Hi! Been enjoying Endless Space (SP, Vanilla) but found that the diplomatic trades are easily exploitable, wanted to mention how and why. If this has been discussed previously, please forgive me and point me to the right place.



Unlike many exploits, these are hard to avoid without abandoning diplomacy entirely. One of the 4 Xs is exploit, after all-- part of the game is buying low and selling high.



Exploits:



My trade goods for all of your income. Oh, and that tech please.: Load up enough trade goods to counteract as much negative income as they're willing to accept. Since that negative income will eventually drive them bankruptcy, you're guaranteed a quick end to the agreement, without having to shoulder any of the relationship loss from breaking it. Since you're guaranteed a short-lived (usually 1 turn) agreement, throw in the rest of your trade goods and get some tech for it.



Here's a gift! Oh it's just enough for what you're going to owe me, how fortunate. When AI are willing to accept a trade for all of their savings, they don't care whether that's one credit or one million. That makes it safe to give them a gift of any amount before the trade. This is handy in conjunction with the other tricks, since AI won't accept a per-turn cost larger than their current savings. (This is probably bugged; I imagine the intent was that they shouldn't accept a negative income in excess of their current savings.)



All of your trade goods for 1 credit per turn, because beggars can't be choosers. Once you've devastated an AI's income, probably with a deal guaranteed to expire the next turn, they are willing to accept ridiculously low offers for their own trade goods, because they are comparing a handful of credits per turn to their zero or negative income.



Why they work:



AI value trade goods very, very highly. In particular, they're willing to give up their entire income for trade goods, which is a bad idea.

AI value credits and income in relationship to the size of their credits and income respectively. Since these are factors affected by trade, a player is set up well to exploit this valuation.

AI are honorable. They very rarely reconsider on-going deals, at least, when they feel good about the faction they're dealing with.

AI are willing to consider one-time trades in the same offer as on-going trades. They readily trade tech and savings, thinking that the long-term is worth it, but they are poor at predicting the future-- unable to realize that a given exchange will, in fact, not last through their next turn.



What to do about it:



I don't have a perfect understanding of how diplomacy works in Endless Space. I'm going to suggest a few things because I think it's crappy to detail a problem without offering any solutions. But the solutions I offer aren't guaranteed solutions, just whatever I would try first, if I was the developer.



Change valuation of savings and income to become based upon average amount of savings or income in the universe. This could probably be plugged in pretty easily. It would also prevent single turn fluctuations in valuation, which are easy to exploit.

Increase valuation of negative income. Entering debt endangers all per-turn deals. Ideally, each deal should have an expected duration based around expected game length and treaty-breaking behavior. This would allow debt to be valued in context of the current deal (will only last x turns based on money and income) as well as previous deals.

Evaluate existing treaties on a regular basis, with existing diplomatic weights. I'm not sure to what extent this is already being done. But existing treaties need to be continually evaluated in the same way that a proposed trade would be-- with a modifier for cost to relationship, of course. It makes little sense to maintain their 1 hyperium for 1 credit per turn cost to me, when they're willing to buy 1 unit of hyperium back for so much more.



I'm interested in this as much as a discussion of abstract mechanics as I am in the context of Endless Space. Diplomatic weights are an interesting problem.

I suspect the biggest problem is valuing dust and income based strictly on current dust and income, rather than being tied in some way to their actual utility. I once discovered that an AI with a lot of dust on hand but negative income was willing to give me a flat amount of dust in exchange for dust-per-turn that wouldn't have paid back the flat fee even if the deal lasted until I could have canceled it at no penalty; even under a naive analysis, that deal couldn't possibly have helped the computer. That's what comes of trying to optimize each variable separately, without considering the (rather trivial, in this case) relationship between them.



Ideally, the value of dust should probably be calculated based on some specific plan the AI has for how to use it (like "if I get a large cash pay-off this turn, I'm going to use it to buyout X and Y, therefore it will be worth Z to me"). But if that's too complex, I think it would probably be better to treat the value of dust as constant, and scale the value of resources based on the size of your empire (e.g. a luxury resource that adds +2% food will be worth twice as much if you double your food output).



If we're aiming to make the AIs good at the game (rather than making them role-play even when it hurts their chances of winning), there should also probably be a distinction between "I don't want to give you X because it will hurt me" and "I don't want to give you X because it will help you", and the latter part should be scaled based on how large of a threat your trading partner is. For example, your fifth copy of a strategic resource doesn't help you (except as a backup), so the AI shouldn't mind too much trading it to a player in last place and in the far corner of the map, but should still be reluctant to trade it to a neighbor in first place (especially if it will be that neighbor's first or fourth copy).



(Since the "hurt me" component is zero for technologies, that should lead to rampant technology-trading in games with a large number of players, but that's a problem with the rules for technology trading; under the current rules, trading techs aggressively is the rational strategy in a game with a large number of players.)