FreeOrion

What's important is starting with the basics of diplomacy outside the AI system, such as the current status between two players (i.e War, Truce, Non-Aggression Pact, Trade Agreements, Defencive Alliance, Full Alliance), and what kind of options it allows for individual players (trade agreements bring variable income from surplus resources and luxuries, war automatically enters into battle mode, allies assist, full alliance allows movement of warships through empires without civil provocation warning, and access to allied player's intelligence).

Then after, we assign variables to numerous objects determining diplomatic effects, for example: the number of AI player ships another player has destroyed, and how much importance each ship represents, with various decrementations as turns passes. Another example would be if a player gives a generous offer or a ridiculous one, and keeps pushing it, All these weights including the casus belli would determine what kind of relations is most probable with the two players. If it's very negative, it can lead to communications breakdown, or if it's even leading to war, then the AI player determines if it can afford to declare war or if it needs to be able to use a certain amount of subterfuge in order to build a fleet rivalling the offensives player and acquire a specific target (leading to war strategy AI).

I think i'll clearify mine as well, In my system you never could put it 'I'll sell for 100 or 50.'
With most clauses the choice is strictly binary,
Yes this is perhaps possible, no it is not at all possible.
With other things, such as clauses which are semi-variable peace treaty of 10 turns vrs 20 turns vrs 50 turns, one has want a lot, want a little, would avoid, veto.

and then with resources you have "willing to sell very low numbers, willing to sell low, willing to sell, and I've got plenty"(and their equivalents in the buying)

Now, if you said willing to sell very low numbers, that would limit the max number which could be used to balance. Net effect, the trade is significantly smaller (perhaps a tiny bit more profitable for you, but only the smallest amount, and only if we use non linear equations to solve the min-max curves).

First order approximation: setting those would only determine the magnitude of the trade not the profit per unit sold.

For example, you state you want lasers, and you see they want food.

you could click willing to give Lots of food, run the equation and get something like

Tech for 30 food.
changing to "I value my food very highly" would probably return failour, if not it would at best return
Tech for 29 food.(perhaps, but far more likely to return fail, unless you actually have a huge surplus of food. so that 'I'm saving my food' is mostly ignored by the equation)

But if you were trading foor for minerals it would look like this

30 minerals for 30 food. (low value of food)
7 minerals for 7 food (low excess food, or low need for minerals)
and if you demand lots of minerals for low food, well that's not very diplomatic of you is it, and probably will result in failour. (if not then 7 for 7 again, or 8 for 8 perhaps something like that).



yup it risks no deal, which it turn means it risks [slight] political ramifications, (if we implement such[i would myself])



Where as in my system that is all finished, there is no play room for holding out, the prices are set by the political situation on both sides of the table, although one *might* be able to weasel an additional 1% gain out of the system, it would hopefully not be worth it, and would significantly risk failour. (which I would have block talks for a few turns, and carry some loss of respect between the two nations).



I would not use the same criteria.

I would have something like

1. A system for elegantly building complex 'real'esk diplomatic treaties.

2. Dynamically determined pricing, which reflects both sides production/consumption, both sides access to alternative sources, and to a lesser extent future plans of both sides.

3. Un-manipulatable results. (this is to make human-AI, AI-AI, and Human-Human trades on equal ground). Trying to build an AI which is able to 'guess' what your willing to sell something for or not, and
try to accept/refuse/play until it 'feels' like it's 'won' is probably beyond our scope.
If treaties are 'manipulatable' by refusing them, then two humans playing one against another will no longer be the same as an AI and a human.

4. A system which easily incorporates differences in diplomatic ability/skill/tech's for the empires in question. A non manipulatable system can force even human vrs human trades to be equally one sided as Human AI trades if one side has huge diplomatic ability and the other very little. The low diplomatic side knows it( s/he will always get the short end of the stick no matter who they trade with they will get less for their product because they chose/are a low diplomacy race.)

I would also have 'unlimited'*as long as you don't trigger a failed talks event by demanding too much* number of runs of the treaty talks/ turn. Since an un-manipulatable system means you never NEED to do more then one launch *and won't profit more by doing so. (I would although have the system have a higher chance to fail each time its re-launched in a turn to prevent too much spamming looking for that last half % gain)*.

That's how I'm looking at it . . . and those are my priorities.

Best wishes Robbie Price

PS, those weren't my priorities for diplomacy, they were my priorities when there was a barter system (where you construct a treaty) [which excludes #2]

#2 is a REAL problem, because How does the system know what my plans are... the only way is to self report it, why would I not lie.

#3. I think it is perfectly possible for an AI to guess how much I want something... It just calculates how much it would want it in MY position (as far as it knows my position)... possibly using a 'variable type AI' that has a strategy similar to mine (based on actions am I militaristic, etc.)

#4 is NOT necessary, if we have an Allegiance system. (Diplomacy is basically your political power over various planets, including those of other empires) It would be nice, but is not Needed.


Let's say I'm willing to pay 100 for something, and someone is only willing to sell it for 50, but the "system" says the price is 30... It needs some method of 'readjusting'

Also the Price determined should not be based on my production consumption... it should be based on what the other side KNOWS about my production/consumption. (I don't necessarily want to let you know I'm short of X or have a lot of excess X)


Basically our systems seem the same in that
1. Values of goods in a deal would be calculated based on game state factors (yours=state of the empires, mine=what empires know about the state of the other empire)

2. The players would limit what could go into a deal, and would say what they want out of it (yours=little, minimal, normal, surplus, etc. mine = maximum/minimum absolute values)

3. The deals would be generated and then either accepted or rejected (in your case at increasing penalty over a single turn, in mine at increasing modification of the price)


I think key points in a diplomacy system are

1. player must be able to produce a treaty that is acceptable to both sides if both sides cooperate in working towards it.

2. the player must not be able to use diplomacy as a no cost way of gathering information about an AI manipulate an AI (so you should not be able to propose a treaty you are not going to accept)

3. the offer/counteroffer should be reduced as much as possible (both to avoid manipulating the AI and to reduce micromanagement)

now #3 requires A limit on number of proposal versions youcan reject before no more proposals come that turn

For that part of #3 not to be super annoying there needs to be some way of rapidly reaching what a possible compromise point should be (either an advisor that lets you know a deal is bad, or have all deals constructed by a third party [our systems])

#1 Is important I think though, because the information available to the "constructor" of the agreement is either

A-(player generated proposals)=player knows their empire, their plans and limited info on other empire

B-(system generated proposals)=system knows Both empires, and what info the empires have on each other, and some form of player input

The issue is what info you as a player want to give, and if you can manipulate the system to give a result you want.

I think you need to be able to do that, at a cost. The cost is rejecting the deal for a turn, because that means you REALLY want a better deal, because you actually are willing to give up a worse deal for 1 turn. (the way to tell if you are willing to wait for a better deal is if you wait for a better deal)

Fair enough, i was generalizing, same difference in the end.



The goal would be to make a system which lying doesn't work, or isn't worth it.

If you say your only willing to trade a small quantity it wouldn't change the price, just the volume. If that's not sufficient in quantity then the negotiations fail, and you lose face, or just the chance to have that contract for a few turns. (I would have) having negotiations fail block negotiations for some number of turns.


For 3 I agree. I just have the price be always limited to that Modified AI guess.

If we include races which have bonuses to trade, or bonuses to diplomacy, or techs which affect either of these. . . we'll inevitably need some way of incorporating them into game play. This is just one suggestion as to how. If somebody can suggest a better, I'll be all for it.


This, i think, is were you and i differ most. in fact i think this might be the only significant point of disagreement.

For me, It doesn't matter what the user is willing to sell the tech for, or what the tech is worth to the user.
I say this because if you allow the users interpretations of value play into it forceably then human-human trades will be differnt then Human-AI trades. Since Humans can engage in extra game bartering. and also because humans are naturally build to barter, but building a AI which barters well is almost as hard as building one to pass the Turning Test.

For me, If the user puts a tech for sale, 'the diplomats'(the subroutine) calculates a value for the tech, The value would be determined by such things as; How many turns it would get to research the tech themselves, how high the tech is on your tech tree, how high it is on their tech tree, general empire status, diplomacy ratings, AND the perceptions of each other in those categorizes.

This price then would be the 'fair price'/'best price', given the status of the universe. If the price isn't good enough . . . then don't sell. To me the only thing the user should be able to do to change the price of the tech is to change the condition of the game, or terms of the treaty.

The reason "I" don't want to give the user the ability to wait a turn and get a better price is because i see trades as happening between empires, not emperors(/users). In the end the user give the treaty the final thumbs up or thumbs down, but not more. This forces AI-Human trades to be the same as AI-AI trades and Human-Human trades. Which may not be as important for FO as i think it is. I don't know.


I would have both if both where possible to include. What is real, and what is known, both have importance.

I'm not sure how you would implement this . . . Your method explicitly has people proposing and rejecting treaties. As they do so the rate which the other side gives in by definition gives you information. (free spying), and you can at any point stop proposing, and you've not accepted. . .

I'm not trying to criticize, I just can't get the two to work together in my head, how are you envisioning it working?? maybe I'm not seeing your method as you are. . . do you have some form of negative consequence(other then lost turns) for not accepting treaties, or for abandoning treaties during negotiations? If so I've missed that part of your plan.


Fair enough, If we decide on a final system where wanting a better deal means you can get a better deal, then i concur your system is much more suited for that. If we decide to be more strict with the users, and they get the deals the situation between the two empires merit regardless of the user/emperor/race council 'wants', I think a system more like mine might be more appropriate.


In the end this sort of fundamental difference in direction might have to be put to the wider community.


Krikkitone and I have rather dominated this area for a while, are there any other Methods for approaching trade/diplomatic treaties which people are preferring which have not gotten much airtime??

m_k, if your still reading these what are your thoughts now?? Anybody else??

Best wishes all. Robbie.

Well summing up some of the ideas for others. I think we are looking at four basic systems
[in all cases, player input impacts "what's on the table"]

1. RP's:
Values for Deals are calculated based on game state values, player input as to value can be included but risks no deal

2. K1 auto:
Values for Deals are calculted based on game state values, player input is primarily done through rejection of deals

3. K1's barter:
Deals are proposed by a player, proposing a deal means you are bound to accept it. Only one (or some small limit) of proposals may be made a turn.

4. Standard:
Deals are proposed by a player, but even if the other player accepts the proposed deal, you must also accept it.
[this has the bad effect of no cost diplomatic spying]


In terms of diplomatic spying, one thing No calculation system can include is player plans (yes it can know the AI plans, but it can't know player plans so those are part of the system)


RP Quote
"In the end the user give the treaty the final thumbs up or thumbs down, but not more"

I am saying that too... but I think that thumbs up/thumbs down should affect the value of future treaties (since that thumbs up/down includes the 'player plans' part of the value... which is Very important especially for things like peace treaties.)

And that is the fundamental difference we put to others.

That pritty much sums up the situation. I would be a bit more precise on 1 and 2,

1. Sides input to what degree(magnitude or general willingness) they are to have each clause in the final document. Exchange rates between each clause, in there basic form (/resource, /treaty_day, etc...), are calculated by game state, and can not be altered by a user without changing significantly the treaty or the relationship between the two empires. There is little, or no, room to gain 'extra profit'.

2. Sides input max/min absolutes, clause absolutes, and clause vetoes. Program calcs a midpoint between the two offers on the table. Both parties are then presented a semi midpoint offer, and they may chose to accept or reject this midpoint offer. The rejecting of a midpoint offer is remembered by the algorithm and next turn, if possible, a more tempting offer is presented, (within the bounds of both sides min/max's of course). The midpoint offers take into account both the desired prices of all parties, and the value as calculated from game state, with some emphasis on what each empire thinks they know about the other.



Krikkitone, would you agree with my summery of the first two?

Best wishes again

Pretty much

One possible solution would be to implement this in a rather simple fashion stemming from the idea that even the most dissimilar races can get acquainted to each other (like the most dissimilar human characters aka people can) when knowing each other for a long enough period of time.

The feature is called "Acceptance" and it's meant on a population level, i.e. if your population accepts the alien counterpart. At first encounter you always have 0 acceptance since you basically don't know anything about that new strange alien being other than its existence. By signing a friendship treaty, that acceptance begins to increase with a speed which is determined by certain race characteristics. We have a set then of let's say 8 characteristics where each race can only have one (its major characteristic) so a farmer and a war-like race will not get acquainted that fast as two scientific ones for example. It's all determined in a 8x8 matrix.

So by having long agreements, that acceptance grows and grows and at some point it ain't possible to declare war on that race anymore (the utmost consequence would be your usurpation as leader of that empire so in fact game over then). Of course this acceptance can develop in reverse when bad accidents (random events or intel-induced deliberate plots and propaganda) happen but they take a while, i.e. some turns, to actually influence that acceptance percentage.

Of course, acceptance can also take negative values and then lead to the fact that you could not offer a peace treaty anymore to that race when having too high negative acceptance (for example when your population suffered a lot of casualties and losses in a long-waged war against that race whereby the threshold for this point of no return is depending on the race's specific nature, i.e. peace-loving farmer and scientific races will always be able to call for peace since they are basically willing to forgive or let rational arguments count about future losses for example).

That way, you have a certain continuity in your diplomatic treaties and relationship development but at the same time keep the possibility to revert a current state of "uber-friendship"/"uber-acceptance" by using intel propaganda methods (which btw. can be counterattacked, detected and interfered by your friend's/enemy's intel so he knows about that treacherous development provided he's producing enough intel and allocate some on inner security resp. on spying its friend's empire).

I think that's pretty elegant and I hope it solves all problems regarding diplomacy. Gifts and thing have no effect on acceptance, only at your relationship barometer with that race (that barometer is not the same as population acceptance, it only shows what the empires' leaders(!) think of you). If that barometer and population acceptance differ too much, then negative morale strikes both empire's systems so they are forced a bit to do what their peoples demand. Of course, a race with especially big intel production and capabilities can mitigate this effect by propagandizing its population.

In all other terms of diplomacy, an approach much like good old Birth of the Federation, so nothing fancy in fact, would be best I think for starters.

Goodmorning all

to M4IV, you might find that the thread on simulating citizens is more closely related to what your trying to achieve ... your comments are more about the response of citizens to a given proposal then the value of proposals and their exchange rates and how to build them on a pre-User interface level. Which is more along the lines of what this thread was trying to work out. . .


To everybody else, has anybody given any further thought to this thread? does anybody have an preference either way?

Best wishes, Robbie Price

Don't understand. My system is pre-User interface level. The value of proposals and their exchange rates should be dealt like Birth of the Federation did (I hope everyone knows that game). For exact values I'd need to be a regular programmer here not to tamper unknowingly with other balance aspects of the game.

I have read through the simulating citizens thread but the underlying system here is somewhat different plus it's related more strongly to diplomacy than the other thread.

I think Geoff's 3rd wanted thing in diplomacy can be reached by what I described here.

My apologies, I was rather curt previously, i more detailed explanation:

a very similar system, though not nearly as strict, has been proposed, and is under discussion regarding the response of civilians to diplomatic decisions.

In short, rather then having a new meter, your citizens would like or be upset with political choices you make, before you decide to offer/or declare a political decision you are informed 'generally' how your people will respond, which can be as negative as to cause happiness to drop low enough to cause riots. Citizens would 'remember' previous political activities, and have bias for and against various other empires and respond accordingly.

Hence why i suggested looking under citizens.

Best wishes, keep up the good thinking

Robbie Price

A few more ideas that might be nice to have in our diplomacy model.

*A Direct way to request help from an AI empire

I find it a bit irritating that in most strategy games that I have played, even though you have a defensive alliance/full alliance with an AI empire and they declare war to your enemy, it doesn`t often actually mean that they are really fighting against this common enemy. They might just be in a state of war against this enemy empire, but they aren`t really attacking it, or if they are actually attacking the enemy empire there is no way that you could ask them to aid you in defending your star systems, or plan a coordinated strike against this enemy, in which both empires participate. I have also always appreciated AIs that are capable of sending some troops to aid in my war effort, even if this usually has been more of a coincidence than a planned action.

So it would be nice to have some diplomatic options that could help co-operation of allies/friendly empires during war times. These options might be something like the following examples:

- Send troops to defend against/fight off enemy troops from my empire.
- Send troops to defend against/fight off enemy troops from sector X. (if there will be galactic sectors.)
- Send troops to defend against/fight off enemy troops from star system X.
- Support my attack on enemy star system X after Y amount of turns. (This might be a little difficult to implement, since there would have to be some sort of an auto calculation mechanism to ensure that the player can truly attack system X after Y amount of turns. Also when the AI would response to this kind of a request it would also be useful if the game could calculate how much time the AI`s fleet needs to arrive at that system. So that both fleets would be able to attack the system at the same time).

This way you could perhaps also evaluate how useful your allies really are and act accordingly.

* Large scale war plans.

It might be nice to have an option to make large scale war plans with your allies/friends before you attack a common enemy. This way you could for example split the enemy`s territory in half before attacking it, thus making it clear to both sides where to focus their war effort. This kind of an option might also be helpful if you are already at war with a hostile empire and want to determine areas of responsibility with other empires that are at war against this enemy, thus making it clear to all participants where to focus their efforts.

_________________
What is your favourite alien species and WHY?
Preliminary thoughts about diplomacy
Some unfinished ideas for specials

I'm for #3. It's pretty much what I had in mind. Making a bartering AI can't be that difficult; They know what they want and they know what they're willing to give for it. All we have to do is make them return the deal the first time at a given comparative value ratio (ex. they are getting 1.5 times what they are giving) and gradually lower it until it is even (or slightly higher). Naturally, if the ratio was greater than this predetermined (slightly random each time?) ratio, they would accept on the spot, and either side would be free to reject any offer or counter-offer made by the other player, which for the AI, would occur if it is getting way less than the other player in the deal (ex. they are getting .5 times what they are giving) and as more back-and-forths happen, the value necessary for them to outright reject would become less and less extreme.

So the player would input exact values for what he is exchanging, human - AI barters could not go on forever, and human players cannot adjust the treaty once the AI has accepted. Players have more direct control over diplomacy than in #1 and #2, which, IMO, makes it a lot less like MoO3 (in general, not the diplomacy system in particular {which also sucked, but in a different way than the rest of it}).

_________________
Warning: Antarans in dimensional portal are closer than they appear.

I'd also like to point out that we should generally make design decisions from the perspective of human-only games, and take for granted that we'll be able to program an AI to deal with it. Any decisions made regarding the diplomacy model should only take into account human players. We have years and years to make a great AI, after all.

_________________
Warning: Antarans in dimensional portal are closer than they appear.

Actually to Barter you need to know 4 things

1. What I want
2. What I am willing to give
3. What YOU want
4. What YOU are willing to give

Now for #3 and #4, I need to know it, I can't trust you to tell me (because you will lie)

any AI would need to be programmed to deal with that.

The ONLY reason a deal takes place is because different parties value something differerntly

If Planet X is worth 100$ to ME and it is worth 90$ to YOU, then I can buy Planet X from you for 91-99$ and we will have a deal.

A Bartering AI will try to get that as close to 91 as they think they can (assuming they are buying, or as close to 99 as they think they can (assuming they are selling)

I must not only make a Profit, (What I get > What I lose)
BUT
My profit must be as big as yours ( My net gain >= Your net Gain)
..especially since this is a competitive game.. but also because otherwise, I could have squeezed more out of you.

I think Avoiding an exploitable AI requires an AI that Lies in diplomacy (you ask them, how much will you give me for planet X.... they will give you the lowest value they think you will accept, assuming they are willing to pay that much.... but if you offered them a deal with a higher price they would accept)

This means you need an advisor that provides a "this is probably how much they are actually willing to pay" value, based on your knowledge of their situation.

This also means "proposed treaties" should be binding on the proposer.

So instead of asking "what will you give me"? You ask... Will this deal work (after your 'diplomat adviser' already gives you what your in-game/empire knowledge says is a good deal... and then you modify that based on your strategic out-game/gameplayer knowledge.)
Then you take their Response and it adjusts your strategic knowledge... and potentially your 'diplo advisers' knowledge (since your 'diplo adviser' only knows what your empire knows, and you don't have complete knowledge of the other empire's current state).