt to ensure net gpt from AI less or equal to 999 information.

* If AI quit from MA (not human quit or kill ) human player not responcible for this actoin.
(we try the best not to screw "system", but If system...)

PS: I think the root of the entire dilemma is the fact that deals are "bound together" to a single unit.
due to this "bug" civ 3 much better then civ 4. Lot of features you may find if one bound things with "peace treaty". I discribed some in my article "War happiness works!"

Niklas

Dec 08, 2007, 02:21 PM

Well, I don't see any reason to think that you have to use the 999 trick if you never try to screw the AI.

I mean, all these deals are ways to exploit the AI - if all you ever do is make trades and not try to screw the AI, you never have to check the 999.
That might be fine if the "screwing" wasn't so powerful in the first place. If these kinds of agreements are allowed, there's no we we could keep from doing it if we want to stay competitive. So if it's allowed, no one will keep from screwing the AI.

Conceptually, you could run into a problem like this for a turn or two with the AI - say you buy steam power for 600 gpt then get electricity for free but screw up and don't sell it back to the AI until the next turn, for 580 gpt. Then there will be a turn where the AI might be running a big deficit, but I don't think anyone would see that as an exploit.
Indeed I wouldn't. It's not the AI's deficit alone that makes it an exploit, you the player must gain from it as well to make it exploitish. In this case you are still paying 600 gpt x20, so you're not gaining anything "for free".

I see very little difference between these two trades. In some ways the first trade, if limited, is LESS exploitive than the second. In the first, if limited, the AI gets a Tech and since I don't think they'll ever trade more for a Tech then they think its worth, they'll pay what they think is equal value, or less. And the player, in exchange for a possible trade rep hit, has gotten more of what the AI values the Tech for in exchange. In the second the AI gets NOTHING, and the player gets a free tech. This just manufactured an amount of free gold equal to the player's cost to research the Tech. Since we have always allowed the 'buy a Tech from the AI for gpt, then cut the payment' tactic, getting 'free gold' is not a good enough argument to ban the first trade.
The problem in this case is that the AI is so incredibly stupid that it will buy whatever it can afford (or sometimes not afford). No they won't pay more than what they think its value is, but the AI simply cannot make such valuations. They will happily pay tons of gpt for Music Theory even when Bach's is completed elsewhere. And by taking their research gpt (which is effectively what you do when you "free up" more of his available gpt) he will also have lost the ability to research something else. My previous comment about door salesmen and senile old women still stands.

Also, turn it and see what the AI loses. In the first case the AI loses nothing (except the ability to trade that tech to you), in the second case the AI loses a lot of gpt, potentially crippling it.

That's the first step of the manipulative gpt trade, and I thought everyone feels that part was OK. It's really the 3rd step, when the gift/trade of gpt is broken, where the exploit can occur. The problem is there are lots of ways to get from 1 to 3 without being exploitive, or that aren't covered in any of the rules discussions; even the spirit of the law may be clear to most but vague :) to others. So I'm proposing a comparison that looks just at Step 3, and if we can come to a general agreement on how much of the total gpt available to the AI can be snagged, we can avoid causing a negative gpt and crippling of the AI.
It is really the combination of the steps that creates the exploit. I don't think cutting a luxury deal is an exploit, but it would be if you received bloated gpt by cutting it. That's what I mean by speculation - the combination of artificially increasing the AI's available gpt and then trading for it. If you don't cut the trade route or alliance, there's no speculation, so the first step would be ok. If you haven't bloated the AI's available gpt, again there's no speculation, so the third step would be ok.

I think this is fine as long as the AI doesn't pay more than it thinks the Tech is worth to it (and I don't believe it ever does that), and you don't take more gpt than the AI can afford. Checking the 999 number and applying a generic ratio is not a sure thing, but I believe it can generally prevent the negative gpt, AI crippling situation, and give the player a guideline on how far they can go.
See again the comment on stupidity in AI tech valuation.

At the moment it looks like there are two parties here: the ones who want to ban Emsworth Agreements altogether (as described by Niklas) and the ones who just want to ban those Emsworth Agreements that criple the AI "too much" (as described by civ_steve). I would like to add a few more ideas to the current discussion (let me use the abbreviation "EA" for Emsworth Agreement in the following):

Some have said, that if we allowed EAs, then everybody, who wants to be a competitive player, will need to use them. Well, what is so bad about that? These days everybody, who wants to be competitive, needs to know about the "tech broker strategy", about 4-turn settler factories or palace jumps and many more strategems. EAs would enrich our arsenal of strategic weapons and would make the trading and diplomacy aspect of the game more interesting/colorful. If EAs are well documented in the strategy forum, everybody can use them. I don't have a problem with that. This would be a point in favor of civ_steve's idea.
Actually, that's a very good point, but...
However I don't like the idea of 999-checking, calculating ratios and determining how much I have to pay back to keep a clean consciousness... :) But if that can't be avoided, so be it.
... I really don't like this part of it. I simply don't think we can find a limit that would be suitable, since even at low ratios you can potentially create incredible amounts of gold, and cripple the AI.

I did some digging on the 999 trick (so yeah, now I know it, how quaint) and it seems to me it contains a lot of cans of worms. First of all, just so we're clear what we're discussing, as Marsden pointed out the value listed is the Big Green Number (BGN). That means the total of all beakers, gold, happy faces and corrupted commerce that your towns generate each turn, plus income from taxmen (but not clowns and scientists), gpt from other civs and Wall Street interest.

A first thing to note about this number is that it is not fixed but depends on the slider settings. A civ with lots of libraries but no markets has a higher BGN running a high research than a low research. The lowest value is of course when running 100% luxuries since there are no multipliers for that. Sampling a few of my own games, the difference between 100% research and 100% taxes varies 5-15% of the (highest) BGN (20% of the lowest). Technically it could go higher, but that's unlikely. But considering a Sid AI who has Literature but not Currency (and builds libs ridiculously fast), the value could potentially vary close to 20% (of the highest BGN).

Now, what would be a fair number for the AI to have left? Again sampling some of my own games, the value I would have available for trading, that wouldn't make me start losing units or make all my towns riot, varies greatly. In some games it is as high as 70% (0% lux rate, high income from other civs), in others it's as low as below 30% (high lux rate, lots of units OR a large gpt deal to an AI). And these are clearly not extreme values since I just sampled a few saves I had at random. And furthermore, I rarely (never) play as extremely as the AI does, building all improvements it can in all towns and building lots of units as long as it can afford them. And paying gpt deals to the player or other AIs, since those expenses are counted after the BGN. A redeemeing factor though is that the AI rarely uses a high lux rate, preferring clowns for some reason (at least that's my experience).

So what would be a fair number? I think 25% of the BGN is too much, given that even I in my fairly moderate unit-focused moments can go below 30% (and that's of the lowest BGN, meaning it could be more if we compare to the visible BGN - it just takes one library for the values to differ - though in my sample those values were equal). And it just takes one gpt deal paying off most of the available gpt to some other AI for the actual ratio to drop dramatically, closing in on 0%.

So to sum up my thoughts on this - if we are to set a limit, it has to be 0%. And what would be the point of it then? Then we would outlaw plenty of "fair" cases, like PaperBeetle's Byz-Aztec-Japs scenario, or when an AI stupidly puts himself in the situation by signing peace and breaking an MA that you have would preferred to retain. So I don't see how the BGN/999 trick could help us out here.

I stand by my previous opinion, that we need to outlaw the practice and not try to impose limits on it.

Please consider the following scenario: It is perfectly legal to buy an expensive tech for gtp and then declare war on the AI (or provoke it into declaring on you). Let's assume you are planning to do this, and you need a certain tech very badly, but you are not yet in a position to declare immediately, because your defensive troops are not yet ready. So you buy the tech now, pay 200 gpt for the first two turns, move the last remaining troops into position and then declare at the end of turn 2 (so you can at least save the remaining 18 payments...)
That would be perfectly legal as well, wouldn't it? But in the meantime your victim might have made a couple of deals with other AIs, for example spending your 200gpt for a few other techs. So when you declare, you might drive him into negative spending!
Yes, this certainly becomes a problem. But at least a redeeming factor is that the "free" gold won't go to the player but to other AIs. But of course you could then set up gpt deals with them and get it via proxy... So yes, I agree this is a problematic case. Maybe we should outlaw it as well? Meaning that if you want to declare and cancel a gpt deal, you had better do it immediately.

What I'm trying to point out here: I don't see a big difference between a) canceling a deal in which you give gpt to the AI while another deal is active in which that AI gives gpt to you and b) canceling a deal in which you give gpt to the AI while no other deal with that AI is active. In both cases you are stealing the same amount of money from the AI! And in my opinion buying an expensive tech and then not giving them the money which is rightfully theirs, is more "morally reprehensible" than giving them money for a MA and then no longer paying it when the MA is no longer active...
I do see a big difference, in that you don't actually take something from the AI in case (b) except virtual (promised) money, not actual money as in case (a). But as you pointed out above, that can indeed be bad enough, if he uses that virtual money to sign deals with others.

PS: I think the root of the entire dilemma is the fact that deals are "bound together" to a single unit. This looks like a bug to me. For example what I don't see is: if you buy a tech from the AI and pay with a resource and an amount of gpt and then cancel the resource part of the deal, why the gpt part of the deal should be canceled too. If we could fix that and have every element of a deal been treated as a single independent item, the problem would be gone. Do you think we could get the source code from Fireaxis? After all they are no longer fixing and selling it. And from what I've seen on this forum here, it appears that some of you have really close contacts to some Fireaxis developers...?! We could make a patch and require it as prerequisite for the GOTMs.
I doubt we'll get the source code or any some such. But even so, it's not really a bug. For instance if you were paying gpt+luxury for a luxury from the AI, and cut your own luxury, do you still think the whole trade shouldn't be cut?

AutomatedTeller

Dec 08, 2007, 03:22 PM

oh - I had thought that the 999 number was the base commerce and wouldn't change, but it clearly isn't - given that, I would agree that there is little chance of knowing what the actual amount is.

I've never seen the AI move the lux slider above 0. I think the reason it uses clowns is because score is figured on happiness, so clowns will maximize score.

Beorn-eL-Feared

Dec 08, 2007, 08:13 PM

Whatever the wording, I strongly think this needs a ban. Not only for the gold spurring from the trees part, for the intent as well.

civ_steve

Dec 10, 2007, 01:30 AM

I've had a lot of family engagements this weekend, so I'm a little delayed in getting back to this topic.

Lanzelot has pointed out a possible strategy where the player could be providing gpt to Civ A, and manipulated it so that he's receiving the gpt back through another civ. Result is the player could cut the gpt going to Civ A, still gain the gpt but not have direct gpt payment from the civ getting crippled. I've also been getting PM's about other similar manipulative strategies, and there's been some GOTM Staff discussion going on as well. A general rule to apply here is very difficult to formulate; there are very many similar tactics, and some have been acceptable for a long time.

Key to the rule must be that the AI not be forced into a negative gpt status through intentional actions of the player, or through actions or lack of actions that the player knows are highly likely to establish the negative gpt status. This is likely caused due to gpt being provided to a Civ by the player, and that gpt then being cut by whatever reason. I'm coming around to the view that when the player makes deals providing gpt to a civ, they must be made with the intention of seeing them through. One exception - when an AI is moving units into attack position and the player buys a Tech from the AI for gpt; however, the player may not make a Remove Demand in this case; the AI has to Declare War freely.

I'm thinking that a player providing any significant amount of gpt to an AI Civ should have an idea of that Civ's 999 number. If the gpt is cut, the earlier 999 number should be compared to the current value; if it's ratio is higher than a certain value, the player must restore gpt to balance it, or gold to equalize the difference. Note - in this case the player might not be receiving any gpt from that civ!

Also, the player is not responsible for an AI going into negative gpt through interactions with other AI, assuming the player isn't attempting to manipulate this status. Under no circumstances may a player attempt this to occur.

I'm thinking the ratio might be about 2:1; if the post-gpt-cut 999 value is less than half of the pre-gpt-cut value, the player must restore to that ratio. This is just a generic estimated value, to give the player some responsibility for cases where the gpt is cut through non-intentional and non anticipated actions. The player still must intend to carry the gpt trade for 20 turns when made.

Just some more thoughts - are we getting any closer?

Lanzelot

Dec 10, 2007, 11:21 AM

Just some more thoughts - are we getting any closer?

No, I think we are going in circles...

Ok, seriously, here's some more input: during my preparation for COTM43 I've been studying SirPleb's excellent article about his Sid level HOF game. Here I stumbled over a very interesting piece of information, which has been reported by SirPleb and confirmed by Arathorn. Here is a summary of the findings in my own words. The original information is a bit scattered across the entire thread.

SirPleb noted that at some point in his game the AIs were no longer able to offer him gpt for techs. Later he attributed this to the fact that the AIs were running a negative income because of their excessive amount of units, for example an AI with 18 cities had a military force of 1300 units, which would require something like 1000gpt just for unit upkeep:

I think this is a new way for the AI to self-destruct. A cancerous sort of military growth peculiar to Sid level. I suspect that because the production bonus is so high, the AI has built every possible improvement in all 18 of those cities. And then rather than produce wealth it keeps on producing military? Since it is isolated on an island, it never gets into a war which would prune that military. Eventually I expect it would over-run its ability to support itself. In the meantime its science rate is probably getting slower and slower.

Arathorn added some more information he gained by spying out AI cities:

The huge AI armies is why they never have gpt to pay for techs or luxuries or whatever, but it is NOT a huge detriment to their ability to research. Recall that the only penalty for having a negative income is the loss of one building and one unit, which is very easy to replace with Sid discounts. My guess is that, if they were checked, the AI would have very large negative income every turn but would still be researching at a reasonable clip (60% or so). Whenever I've investigated a Sid city, even when I *know* they're beyond broke, they have a high level of research going.

So it looks like even without "evil manipulation" by a human player the AI on Sid level sometimes goes into negative income all by itself!! And even on a large scale: they run at minus 1000gpt, pay for it with one unit (which is no problem for an AI that easily produces 10 or more new units each turn...) and happily research away at 60%... These observations have a couple of implications:

At Sid the AI is doing heavily, what the GOTM rules have forbidden: running a large negative income. So shouldn't it be allowed to the human player, at least on Sid?!
They may even be doing it on their own account, whithout being "Emsworth manipulated" by the human player, so can EAs really be called an exploit?! And can we allow something for Sid and ban it for the other levels?
If an AI is running such a large deficit, and we are gifting it let's say 500gpt, they will still be broke, so we still won't be able to sell them anything for gpt... That means on Sid EAs may be worthless...!
I see a problem with the following proposition by civ_steve:
I'm thinking that a player providing any significant amount of gpt to an AI Civ should have an idea of that Civ's 999 number. If the gpt is cut, the earlier 999 number should be compared to the current value; if it's ratio is higher than a certain value, the player must restore gpt to balance it, or gold to equalize the difference.
Let's assume the 999-method shows something like 50gpt for an AI. We make a deal with them which includes 100gpt from our side. Shortly after that the AI decides, it does no longer care for having a positive income, and pushes the tech slider up to 60% and starts producing like 50 units per turn. A couple of turns later they are at minus 1000gpt and then our deal gets canceled (non-intentionally). We check their 999-value again and get a big shock... :eek: This rule would then make us pay for their entire unit upkeep and research expenses! This can't be your intention, civ_steve... :lol:

I think it's becoming more and more hopeless, the longer we think about it... Why not simply reward the human player's "creativity and inventivness" and give us as much freedom in our trading with the AI as possible?
If we just put a rule into effect that prevents exponential growth via "iterative EAs", most of the sting would be taken out of the exploit. I think, we should setup a clear-cut definition of what EAs are, like the one given by Niklas, perhaps modified to include trade embargos as follows:

Emsworth Agreements
It is possible to make an AI pay much more gold per turn than he normally has available by

giving him gold per turn through a deal that includes a resource or a military alliance or a trade embargo
trading back that gold per turn for techs in a second deal
breaking the first deal by disconnecting the resource route or breaking the millitary alliance/trade embargo, either deliberately or through circumstances known to be about to happen.

If you do something like this, you have to wait the full 20 turns until all involved deals have expired, before you are allowed to do it again.

Cheers, Lanzelot

civ_steve

Dec 10, 2007, 01:24 PM

No, I think we are going in circles...Me too! :crazyeye:

So it looks like even without "evil manipulation" by a human player the AI on Sid level sometimes goes into negative income all by itself!! I've seen it at other levels and very early on in the game.

At Sid the AI is doing heavily, what the GOTM rules have forbidden: running a large negative income. So shouldn't it be allowed to the human player, at least on Sid?!No. The AI follow their programming, which leads them to this state. The player will abuse the privelege if given.

They may even be doing it on their own account, whithout being "Emsworth manipulated" by the human player, so can EAs really be called an exploit?! And can we allow something for Sid and ban it for the other levels?Yes, EA's can be called an exploit. The player uses a series of trades to lower the AI's BGN (999 value) artificially. The AI increases it's costs vs the BGN to go negative, a consequence of the decreased costs at higher levels.

If an AI is running such a large deficit, and we are gifting it let's say 500gpt, they will still be broke, so we still won't be able to sell them anything for gpt... That means on Sid EAs may be worthless...!That would resolve the argument for this game at least!

I see a problem with the following proposition by civ_steve:

Let's assume the 999-method shows something like 50gpt for an AI. We make a deal with them which includes 100gpt from our side. Shortly after that the AI decides, it does no longer care for having a positive income, and pushes the tech slider up to 60% and starts producing like 50 units per turn. A couple of turns later they are at minus 1000gpt and then our deal gets canceled (non-intentionally). We check their 999-value again and get a big shock... :eek: This rule would then make us pay for their entire unit upkeep and research expenses! This can't be your intention, civ_steve... :lol:

Very extreme thinking, Lanzelot, which is what this discussion requires! :) However, the BGN only shows the gpt coming in, the positive side of the equation; the increased unit costs will force the overall economy of the AI to be negative, but not show up in the BGN value. If the 999 value is the AI's BGN, it shouldn't be directly affected by any increase in the 'expense' side of the ledger. But it can be indirectly affected if the AI decides to adjust it's slider (if it has more libraries than marketplaces, changing the Science rate will change the BGN.)

The difference in the BGN should be mostly due to the loss of gpt the player was providing. The player would only be responsible if the amount of gpt being given to the AI was excessive - in this case, it would have to account for more than half of all incoming gpt before this limit is reached.

I think it's becoming more and more hopeless, the longer we think about it... Why not simply reward the human player's "creativity and inventivness" and give us as much freedom in our trading with the AI as possible? If we just put a rule into effect that prevents exponential growth via "iterative EAs", most of the sting would be taken out of the exploit. I think, we should setup a clear-cut definition of what EAs are, like the one given by Niklas, perhaps modified to include trade embargos as follows:
...
I'm trying to make this as simple as I can. There are far too many examples of how to get gpt from the AI that are marginally to incredibly exploitive, and I'd rather have the GOTM focus on improving player's abilities to do smart strategic play, not see who can take best advantage of the exploit of the month!

I think the easiest way to do this is to make a rule that 20 turn deals should be entered into with the full intention of them lasting 20 turns. They should never be terminated directly by the player, and if they do terminate early the effect on the AI should be determined, and if needed, remedied for the remainder of the term of the deal. One of the few allowable exceptions is the case where the AI is about to attack and player pays gpt for Tech; in this case the AI has to declare freely to absolve the player of the gpt deal. In the general case of giving gpt to the AI, this should usually be no problem since the deal was entered into with the expectation of paying the gold; at worse case you'd pay the remaining term. (This is a change to what I said before; clearly the player can not be expected to pay more than they would under the original trade deal.)

I. Larkin

Dec 10, 2007, 01:28 PM

OK, I decided to see how EA works in COTM43. Now I know, why Steve smiled...
At Sid level "system" protect itself from EA...
I think 50% fair enough. HoF's are using 100%...
I also reconed, that if AI have more than 999 gpt, 999 trick shows 999, or 999999...
So may be fair to assume that human income from AI must not be larger, then 999 gpt.

Elear

Dec 10, 2007, 02:12 PM

I think the easiest way to do this is to make a rule that 20 turn deals should be entered into with the full intention of them lasting 20 turns.

This is ridiculously simple. Why can't we all just do this?

Now, if you pay huge GPT for a tech, force war in some way, and get relieved of your payments, then there SHOULD be a rep hit, and no AIs will want to trade for your GPT anymore. This is a big penalty, and why many SGs have rulesets around honorable play to preserve reputations.

If you force war to get out of a deal, then you will get a rep hit and you can continue on. But you should never do it purely to exploit the AIs.

It all gets a little confusing at this point, because I don't see WHY anyone needs to use this techinque at all. There are countless other tactics that are purely just that: tactics. Take the beachhead for example. Get a big stack of defenders. Maybe a few armies. Rush barracks and walls. Deal with the AI stacks by bombarding and moving your units deliberately and cautiously. Slowly advance when you start to be able to take the advantage.

Let's go back to the basics, the core elements that sets games apart. REX...micromanaging, tech trading, city building, war strategy. Not the obscure deal manipulations to utterly destroy the essence of each AI.

Of course, there will be those who make the argument that to have truly great games, record-breaking games, you have to think 'outside the box' and advance with complex exploits (that such people will call clever strategies).

I guess I fall in the 'Niklas' camp, so to speak.

Lanzelot

Dec 10, 2007, 02:22 PM

The more I think about this, the better I like my original proposition, perhaps modified like this:

"In a deal which includes a resource, a millitary alliance or a trade embargo, the human player is not allowed to pay the AI more than the AI demands."

This would clearly put an end to all those kind of tricks, which involve getting the money back via "proxy AIs" or by provoking the AI into canceling the deal itself, or which try to damage the AI without actually getting the money back (e.g by just waiting until they have spent it and then canceling the deal). It tackles the problem at its root, namely the point where "artificially bloated" money is injected into the system.
In addition it would have a couple of advantages:

It still allows us a wide variety of strategies which do involve breaking a deal, but in a non-exploitive manner
The rule is easy to formulate and (hopefully) hard to bypass
We don't need to use inconvenient extra measures like checking stuff via the 999 method and comparing against earlier values
It would still allow us to gift money to an AI, if we want to set him up as a "research partner" or help him to put up a better fight against a common enemy. We just would have to give him that money in a "non-breakable" deal (one which does not include resources/alliances/embargos).
We don't need to worry about having to pay something without getting anything in return. Just consider the following not so unlikely scenario:

A super-power declares war on me, and as don't I want to receive the full blow of their attack, I buy a millitary alliance with one of their stronger neighbors. I really try to get the cheapest deal, but they won't help me for less than 50gpt. Sometime later I get a tech which they don't yet have, and as I urgently need some cash, I sell it to them. Again shortly after that the super-power surprisingly is willing to negotiate and would accept peace for a small tribute. As this seems the best option for getting out of this war unscathed and staying alive, I gladly accept, even though this pisses off my former alliance partner.

This scenario is no exploit in my opinion. My trading partner got the tech and is paying the fair price for it, I paid my fair amount of money while the MA was active and while he was helping me in that war, and now that he is no longer obliged to help me, I'm no longer paying. And I got the reputation hit for breaking the MA, that I rightfully deserve.

Under my rule this would be perfectly legal, because I refrained from that point which makes it an Emsworth Agreement: injecting extra money into the system. Under your rule, however, I'm afraid I would possibly have to pay some kind of remedy to my former alliance partner. Even though no EA was intended at any point, and even though I only chose those actions, which seemed most appropriate for my overall strategy at each point.

Ok, but now I'm finally off to start COTM43...!!! :eek:

I. Larkin

Dec 10, 2007, 02:42 PM

A super-power declares war on me, and as don't I want to receive the full blow of their attack, I buy a millitary alliance with one of their stronger neighbors. I really try to get the cheapest deal, but they won't help me for less than 50gpt. Sometime later I get a tech which they don't yet have, and as I urgently need some cash, I sell it to them. Again shortly after that the super-power surprisingly is willing to negotiate and would accept peace for a small tribute. As this seems the best option for getting out of this war unscathed and staying alive, I gladly accept, even though this pisses off my former alliance partner.
In this scenario you will get sever MA reputation lose, after that nobody will make MA with you for gpt. As you will be penalised by "system" badly it is not exploit at all.

Chamnix

Dec 10, 2007, 03:03 PM

I think the easiest way to do this is to make a rule that 20 turn deals should be entered into with the full intention of them lasting 20 turns. They should never be terminated directly by the player, and if they do terminate early the effect on the AI should be determined, and if needed, remedied for the remainder of the term of the deal.

That is quite a dramatic change in the rules, but I think I like it :thumbsup: (although it’s a little late for the current Sid game :shifty:). Many previously allowed exploits perfectly valid strategies will obviously be caught up by that rule: techs for gpt then declare, fake peace treaties to gain a couple cities before redeclaring, RoP abuse(?), etc.

It certainly has the advantage of being simple and probably easy to check for – the player’s reputation should be clean at the end of the game, and it seems likely that one of the gurus here would know how to find that in the save file. If the reputation is not clean, it doesn’t necessarily disqualify the game, but the player should be able to explain what happened (maybe make it required to keep a save from right after you notice your rep has been shot).

I think it seems like a preferred way to play. I abuse the heck out of the AI in my XOTM games because I assume at least some of the competition will as well, but I think requiring a clean rep (or an explanation of how it wasn’t your fault) is worth implementing at least on a trial basis. It sounds like more fun to me!

Of course, there will be those who make the argument that to have truly great games, record-breaking games, you have to think 'outside the box' and advance with complex exploits (that such people will call clever strategies).

In my opinion, it is no big deal not to have record-breaking games in the GOTM – we’re not really competing against past games as we are in the HoF.

"In a deal which includes a resource, a millitary alliance or a trade embargo, the human player is not allowed to pay the AI more than the AI demands."

I have to admit, I like this rule as well if we want to ban Emsworth Agreements but keep the ability to abuse the AI by breaking deals. There may be a couple minor loopholes in it (I think Niklas and PaperBeetle found some potential ways around it), but I think they are rare enough that it won’t be that big a deal. The main problem is that really the penalty for a blown rep in civ is not nearly big enough to compensate for the amount you can gain through dishonest dealing so I like civ_steve’s proposal a bit more :D.

Just consider the following not so unlikely scenario:

A super-power declares war on me, and as don't I want to receive the full blow of their attack, I buy a millitary alliance with one of their stronger neighbors. I really try to get the cheapest deal, but they won't help me for less than 50gpt. Sometime later I get a tech which they don't yet have, and as I urgently need some cash, I sell it to them. Again shortly after that the super-power surprisingly is willing to negotiate and would accept peace for a small tribute. As this seems the best option for getting out of this war unscathed and staying alive, I gladly accept, even though this pisses off my former alliance partner.

This scenario is no exploit in my opinion. My trading partner got the tech and is paying the fair price for it, I paid my fair amount of money while the MA was active and while he was helping me in that war, and now that he is no longer obliged to help me, I'm no longer paying. And I got the reputation hit for breaking the MA, that I rightfully deserve.

Under civ_steve’s proposal, you really have to think twice before signing in your alliance partner – is it better to fight alone but be able to get out sooner, or to have an ally and have to fight for 20 turns? It is a tougher decision compared to either fighting alone or possibly blowing your rep – it’s almost always going to be better to blow your rep since playing with a trashed rep is not usually that significant (especially since it only blows your MA rep as I. Larkin posted).

It certainly makes the game harder for all, but as long as the playing field is level, I think that’s OK.