Discussion thread for exploits and strategies

[ShannonCT]

sorry, I post here again :

I think I have found a 6th point in my reasonning:

6) the ending result of the "bug" can be achieved otherwise : so that the result : a civ paying eternaly +16-24gpt for 1 ressource is not a bug. (I haven't yet the willpower to test if it works but in theory it works)
1) say you sell gems to civ A for 3gpt
2) If you wait ten turns (as the contracts reach an end) you notice that the Ai has +3gpt available to exchange against another ressource.
3) Cancel the trade for gems, the Ai will then have +6gpt available that he will pleasantly trade again with either the same gems or another ressource.
redo steps 2) and 3)

Obviously, this is a bug, too.

I guess our entire capitalist system is a bug then. If I buy season tickets to the New York Yankees for $2000 this year and next year am willing to pay the new price of $2500 for the same seats, am I buggy? Are the Yankees buggy for asking me to pay more for the scarce resource?

Using the subsidy/trading "exploit" is just correcting a design flaw* in the trading screen: the AI is not able to immediately rebudget in response to new information. If I hook up my second rice and decide to offer it for sale, and I find that none of my AI rivals have any GPT available, is it because rice is worth nothing to the AI? Certainly not. If the AI has any cities with negative health, they would benefit from having the rice, and rebudgeting or changing their allocation of citizens would allow them to obtain the rice and end up with more surplus food and no less GPT than if they had not done the deal at all. If I use the "exploit" to get the AI to pay me 1 GPT for the rice, the AI is certainly going to benefit more than I. If I can use the technique to get 20 GPT from my friendly neighbor, I am probably benefiting more than they. But how is it different from asking an AI for a gift of gold?

I often find that AIs who are pleased or friendly with me will grant my requests for gold, even if I am much weaker in power then they. This is just a reward for maintaining good relations, and this option has been deliberately offered by the game programmers.

*I apologize in advance to the CIV4 programmers for the use of the term. I love you guys. This is just one point in the game that is unrealistic and where the AI seems unable to do any cost/benefit analysis.

 

[Thrallia]

the fact that you cannot get an AI to give you that much gpt without using the gifting exploit should show to you that it is indeed an exploit.

I agree that cancelling a trade and reagreeing to it for more gpt is not an exploit, simply because the amount the AI had available to trade had gone up because of the AI changing its sliders or its economy improving.

However, the simple fact is that an AI is unlikely to ever have 20+ gpt to trade you normally(which should tell you that it is indeed an exploit: despite the value the AI places on the resources, they will never be able to give you as much as they feel it is worth unless you gift the extra to them at first), although I've gotten a strong AI to give me 10+ gpt for a resource without using any exploits.

Finally, an AI doesn't have to be pleased or friendly with you for it to be willing to give you gpt for a resource. They may not be happy about it, but they'll still trade you gold for a resource.

Beside the fact though, I gifted an AI some gold in the latest GOTM to keep him happy with me(I was scared of dying lol), but when I decided that it was no longer necessary, I couldn't cancel the transaction, so I funded him 5 gpt for the last 2000 years of my game lol

 

[DaviddesJ]

DaviddesJ, I might have misunderstood you but do you propose that breaking a deal and amaking at again at higher price is a bug?

If the AI is just willing to pay more for resources over the course of the game, there's nothing wrong with that.

But it's a bug/exploit if the AI is only updating its available cash once per game turn, so, when you cancel a deal where you are sending it resource for gpt, it suddenly is willing to pay more for resources than it otherwise was. Essentially, it's a weakness in the AI method for deciding how much to offer for resources, which is based primarily on "available gpt", a quantity which can be manipulated, rather than on the actual value of the resource to the AI player.

 

[Lexad]

Ahh, now clear. Yep, we agree here on both points.

 

[Calavente]

disagreed :

the decision for the AI is firstly made :
1) Is it under the cost I would like to pay : say 16-20gpt (someone founded it hardcoded with a value + a formula of diplo points and type of ressource but I do not remember where.)
this value cannot be easily manipulated by easy and cheap tricks : and it is the actual value the AI give to the ressource,
then it says
2) How much can I offer NOW?
this means : how much surplus do I have : 0gpt ? 2gpt ? .. etc.
as the Ai balances well its economics, to use it in the most efficient way (between specialists/ culture-gold-science sliders) it moslty never have 16-20gpt available, only 2-7, maybe 10 if you are lucky. (and the player will only see this part as the only part being visible)

3) then it proposes to pay up to its available surplus!
that is : 2-7 and not the value it would have proposed if he could change its slider during the trade proposal.

So yes it is a bug in the programming (maybe), as the Ai cannot allow for higher bidding than already available surplus. but no it is not an exploit.
Yes, while cancelling the deal, you "allowed" the AI to see that it had more surplus than thought. How is it an exploit ?
by doing this you are doing no harm to the AI, only restoring a wrongness inflicted unpon the player.

I think bribing a civ A into war with B and then attacking A with your own forces is much more an exploit.
In those terms, even bribing an AI to do anything other than a phony war is an exploit.

So if you were an AI with a sword when the fight should be barehanded, it would be ok, but if I come up as a player with either a way to get a sword of my own or kicking your's out, then it is an exploit.

I can't understand this.

Arg re-read thrallia's post...
How can people reason like that ????
the game has a loophole/bug that prevent the AI from being able to pay the price it would have liked. I is made to prevent players to utterly push an AI into misery. But it was done the harsh way. another harsh way would have been to reduce the value of ressources... in the range of 3-10 nobody would have to try the new solution.
Now people found a difficult way so that the Ai will pay the price it think is right.
And you say :

However, the simple fact is that an AI is unlikely to ever have 20+ gpt to trade you normally(which should tell you that it is indeed an exploit: despite the value the AI places on the resources, they will never be able to give you as much as they feel it is worth unless you gift the extra to them at first),

For me it really looks like :
There is a bug in the game doing A instead of B for the AI. then someone comes with some ways X Y and Z to get B. and then you say : 'it is an exploit, obviously not using your trick you would only have A and not B'.

and so what : there was a bug. if you use the ways X Y or Z you correct part of the bug, if you do not use any of them, you are letting the bug play.

the fact that you cannot get an AI to give you that much gpt without using the gifting exploit should show to you that it is indeed an exploit.

"the fact that you cannot get the lion to lose against you without you having a spear should show you that it is indeed an exploit for you to throw rocks at it"
is it true ? no : an balanced fight would not be a naked man vs a lion, throwing rocks is only a feeble way to compensate the fact you were not given any weapons. it is not an attempt to do ranged fight instead of close-fight.
the same is true for this so-called exploit. the way it is, there is an issue. something should be done.

you are using a flawed thing as proof that the remedy against it is an exploit. Is resting in bed and taking medicine an exploit against the fever ?

 

[Calavente]

oh and the AI do not have to be happy with you to pay gold for ressources, but it have to be happy if you ever want some heavy amount of cash.

And yes the amount available has risen when you cancelled a trade 10 turn latter... but with the exemple I proposed the Ai's economy would not have time to really improve. If I had not done the first trade on turn 0, then on turn 10, instead of being proposed 4(from trade at t0) + 4 (from arrangement of the sliders) I would only have been proposed somthing like: 2gpt : what the Ai would have as a surplus.
so in one case it would have only 3gpt, and in one other case it would have 8 available ?
how is it possible if all I did was ewchanging 1 ressource for some money?
only if there is a loophole somewhere.

another thing : If the Ai can only spare 3gpt for a ressource without going bankrupt (the essence of why it could be an exploit), why can I get 3gpt on turn 0 then 5 gpt for another ressource 1 turn later? because the AI put more value on the second ressource? no they are both happy ressources. because its economy has improved a lot? not in 1 turn. because it has moved it slider? OH YEES !! as anyone but the Ai would do. but why didn't it propose 8gpt for the first ressource ? remember the Ai was hardecoded to accept up to 16-20gpt if money available.

on turn 1, isn't the AI spending 8gpt for ressources?
on turn 2 it would be 12 gpt for ressources.

the programming says : spend up to 16-20 gpt for that ressource in those condition if money available. and on turn it has 12gpt available, so retrospectively it should be able to spend 12gpt for 1 ressource.

Tell me now that the AI couldn't at turn 0 spend 12gpt on ressources without crippling its economy.
2 turns latter it could. how? by magic ? If at t2 it can spend 12gpt, at turn 0 it would at least be able to spend 10gpt, not 3!

But no. theres a bug.

a bug is an issue in a game, people have to live with it in any way they can.
an exploit is a bug that for the human but that he can abuse to gain an unchallanged bonus vs the AI.
so an exploit is a bug but all bug are not exploit, especially if they play against the human player.

What you call the "gift exploit" is a way to live with a bug that played against the player and the gameplay even if it is not the most elegant or honorable way to do it.
(a real exploit would have either to be able to be paid more gold than the value the AI think it possess more than a couple dozen ie 100 gpt or something else., another one would be a way to sell ressources to an AI way more -20 gpt- than what you paid them -10gpt- The so-called exploit does not do anything of those two)

You have four possibilities :
-either you use those technics.
-either you show the programming team of firaxis what seems to go wrong.
-either you mod an answer to this issue
-either you do nothing
But in no case you have the right to tell people they are exploiting a loophole.
WE are victims of a loophole !!! do not transforms things.

If you still disagree :tell me using other arguments why it is an exploit; what is exploitative? the unit gifting bug is an exploit, the perpetual anarchy was also one. What here is an exploit, even a minor one. (if it is the fact the AI cannot cancel the trade it is not an exploit as if you don't have enough diplo, someday the AI will cancel it.)


PS : sorry for the 2 (oups 4) very long posts...

 

[DaviddesJ]

But in no case you have the right to tell people they are exploiting a loophole.

Of course I do have the right to say that. It's called "freedom of speech". Look it up.

The value that the AI attaches internally to resources is often way too high. This is a weakness in the AI implementation. This weakness is mitigated, in ordinary play, by the fact that the AI limits its trading to "available gpt". So, even though the AI often puts much too high a value on many resources, that's usually not a big problem. It lets the AI trade resources one for one (because it places a very high value on both the resource it's trading and the one it's receiving)

But, when you find ways to get around the cap, you're exploiting that AI weakness of attaching too high an internal value to the resource.

 

[Thrallia]

and yes, in one turn an AI can indeed improve its economy enough to afford more gpt.

In many instances I've built 2-4 banks or markets in a single turn and boosted my gpt by well over 15.

You are comparing the exploit to an errant real life scenario. A more accurate one is that you want to get a Cable internet connection, but can only afford 10 bucks per month and it costs 30 bucks a month. Certainly you value it enough that you'll pay 30 bucks a month...IF you had enough to spend, but since you don't, you've either got to do without or the cable provider lowers his price for you.

Now say someone comes over and is willing to pay the difference for you, you eagerly accept and start zooming around on the internet. However, you signed a 2 year contract for 30 bucks a month and after 2 months your friend stops helping you pay the bills. Now you are stuck paying 30 bucks a month for something you really want, but can no longer afford. This ruins your credit over the course of the 2 years because you have to backpay over bills to afford this one.

That's the example you should be using for comparison. Obviously, there is no contract after 10 turns, but the AI is unable to see that it is damaging itself, so I likened it to a contract for similarity's sake.

As far as what an exploit is...in terms of security, it is taking advantage of a loophole or manipulating something to your advantage.

Since the AI will never have 16-20 gpt to spend on a resource(regardless of how badly they want it), it is indeed an exploit to artificially and superficially increase the gpt an AI thinks it has to spend.

 

[ShannonCT]

That's the example you should be using for comparison. Obviously, there is no contract after 10 turns, but the AI is unable to see that it is damaging itself, so I likened it to a contract for similarity's sake.

As far as what an exploit is...in terms of security, it is taking advantage of a loophole or manipulating something to your advantage.

Since the AI will never have 16-20 gpt to spend on a resource(regardless of how badly they want it), it is indeed an exploit to artificially and superficially increase the gpt an AI thinks it has to spend.

I've never been able to use this exploit to get more than 10 gpt for a resource. And I can think of plenty of scenarios that I would pay 10gpt for a resource, especially a luxury like gold. If getting a resource means having 1-2 extra happiness in every city and being able to put those extra citizens to work producing more GPT or hammers, then paying for the resource can result in a net gain in wealth, beakers, GPPs, and productivity. Just because you find a way to get the AI to pay a reasonable amount of gold for a valuable resource doesn't mean that you're tricking the AI into harming itself. Maybe the AI needs to be programmed better to recognize when a deal is no longer in its best interest. But most of the deals I get the AI to take with the trading exploit are deals that I would quickly take myself for promoting the growth of my cities.

When you subsidize the AI, there's always a point where further subsidies make no more GPT available for trade from the AI. Where does this number come from? Could it be that the AI is programmed to know the most it should pay for a resource and that this subsidy cap is the value of the least desirable resource that the AI might be offered?

 

[Thrallia]

no clue, all I know is ever since we went to monarch and above on the difficulty levels, the AIs in my games have always had plenty of gpt to share with me. I've had no trouble getting 5+gpt from a single AI for multiple resources, and I even got 14 gpt from Khan for Iron recently, so unless you're being greedy I'm not sure when you'd even NEED to use this exploit

 

[DaviddesJ]

When you subsidize the AI, there's always a point where further subsidies make no more GPT available for trade from the AI. Where does this number come from? Could it be that the AI is programmed to know the most it should pay for a resource and that this subsidy cap is the value of the least desirable resource that the AI might be offered?

There's a limit that seems to be related to the size of the AI's economy, level of economic development, etc. It's quite simplistic---e.g., it doesn't seem to take into account whether the AI has plenty of health resources but few happiness resources, or whether the AI actually has any use for a particular strategic resource, etc. Some people have looked at the code in the SDK and can say more about it.

 

[occam]

But, when you find ways to get around the cap, you're exploiting that AI weakness of attaching too high an internal value to the resource.

DaviddesJ,

First, a linguistic observation: you use the word 'exploit' there, but it sounds a little broad to me... I could say that my invasions exploit/target AI poor tactical troop placement.... but neither one of us would be using 'exploit' in the regulable sense, just in the opportunity sense. It is perfectly allowable to attempt "exploiting AI weakness of attaching too high... value to the resource."

Which brings us to your "cap" (sic). I find your claims about the true purpose of this "cap" to be speculative.... have we come across any hard evidence about the intended purpose* of this "cap" since the last time we danced these steps?

Do we all know the moves yet?
Pro says: "it is reasonable to try to influence the sliders of the AI, and we'll even pay a GPT cost"...
and then con says: "but it circumvents the intent of some cap in the code"
and then pro says: "how are you mind-reading that intent?"
and con says: "you don't have a dev quote either"
and pro says: "ah, but we don't need one - don't you understand burden of proof here?"
and alanH says: "this has not been determined to be an exploit"

So, do you have new evidence, or are we just dancing?

- O


* Heck, we don't even have any hard evidence of the friggin' cap -- let's bump that up a little higher on our agenda, all right CON people? I'd love a chance to speculate about intent too, but as long as it's vague, magic code, it's a lot easier for the CONs to make up those stories, isn't it?

 

[occam]

As far as what an exploit is...in terms of security, it is taking advantage of a loophole or manipulating something to your advantage.

(ellipsis in original)

Speaking of linguistic spread of the definition of 'exploit,' Thrallia provides the poster child.... c'mon, you can't REALLY believe that we should regulate "manipulating something to your advantage" -- should we ban manipulating your troop locations to your advantage? How would one comply with that ban?

I believe that clear and rigorous thought about the word 'exploit' is a prerequisite for any sincere attempt to regulate. Vague and slippery thought tends to herald a rush to judgement.

- O

 

[DaviddesJ]

It's quite obvious that the gpt that the AI will offer for a resource is the minimum of "available gpt" and the "gpt value" that the AI assigns to that resource. This isn't "speculation", it's just observation.

I also disagree that there is any difference between "regulable exploits" and "non-regulable exploits". The GOTM staff can make whatever rules they want for what players can and can't do.

The only "new evidence" I would be interested in is evidence about effectiveness. The more effective this particular tactic is, the more sense it would make to forbid it. Currently, it doesn't seem very effective, so there's not a big rush to forbid it.

I don't think "designer intent" matters at all. If you read my postings, the only thing that matters to me is how the possible AI manipulation affects gameplay. Clearly, if the manipulation has a huge effect on gameplay, then it was unintended by the designer. None of the people involved would deliberately design a mechanism where players can get a big advantage against the AI through a counterintuitive tactic of giving them extra gold---I know them well enough to know that. When I discovered RCP in Civ3, certainly I didn't think to myself, "Hey, maybe the designers deliberately made rank corruption equal to the lowest value for cities at the same distance, in order to encourage cities to be constructed in rings."

Once anything rises to the level that it would be considered for banning by GOTM, we already know it is outside the designers' intentions. But, even if it weren't, the effect and the result is the same---whether something should be banned, or not, depends on how it affects play, not on what some designer thinks about it.

 

[Shillen]

The only "new evidence" I would be interested in is evidence about effectiveness. The more effective this particular tactic is, the more sense it would make to forbid it. Currently, it doesn't seem very effective, so there's not a big rush to forbid it..

Eh, I think it's a hugely effective exploit. Almost every spaceship game I play I'm trading like 5-6 resources for 3-6gpt each. If I could be gettin 20gpt for all of those that's an additional 70+ gold per turn in my coffers. That's probably enough to move my slider up 10%. I don't do this, though, because like you I consider it to be an exploit. I don't care if it's within gotm rules or not, I don't use exploits. Just like I won't "intentionally" use the production bonus whipping exploit, or chop settlers without halting growth at all, or partially chop forests or any of those things that are allowed by GOTM rules. I palace jumped in one GOTM for civ3 and I felt dirty doing it and never did it again.

 

[DaviddesJ]

Eh, I think it's a hugely effective exploit. Almost every spaceship game I play I'm trading like 5-6 resources for 3-6gpt each. If I could be gettin 20gpt for all of those that's an additional 70+ gold per turn in my coffers.

But could you? Given no actual evidence of such results, there's going to be little support for a ban. That is my point.

If one has never attempted such a strategy, and has no empirical evidence that it works, then it seems to me very premature to call it "hugely effective".

 

[ZerrorR]

People from Firaxis surely monitor this forum, so if they find this an exploit, they'll surely fix it in the next patch... Can anyone confirm is this posible in Warlords...

Discovered by player Rustavelli from Civru.com

If quick moves are turned off and you give a GP a command to discover science/join city/finish production/build academy or great shrine or art masterpiece, while he's waving his hands you can select another GP in this city and activate GA. The first GP would both complete the assigned task and aid in Golden Age. You get free Great People this way.

This exploit should be monitored and banned in the official games.

The relevant bug thread

Edit: it is fixed in Warlords.

Designers did know about this bug and fixed it in cIV:W
And IMF is long enough on a forum to be changed

Z

 

[DaviddesJ]

People from Firaxis surely monitor this forum, so if they find this an exploit, they'll surely fix it in the next patch...

Lots of exploits never get fixed. Some things are easy to fix and some, especially in the area of AI weakness, are much harder to address. Some issues in Civ3 were never patched.

Besides, as far as we know, there will never be any more patches to vanilla Civ4, so, as long as there are vanilla GOTM4's (many years, probably?) this will be a potential issue.

Anyway, it doesn't actually matter whether Firaxis thinks it's an exploit. What matters is whether the GOTM community thinks it's an exploit. And, if so, whether it's worth addressing.

 

[Shillen]

People from Firaxis surely monitor this forum, so if they find this an exploit, they'll surely fix it in the next patch... Can anyone confirm is this posible in Warlords...

Heh, they still can't save my leader name/civ/opponents and half the map settings when I make a new game. That's been complained about since release and they didn't fix it in any of the patches, nor did they fix it in warlords.

 

[Calavente]

@thrallian :
the comparison would be :
I hav only 30bucks to pay for the adsl.
and when the provider reduces its cost the next day, I accept the deal.
the day after I go to the same provider and ask him : "can you reduce the amount to 30bucks for the unlimited phone too?" yesterday I couldn't at all spend more than 30 for adsl even when I truly wanted to pay $90, but today I can spend $30 more per month for the ilimited phone." Try that once in RL : the first time will be ok, maybe someone gentle or very generous would allow you to have the adsl for only 1/3rd it's price. But when you come with the second demand he will answer : "WTH !! no deal!" or even better : "now you have 60bucks to pay for adsl, pay me with it" or even "I cancel our deal you cheater !"

yes ! exploiting the game is using a loophole/a misprogramming in the game. But when the loophole plays agains you in normal time and you find something to attempt to do it right what will you do ?

Using attacks with stacks was supposed to be outed from cIV with the arrival of collateral damage...Etc. but now we see that it plays not like that. In fact it is even worse. And it changes the gameplay a lot. So you would forbid also attacks with stacks? no. and they are an exploit of a sort while the IMF bug is not so clearly against the (supposed) intentions of the programmers.

oh and to this :

Eh, I think it's a hugely effective exploit. Almost every spaceship game I play I'm trading like 5-6 resources for 3-6gpt each. If I could be gettin 20gpt for all of those that's an additional 70+ gold per turn in my coffers.

if you really have 5-6 ressource to 1 guy : try cancelling all trade the propose one or two of the ressources again. see how much you will earn...
but you won't earn 20gpt each ressource but 20gpt 1 ressource. and the other you won't be able to trade.