Explotive gpt trade with AI

I was initially in the boat of allowing limited abuse through manipulative gpt trading. What's changed my mind is the existence of techniques to cut the luxury+gpt deal without suffering a rep hit, and the version of the trade where you get the gpt back through a 3rd party. Outlawing EA doesn't go far enough, and there's probably a whole slew of new trade abuses we haven't covered yet, which is why I'm converting to the more 'honorable' trading method as a rule.

I'm willing to give some latitude in the case of MA's; in times of war the strategic tide can change fairly quickly. If the player entered into the MA with the intention of being in it for 20 turns, but something serious has changed the situation, it should be allowable to take actions deemed necessary or expedient.

Accepting what the AI demands is a good starting rule, but I think it needs follow-up in case the demand is excessive. If you make a big gpt payment, and check 999 to find out your gpt to the AI is > half the AI's BGN number, you've entered into a situation that might require remediation if the gpt deal is broken. Otherwise, you're somewhat in the clear.

I don't employ this style of trade, so I'm not directly experienced with the power or usefulness of it. I find it quite sufficient to build up my commerce situation and gradually start taking gpt from the AI through Tech trades. However, this is more of a later game strategy than an early conquest/domination type of play. I'm curious to see if adopting a more 'honorable' trading style would be considered overly suppressive to a more aggressive domination/conquest form of play.
General remarks:
There are two points involved:
-using the AI weakness (and that is what it's all about)
-making use of a bad system (exploit, i.e. for me generating gpt out of thin air)

I'd be fine with an approach from both sides.
I.E. never pay an AI more than it demands. And check if a deal is broken if it is too much(say half of 999 value) gpt that you are getting net.
If an AI cripples itself if this criterion is fulfilled then it is the AIs fault, not the players fault.

Then there still might be some loopholes but at least the use is limited very much. And no money comes out of thin air (which is the main exploit for me)
Steve, I'd like verify my situation but it is in MA so I put question there.
I may be tired right now, but I can't see any problems at all with the rule Lanzelot proposed. And the reason I think I may be tired is that it seems so obvious that I wonder why none of us came up with it until now if it really is so good.
So am I. Unfortunately, in COTM43 I started use STEVE "rule", that may be not bad for Sid and If use only for MA. But I am very much against (and will not use) gpt+lux "gift", tech trade and route cut. It is really bug and exploit, and different from canonical EA. We urgently need to decide what to do for SGTOM 13. I think Lanzelot proposal is good enough.
Can AllanH veryfy it and put it "as a proposed rule" for SGTOM 13?
We have 4 days to discuss it...
"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."
If this is what you all are talking about (I lost track some time ago...) - I like it and favor it. :thumbsup:

Any further payments (to encourage research, to improve attitude - that was my only initial concern) are more effective in seperate trades anyway.

With this rule some gain is still possible as a bounty for tricky diplomacy - but this is certainly not game breaking and appropriate to the effort.
I haven't been round these parts for quite some time, but this thread has piqued my interest. This "Emsworth rule" sounds a lot like a trick I noticed a long while ago (from a dec 2005 thread):

A really unbalancing thing I found out about is to pay gpt for alliances:

for ( int i = 0; i< AIbankrupt.size() ; i++){

pay 50 gpt to AIi for an alliance v a dying AI;​
sell a tech to AIi for 48gpt (he can afford it now);​
then kill the dying AI;

You end up with loads of money. The AI end up with negative gpt.

Clearly this shouldn't be done, but you would have to search the rules pretty hard. Really I just mention this to make sure no-one is actually doing it, and to illustrate that there are probably lots of tricky little exploits not quite covered. Basically, if it feels wrong, it probably is.

I was pretty convinced back then that this could be powerful but no one seemed to believe me. I spotted it in GOTM38 (dec 04) and sent a pm to Ainwood about it. He was obviously very busy and I am still awaiting a reply. I take it that someone has now unleashed this trick to great effect. I always felt it was an exploit.

Sadly I haven't really got time to play civ anymore; I didn't really have the time then either. It's a pity that Civ4 seems such rubbish, although since I never came close to finishing a game I might have been hasty.
Offa is right: that trick isn't that new! Perhaps we should name it "Offa Agreements" then? :king:

But then: the famous formulas for the roots of degree-3 polynomials have been named after Cardano (they guy who published them), instead of Tartaglia (the one who found them)... That's the injustice of the world.

Anyway, do we have a definite rule for GOTM yet? My COTM43 may soon reach a point where I could use it (in a moderate form), so I'd like to know, whether it's banned or not.

With the holidays over I should be able to put a rule together soon. In general you should assume that putting any AI into a negative gpt situation through trade arrangements, whether intentional or unintentional, is not allowed, and if this does occur you should take some steps to remedy it. This is not very specific, but there are a lot of trades that might result in a negative gpt situation, so I apologize for the ambiguity and hope to be clearer in the future.

In general, if you enter any 20 turn deal and it lasts 20 turns, you are usually OK.
In general, if you enter any 20 turn deal and it lasts 20 turns, you are usually OK.

There is another situation that can be exploited, even if the deal lasts the 20 turns.

If the AI you trade with has the Wall Street, you can put it into a negative gpt situation by donating gold (up to 1000 for the +50 gpt) and then selling a tech for all its cash and gpt.

I don't know if it should be considered an exploit.
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.

Intention is a rather ambiguous and rather big word to be using...How can you judge what someone's intentions are without them actually TELLING you what they are? For example, if you sign a 20 turn deal lets say you ask 300 gold from an AI for 25gpt. The AI has units in your territory. You ask them to leave and they declare war. Would it even be remotely reasonable for someone else to say you actually intended to trick the AI? All you did was ask them to simply leave! You don't want their units in your territory. You asked them if they can leave. They declared war instead...can you help them that your not paying them anymore because they declared war on you? I'm sorry but the 20 turn deal rule is something we've already always been following and is not an argument against the "practice" that were talking about here...You happened to sign a loan with the AI. They happened to declare war over something entirely unrelated...
If anything like that happens, just gift those 300g or 25gpt back later or donate to anybody else or burn it by rushing a large building and switching to a cheaper one... :rolleyes:
Don't forget to take screenshots as proofs. :p

You can't make this exploit (and others) fool-proof.

We all assume that players know what they do and you cannot exclude every misuse. But you can spoil the fun if you try to make rules for every possible event.

If a player does not know what he does his exploit will most probably not have much impact on the success of his game and the results of the competition.
If anything like that happens, just gift those 300g or 25gpt back later or donate to anybody else or burn it by rushing a large building and switching to a cheaper one...
Don't forget to take screenshots as proofs.

Why would you be so anal about it? Besides that, a 20 turn deal is a legal contract. If one side breaks it, the other side does not have to do his part either. That's common sense...

With all do respect, I believe implementing a rule to "compensate" an AI for his foolishness is just that.

You can play your own game on your own difficulty settings with your own rules but remember that not everyone wants to play by your rules...
axehaxe - I don't see a problem with your example. Getting 300 Gold essentially for free, or in an earlier example, seeing a massive AI stack heading your way, trading lots of gpt for a Tech, then Asking the AI to leave resulting in a DoW, are both allowable.

The issue is when you inflate the AI's gpt available through 1 trade, extract the inflated gpt from a 2nd trade, and then the first trade is broken in some manner forcing a negative gpt situation. This is a difficult scenario to define a rule for because there are lots of different ways it can happen, some intentional and others unintentional; and there is no easy way for a player to detect if the AI is in a negative gpt situation. For the purposes of this discussion and subsequent play, Players must avoid driving an AI into negative gpt, and if they suspect this has happened, they must attempt to undo the negative gpt situation, either by giving a proportional gpt gift or an appropriate gold gift.

The only way your trading can create a negative gpt situation is if you are both giving and receiving gpt in different trades and the trade where you gift gpt is broken. At that point you would have to assess the AI's situation to see if they are in negative gpt (they might not be), and compensate accordingly.

It does appear that at high levels the AI goes into negative gpt frequently through their normal unit building activity. So the desire here is not to keep the AI out of negative gpt, but to keep the player's actions from pushing the AI into negative gpt with potentially unlimited effect.

Paul: I agree with what you're saying. For the purpose of this competition we do rely on somewhat on an honor system. This discussion should inform the players in the GOTM competition what to avoid and what to do if the situation occurs.
Oh, that one I definetly agree with - Creating a negative gpt situation, that one's certainly bad. In a way it can be considered cheating but at the same time I think it is bad for the human player as well. I've tried that once in a private non-gotm game once just to see what happens and ended up weakening a ton of AI's that could have potentially helped me to stop a runaway AI. Not a good idea at all. It just makes the runaway stronger and your potential allies weaker. The Iroquois pwned me that game and I was using the Mayans too!

There should be a patch that detects an unnaturally created negative gpt situation and causes the game to exit with an error sign. It can detect it based on 3 conditions:

1) There were 2 ongoing trades you made where one you recieve gpt the other the AI recieves gpt. One of those trades breaks resulting in negative gpt for the AI.

2) There is an unusually low dip below zero in any AI's treasury within the past 50 turns.

3) The total theoretical amount of gold and gpt that is supposed to be in the game is not consistent with the actual amount of gold and gpt that is in the game.

If all 3 conditions hold true then the game crashes and you get an error message saying "Illegal game parameters. The game will close." If 2 of the conditions hold true then you get a warning and if something similar happens again within 100 turns, the game also crashes for that.
Top Bottom