Abusing the AI - Gold before DoW

Quote:
Originally Posted by Greizer85
prohibiting all trades with AIs.

A setting for this would be perfect. At least something. Any solution is better than the way it is.

I once started a deity game playing like this - u can even see video in my link of 2 seesion - france deity - a very good game I think.

U know what happened at some point:
Ai offered me about 500 gold and 2 lux for peace - I pressed "accept" in a hurry without thinking and when i afterthought it - it felt like an huge abuse aswell ..
And this isnt even forbidden in those "great" hof "rules".

I'm curious btw: if you could fix the exploit right now, and you *had* to fix it (whether you agreed to the need for the fix or not) - how would you do it and why?

I guess just banning all gold deals at all is only serious way - pure lux for lux deals are also part of mp so they might be OK in sp aswell - there just is the problem that on rnd occasion they dont work because ai dont like u for some weired reason.

So maybe better ban them at all - or u just have to hope for friendly civs being close.

And obviously RA agreeemntes have to be banned aswell - they allways make the player profit way more as the ai. They are just flawed and I dont see any way to balance them to be "fair" - obviously they should give both civs same number of bakkers.

edit:
and above from a game designer perspective - its really hard as u have to make the dumb and bad players (95% of buyers) think that ai is playing to win and have it make deals (roleplay and so).
The core problem is that there is no easy to code way to make the ai "think" wheater a deal is good or not. The inbuilt variables (which f.e. make em give cities for peace if your military is x stronger as theirs even when on other side of map) atm are just very bad.

So yes the best way d be to make the ai only do lux for lux deals and never offer resources at all (it d not give u resources if its really playing to win). But then that not so fun and roleplai

I got no easy fix for designers to be true.
 
Make it so you only get the diplo hit if you give gpt/lux AND are the party that DOWs. This would still allow you to do the trick if you suspect an AI DOW is imminent - but how often do you reliably see the AI coming? And how many times per game?
Personally, I see thru an AI's "pretended friendship" every time one or two turns before the DOW. It's usually the same civs both in the same and across into different ones that like this.
The exploit isn't worth as much money for the AIs that outright say they don't like you since they won't pay full price.
Of course your right that it would introduce risk for those not as familiar with AI behavior compared to current. I'm a computer programmer by trade so mentally reverse engineering what the AI's code must have been to explain observed results is automatic for me.

That's a good summary of #1 & #2.

But #3 as written would be weird; I would have expected it to be something like trade suspended during the war and automatically resumes post war with same number of turns remaining. (SMAC). That tends to work unless you completely kill off that AI during the war.

I still prefer #2
 
Personally, I see thru an AI's "pretended friendship" every time one or two turns before the DOW. It's usually the same civs both in the same and across into different ones that like this.
The exploit isn't worth as much money for the AIs that outright say they don't like you since they won't pay full price.
Of course your right that it would introduce risk for those not as familiar with AI behavior compared to current. I'm a computer programmer by trade so mentally reverse engineering what the AI's code must have been to explain observed results is automatic for me.
Units on the border combined with denouncement is a sure sign, I've found. But either on its own is not always certain; once it gets in the mood, the AI still likes to spam denouncements left and right, and they often build so many troops that some are bound to be near your borders at all times (on the higher diffs, anyway).

That's a good summary of #1 & #2.

But #3 as written would be weird; I would have expected it to be something like trade suspended during the war and automatically resumes post war with same number of turns remaining. (SMAC). That tends to work unless you completely kill off that AI during the war.
Ah. Didn't think of that; makes much more sense. Ofc you could also stay at war indefinitely... After all there is no war weariness in Civ V (another dropped ball for the devs imo).

I still prefer #2
Me too tbh. Out of the options thought out so far, it's the one with the least cons (arguably none at all). You cannot get the lump sums in any way, period, so it's not as open to loopholes as the other two options.

@tommynt: Don't you think that's a bit too harsh, prohibiting all gold trades? Why not simply make the AI have less happy bonuses so it actually needs the lux? Ofc this should be compensated in some other area so that overall difficulty would stay the same (or as close as possible).

Really though the exploit as construed in this thread is about lump sum trades, not all gold trades. Although it's certainly a sound argument that all gold trading is exploitive to an extent, the lump sum case is infinitely worse imo. Only in the lump sum case can you easily get out of the deal, canceling your obligations with no penalty (other than the war, which is harmless due to the brainless combat AI). There must either be a penalty for breaking the deal, the deal must be impossible to make or it must be impossible to break. These are the #1, #2 and #3 choices; within them there can be variation, but I don't think a fourth scenario is possible. (I may be wrong ofc; I have a jumpy kind of mind and often miss obvious trails of thought.)
 
Personally, I see thru an AI's "pretended friendship" every time one or two turns before the DOW. It's usually the same civs both in the same and across into different ones that like this.

I think it is only fair that one can exploit the AI in this case, because it could also be human behaviour. Let's take Darius (who was next to me in my previous game) is pretend-friendly and amasses troops along my border, it is only reasonable of me to exploit the suspicion of an attack. If he wants to pretend everything is normal so I won't suspect his "surprise" attack, and trading "friendly" - so be it. This is not unlikely for a human player who wants to make his opponent at ease. (Naturally I would also move my archers/xbows/units away from the border to remove any advantage he has in the first turn of war.)

Remember that the AI also can spot an attack, and can demand that you declare war or that your troops are just passing through. If you choose war you lose your surprise attack, if you claim to be passing through (a promise that lasts like 30 turns!) you get a stiff diplomatic penalty with everyone for ages if you decide to attack anyway. This demand/request is something the human player doesn't have
 
Never seen that setting. Can not find it, where is it? I look under the advanced menu atm.

Odd, you're right. I thought there was a specific setting for this, but apparently not. Maybe it was patched out, or I'm remembering something from Civ 4?

There is still a way to play an always war game though. You could put all AIs onto a single team, then declare war as soon as you make contact. AIs would still be able to trade with each other, though.
 
I think the "best" thing would be to impose a small diplomatic penalty (-1) to the civ you cheated, and to his friends. It should last 20-30 turns after war has ended. Not sure what the penalty should read, something like:
"You backed out of a trade prematurely!"
 
Never seen that setting. Can not find it, where is it? I look under the advanced menu atm.

There is an 'Always War' setting in Steam's WorldBuilder utility. it is a free tool from Civ that lets you make/edit maps and create custom scenarios (and mods from other parts of the tool). But it would be a time-consuming task to do that before very game. Plus you'd know the map layout beforehand taking the element of discovery away.
 
Two more solutions I don't think anyone has mentioned:

1. Make gpt deals restore after the war is over. This means the exploit becomes a do-or-die type approach as you either have to keep a prolonged war, wipe out the AI victim or face the huge repayment bill when peace occurs. Doesn't complete remove it, but certainly makes you think twice before using it (unless you're deperate!)

2. Tweak the AI so that when it anticipates a war it only gives deals on a per turn basis. Obviously the difficulty here is how good the AI is at anticpating a DoW - but if it's beginning to march an army towards you they must have the intention of picking a fight and therefore can play its diplomancy accordingly. The exploit would still be available if the human player surprises the AI...however the AI is smart enough to spot troops amassing on their borders - so perhaps with this red flag they should be similarly reluctant to give lump sums. If the human doesn't amass and army on the borders then the DoW can be a genuine surprise and the AI has been 'tricked' in a valid manner. That could

Combining both would work quite well too! Could also cap the AI trading lump sums to 1000 in gpt/lux res trades.
 
Two more solutions I don't think anyone has mentioned:

1. Make gpt deals restore after the war is over. This means the exploit becomes a do-or-die type approach as you either have to keep a prolonged war, wipe out the AI victim or face the huge repayment bill when peace occurs. Doesn't complete remove it, but certainly makes you think twice before using it (unless you're deperate!)

2. Tweak the AI so that when it anticipates a war it only gives deals on a per turn basis. Obviously the difficulty here is how good the AI is at anticpating a DoW - but if it's beginning to march an army towards you they must have the intention of picking a fight and therefore can play its diplomancy accordingly. The exploit would still be available if the human player surprises the AI...however the AI is smart enough to spot troops amassing on their borders - so perhaps with this red flag they should be similarly reluctant to give lump sums. If the human doesn't amass and army on the borders then the DoW can be a genuine surprise and the AI has been 'tricked' in a valid manner. That could

Combining both would work quite well too! Could also cap the AI trading lump sums to 1000 in gpt/lux res trades.

The first two ideas are interesting and don't seem unreasonable. The first one would require an AI that could anticipate human actions without either missing an incoming DoW or assuming hostility when none is present (as is often the case when the AI confronts a player about troop movements).

I don't think capping AI lump sum trades would accomplish much, however. If the cap is too large, a human could just do multiple smaller trades to drain the AI treasury. If the cap is too small, no meaningful lump-sum trades could be made in peacetime.
 
Two more solutions I don't think anyone has mentioned:

1. Make gpt deals restore after the war is over. This means the exploit becomes a do-or-die type approach as you either have to keep a prolonged war, wipe out the AI victim or face the huge repayment bill when peace occurs. Doesn't complete remove it, but certainly makes you think twice before using it (unless you're deperate!)
This was mentioned by joncnunn a few posts back. It's much better than my original #3 imo. There is a catch here though - you could very well use the gold you swindle from an AI to kill that very same AI. I'm sure if AIs could actually think they'd appreciate the irony. ;)

Ofc the current repercussions of totally killing off an AI - namely the entire world turning against you with endless denounce and war spam - might well be enough to deter prospective Nigerian princes (so to speak). But sometimes with many civs around you can still find someone to trade with... And if the trick jump-starts your world conquest then the hostility doesn't really matter much. Some would term this a legit strategy, others would disagree and think it must be prevented altogether, not limited to one or a few uses. I'm not sure which camp I stand in at the moment.

2. Tweak the AI so that when it anticipates a war it only gives deals on a per turn basis. Obviously the difficulty here is how good the AI is at anticpating a DoW - but if it's beginning to march an army towards you they must have the intention of picking a fight and therefore can play its diplomancy accordingly. The exploit would still be available if the human player surprises the AI...however the AI is smart enough to spot troops amassing on their borders - so perhaps with this red flag they should be similarly reluctant to give lump sums. If the human doesn't amass and army on the borders then the DoW can be a genuine surprise and the AI has been 'tricked' in a valid manner. That could
Is there something missing here? The last sentence seems to get cut off rather abruptly.

I'd be very cautious when trying to implement something like this. The AI isn't really 'smart' in that it doesn't really think at all - it is just a series of various rules and conditions that interlace in intricate ways and synergetically lead to acts that appear reasoned - or rather this is the ideal. It can be argued that as far as AIs go, the Civ V AI is plenty terrible. The only reason it can compete against the human at all is its ridiculous slew of bonuses, enabling constant city and unit spam. So when it comes to nuanced decisions like when to accept deals with war hovering in the air, it would be easily exploitable. Just hang back with your troops away from the borders, make the deal, DOW immediately and march your troops to the borders (along with newly-bought surprise reinforcements...).

As a side note, imo the human player should have the option of questioning AI troop movements, too, or something similar. Ofc the trouble here is that you'd always do it, sometimes gaining a huge advantage (delayed attack or a major malus for breaking their word for the AI), and never a negative. There's no decision involved, just tedious clicking whenever you see them mass at your borders (meaning every turn with every neighbor on Deity). Yet it feels unfair that the AI can do this while we can't. I don't know how to fix this problem really.

Combining both would work quite well too! Could also cap the AI trading lump sums to 1000 in gpt/lux res trades.
A cap could somewhat work I suppose - no more lump sum trades for 30 turns with the player after one is made. But the trouble is the more players there are the more AIs you can swindle. And 30 turns is not that long; in fact it's just enough for them to amass a new fortune ripe for the picking. Unless you meant 1000 gold all-game long? Even then a few such trades early on would give a snowballing advantage, especially if you'd buy units to do conquests with. I'd rather do away with lump sum trades altogether tbh. It's safe and clean and arguably adds a layer of strategy (when to save and when to spend - do you keep an 'emergency fund' for sudden DOWs or use all your money on advancing science etc?).

Edit: Gotta love this forum's way of logging you out unless you refresh every 5 seconds. There wouldn't be a way of changing that, now would there?
 
I loved the great fall patch. But there is one major major problem still in the game: Offer the AI a trade gold vs gold/turn. Then declare war.

This ruins late game. If you are pressured it is hard resist doing it, it is just too easy.

Why not just patch this? The solution I offer is: If you make a trade to receive gold you cannot declare war in 10 turns. This would solve the whole thing.

How can we come in contact with the people patching the game?

Please help. :sad:

The problem is that you open the game to a ton of other problems. Like if you can't DOW, does that mean they can't dow either. If they can still DOW it stops practically nothing of the exploit because often you can even tell when they are going to DOW you and still trade for reasonable values. Oh and if the AI can't DOW you because you traded for gold... That's far more exploitable.

Point is, it's an AI. As far as strategic games go, it's gotten better, but you have to draw the line as to what you will exploit in it and the things you won't do just for the sake of not losing.
 
I take it one step further... when going to war with the last civ in a domination game, I'll trade all my luxuries and GPT (approaching 1000 during golden age) for a number of the last civ's cities just before declaring war. Cheap? It doesn't matter... victory is inevitable and I'm just speeding the process up.
 
I take it one step further... when going to war with the last civ in a domination game, I'll trade all my luxuries and GPT (approaching 1000 during golden age) for a number of the last civ's cities just before declaring war. Cheap? It doesn't matter... victory is inevitable and I'm just speeding the process up.

Selling cities is actually another dumb thing the AI can do that it wasn't allowed to in earlier versions of Civ. *
(In one version, the AI would only trade you a city to get out of a losing war and only if human didn't give them anything for it)

* A human may have good reason to sell a city in some circumstances; but an AI can not possibly be expected to recognize those circumstances. Humans would assume if they saw an AI sell a civ to another AI that at least one of those two AIs made a really dumb move. (It's either a worthwhile city spot and so shouldn't have been sold or else its a worthless city and so the other AI shouldn't have been willing to buy it.)
 
No matter how we all like to get gold from AI's for something that benefits in duration(GPM,luxuries) - it makes no sense if you think about it.

If you sell a "one time thing" - you can/should get a one-time payment. Like if you sell a city. Or want AI to declare a war. Otherwise there's no point of keeping it, but god, how much I preffer to get money over GPM from AI's(without abusing DoW staff), because using the money I can already buy a workshop/market/lancer and the money starts "to work".
 
Trading away all the ressources to an AI and then declaring war - I consider this as cheating/exploiting

Trading away all (or some) ressources to an AI that you think will DOW you soon - I consider this as strategy and an AI flaw.
 
I think the real answer is making ALL trades involving cash be GPT only. Money management should be important, not a minor inconvenience that can be fixed with "Oh, I'll just make a stupid trade, and fix my entire empire's financial problems."
 
I think the real answer is making ALL trades involving cash be GPT only. Money management should be important, not a minor inconvenience that can be fixed with "Oh, I'll just make a stupid trade, and fix my entire empire's financial problems."

That would work too.
 
Top Bottom