Diplomacy/AI intelligence

[cccv]

If the "shopping cart" option honestly was the best way to help them, sure it would.

Granted, the best way to fix the problem is to make the AI less unreasonable to begin with, but that wasn't exactly an option in IV. Hopefully in V they've accomplished that to some degree. However, we do know that the AI will still be making demands about how we run our empires. The way we're hoping for this to be done reasonably is to give the player the power that the AI has. If they can say "stop expanding," we need to be able to say it, otherwise they can shackle us then do as they please with no consequence. If they can say "keep your troops away from our border," we again should be able to say the same, so they can't expel our troops and then launch an attack themselves now that we're out of position. That sort of stuff. Redlining is just another potential thing that may help us have more equal footing.

 

[Ahriman]

An alternative fix; make the relations hits temporary. If the AI asks for tribute, you can accept or refuse. If you accept, good relations for 30 turns. If you refuse, large relations hit for 10 turns, good chance of war... unless they were bluffing. But then, relations return to normal.

Accepting/refusing tribute should be a short-term threat; no-one should be holding your refusal to pay tribute against you 100 turns later.

I'm guessing of course that no-tech-trading makes 90% of this problem go away by itself. Demanding some gold, or less settlement, or your armies to move away, or whatever, is quite different from demanding a tech (that benefits them forever).

 

[Aramel]

If the "shopping cart" option honestly was the best way to help them, sure it would.

Granted, the best way to fix the problem is to make the AI less unreasonable to begin with, but that wasn't exactly an option in IV. Hopefully in V they've accomplished that to some degree. However, we do know that the AI will still be making demands about how we run our empires. The way we're hoping for this to be done reasonably is to give the player the power that the AI has. If they can say "stop expanding," we need to be able to say it, otherwise they can shackle us then do as they please with no consequence. If they can say "keep your troops away from our border," we again should be able to say the same, so they can't expel our troops and then launch an attack themselves now that we're out of position. That sort of stuff. Redlining is just another potential thing that may help us have more equal footing.

No, redlining is not a power the AI has. I have explained this again and again. It is just a shortcut for the convenience of the human player. Redlining is effectively just rejecting a demand under all circumstances. The human player can essentially already do this; if, for example, you just decide for yourself that you are going to reject any demand about giving up your iron, just reject all deals involving it personally.

The only difference between redlining and what the human player can do is the fact that the rejection is preemptive. For most deals, this makes no difference, because most deals carry no penalty if they are rejected. Demands are the exception, as rejecting a demand causes a relationship penalty for the human player. Thus, preemptively preventing a deal could be useful for the human player. But just because something is useful is not a good reason for the human player to be capable of it. It would be useful for the human if they could reject all wars preemptively, but that would not lead to good gameplay. If, however, the AI got some kind of advantage from redlining that the human player didn't have, that would be unfair and people would be right for complaining about it.

But does the AI get some advantage from being able to preemptively reject demands? No it does not, at least not as far as the human player is concerned. The reason is because AI demands and human demands are actually very different. The AI has a relationship score that to some extent controls it's actions; the human does not. When the AI makes a demand of the human player, rejecting it affects the AI's relationship score towards that player. However, when the situation is reversed, there is no relationship score for the human player to be affected. Thus, whether or not the AI rejects the deal, it suffers no in-game penalty from this. No relationship score is affected. Thus the AI has no reason to prevent itself from receiving demands. It receives no penalty from rejecting demands. In fact, the only reason it ever agrees to demands lies outside of the game mechanics entirely and has to do with the player's emotions, but even then redlining offers no advantage because the human could become angered at a resource being redlined just as easily as it could at a demand being rejected. So the AI gains no benefit from preventing demands being made of it; in fact, it would have every incentive to just let you keeping making demands and ruining your relationship with it, because you would only be hurting yourself. If redlining gives an advantage to anyone, it is the human player, because it prevents him from making demands that would be automatically rejected, and thus from ruining his diplomatic relations with no possibility of gain.

The only way redlining could possibly offer the AI an advantage is in preventing it from receiving demands from other AI's. If redlining actually does let it avoid diplomatic penalties here, that would be unfair, and would represent a design flaw. However, the solution would not be to let the human player redline, it would be to make the AI suffer those penalties regardless of whether it redlined or not.

Again, redlining gives the AI player no advantage, at least as far as negotiations between humans and AI's are concerned. To give the human player redlining is to give to the human an advantage that the AI doesn't have. There is no reason for this; it would just make things more complicated without improving gameplay at all. Again, Redlining is not an advantage for the AI. It is a convenience for the human player. That is all it does.

 

[cccv]

I could point out that if we had the ability to redline it would also be a convenience to us. I know I got annoyed in late game where I had to tell someone to piss off every single turn. It usually got to the point where I'd even stop looking at what they were asking for and end "negotiations" immediately, friend and enemy alike. It sure would have been nice if the AI knew not to bother to ask.

However, when the situation is reversed, there is no relationship score for the human player to be affected. Thus, whether or not the AI rejects the deal, it suffers no in-game penalty from this. No relationship score is affected. Thus the AI has no reason to prevent itself from receiving demands. It receives no penalty from rejecting demands. In fact, the only reason it ever agrees to demands lies outside of the game mechanics entirely and has to do with the player's emotions, but even then redlining offers no advantage because the human could become angered at a resource being redlined just as easily as it could at a demand being rejected. So the AI gains no benefit from preventing demands being made of it; in fact, it would have every incentive to just let you keeping making demands and ruining your relationship with it, because you would only be hurting yourself. If redlining gives an advantage to anyone, it is the human player, because it prevents him from making demands that would be automatically rejected, and thus from ruining his diplomatic relations with no possibility of gain.

Hmm? This makes sense if you assume the AI wants to have a bad relationship with us. But when they make demands and we reject them, the relationship is hurt and they decide to cancel deals that hurt them too or get into wars they can't possibly win. The AI shouldn't want to hate us no matter what. Oh, and as far as how we can be "angered" by their redlining, that just leads to diplomacy that makes no sense. They're allowed to piss us off to no end, yet our relationship is at "pleased."

Anyway, I've thought about how I'd like demands to work in V and this is what I've come up with:

The AI shouldn't be able to make unreasonable demands (large % of our gold, make war [especially when our military strength is low], cancel deals with friends, or give an important strategic resource) unless it means enough to them that they're honestly prepared to go to war with me if I don't give in. In IV, Ghandi could pester me over and over for something and bring our relationship down, even though I knew he could do sweet crap all about it. In V, I'd like Ghandi to keep his mouth shut unless he can back his threats up, and unless what he wants is so important to him that he may do so (like if he desperately needs the gold, or maybe he's thinking about going to war with me anyway and he's giving me a chance to appease him first). I should know that if I'm not going to give him iron, there's an XX% chance (and I don't need an exact number) that his war elephants will show up and try to take it. But all of his grumblings about how he wants me to cancel deals with my friends? If he isn't prepared to fight me if I don't make that a reality, he should stay home. Otherwise diplomacy would be like it was in IV-- "oh, Ghandi wants something" *end convo.* ...Gee, I hope he wasn't just offering me all of his cities >.>

 

[Dermo]

What I find frustrating about diplomacy in Civ4 was not being able to see how a diplomatic event might affect other leaders. There are many times where the AI would offer a trade/demand and I would have liked to have been able to go to the diplomacy screen and get a better idea of the 'big picture'. There should be a way to access the diplomacy screen from the trade window.
Or at least the option to think about teh AI's request for a turn

 

[Roland Johansen]

but if the system is as complex as you want it, then those numbers will be largely useless anyway. In my diplomatic relations in Civ4, even when I don't look at my relations I *know* there's a hierarchy of things on the like/dislike list. Obviously going to war against a Civ/going to war with a Civ, is going to almost always be at the top of that list. Religion was always right below it & below that was usually breaking/making trade deals. Refusing trade deals or demands was usually below that, about equal with close borders sparking tensions.

Perhaps again the best approach, short of actual numbers, is for the diplomatic advisor to show you the most *important* diplomatic factor currently at work at the top of the list. This could be a function of both its importance to this AI & how recently the diplomatic action occurred. So, for example, if Napoleon has a Militaristic Flavor, then moving troops up to his borders will probably be the key thing pissing him off-even if you did it awhile ago, & so this will probably remain at the top of the list for a while (in red to indicate its negative impact on relations). However, a commercially oriented Civilization might be more pissed off with you for breaking a trade pact, especially if it happened recently, & would therefore be at the top of their list. It would help even more, of course, if the leaders themselves also *tell* you how they feel about you-& what the primary cause is. I just don't think exact numbers are necessary to convey that information!

My post was about the claim made by various posters (amongst which you) that a numerical display of diplomatic modifiers would lead to an AI which could be easily manipulated while removing those values would stop this human ability to manipulate the AI.

It's a somewhat logical thought as the civ4 AI used a numerical display of attitude modifiers and it was easily exploitable. However in two easy examples, I have shown this absolute relation not to be true. I hope that we can now continue to discuss the merits of the various ways to display the AI's attitude towards us without repeating this claim.

By the way, I guess we have at this point no idea at all whether the civ5 diplomatic AI will be easily exploitable or not. We didn't know that about the civ4 diplomatic AI a few months before its release. We actually found that to be true only a few months after its release when several players had (for instance) discovered that friendly AI's would never start to consider declaring war.

I at least hope for a diplomatic model that is transparent in a sense that we can easily see how all AI's feel about us (at a glance preferably) and why they feel that way about us. Furthermore, I would like to have an idea how much future actions might affect that diplomatic relation. And finally, I hope that this diplomatic relation cannot be exploited too much and doesn't dictate the AI's actions. We do have some indication that this last element might be true as AI's can 'lie' about their intentions according to previews.

In my view, numbers are a very easy and natural way to display such stuff. It's seems to me that some people just have a natural aversion to numbers and everything they represent.

 

[Ramesses]

The only difference between redlining and what the human player can do is the fact that the rejection is preemptive. For most deals, this makes no difference, because most deals carry no penalty if they are rejected.

Alright, I see where the communication breaks down.

We're talking about an existing mechanic in CivIV which, as it is applied in-game, bars the player both from trading for it or demanding it. However, the entire discussion revolves around a single aspect of redlining - specifically, the ability to prevent the human player from demanding resources, technologies, and civic/religion changes.

This, right here, is what we want to be available to the human player. We don't mind if they ask to trade for something of ours, even if the deal is ridiculously unbalanced since, as you say, there is no penalty for declining. What we DO mind is the fact that an AI will demand tribute for an item repeatedly, and take multiple diplo hits in the process, all because the item they keep demanding is something that you don't want to part with, except for massive compensation. It gets even worse when they demand something that you are willing to trade but not willing to give away, because often you are then stuck between: a) giving them the thing, then not being able to demand compensation in return; or b) refusing their offer, then not being able to trade it because now they don't like you.

So maybe the solution isn't to give the player the ability to redline, but a different kind of mechanic. I suggest something I call "yellowlining", which means giving the player the power to mark items as being unavailable for tribute to all AIs. The AI cannot demand anything that you yellowline, but they are free to try and make a deal for it.

To prevent abuse of the system (namely, yellowlining everything), a few restrictions need to be put in place. I'm not sure what those should be yet, though.

 

[Aeon221]

I'd like it if the diploscores tended towards zero in the medium term, so that keeping on friendly or hostile terms with the AI required regular effort, be it with insults and threats or goodies and happies.

I want power blocks to form, last for a while, decay, and fall apart. I want an ancient rivalry to remain in effect only because the modern players are still pissing about over modern issues. I want the AI to be like the player and care naught for friends, enemies, past insults or past treasures, only about who it needs to love or hate to win the game.

More feedback, and more immediate feedback, would be extremely handy. I should be informed that I've pissed someone off when it happens, not the next time I go for some trading. But, really, it's all about making the AI forward thinking by making it a partial amnesiac.

 

[mjs0]

I'd like it if the diploscores tended towards zero in the medium term, so that keeping on friendly or hostile terms with the AI required regular effort, be it with insults and threats or goodies and happies.

I agree with this in general...I would like to see any negative or positive 'feelings' the AI may have towards me decay over time, however I would prefer they not trend all the way to neutral but rather back to some small but non-zero residual effect.
So, a handful of border skirmishes or diplomatic incidents several hundred years ago would have very little effect on our current relationship, but the residuals of a very long history of diplomatic and/or military hostility would accumulate over time to leave a small but significant undercurrent of distrust, not to the level of scuppering real efforts to be friendly but to a level that prevents the default relationship (i.e. without recent efforts to be friend or foe) being neutral.

I want power blocks to form, last for a while, decay, and fall apart. I want an ancient rivalry to remain in effect only because the modern players are still pissing about over modern issues. I want the AI to be like the player and care naught for friends, enemies, past insults or past treasures, only about who it needs to love or hate to win the game.

Yes, but with reservations...as I said above a genuine long lasting rivalry should leave some residual ill will, even if it is ancient, just not at (or even close to) it's original intensity.
I'm also not sure about the characterisation of the AI, it actually depends on the type of game I am looking for...sometimes I want to roleplay and have the AI do the same...sometimes I would like the AI to behave more like a genuine competitor. Perhaps an option could be available to allow for that choice???

More feedback, and more immediate feedback, would be extremely handy. I should be informed that I've pissed someone off when it happens, not the next time I go for some trading. But, really, it's all about making the AI forward thinking by making it a partial amnesiac.

Yes, instant feedback within the game world, i.e. without breaking immersion, would be great...even if it's simply my foreign advisor telling me I may have made an enemy today.

 

[tom2050]

Civ is not being dumbed down... that really isn't the right word. It is headed away from micromanagement and towards a macro style. This is good in many ways, as it lets you do everything much easier and without the excessive clicking. Although my perference is somewhere in the middle (leaning toward micro side); I fear they may go to far with the macro to get the casual gaming audience. But doing so would be good for them to expand their audience.

In all respects, the programming of the game and models are much more complex than any previous iteration (I would guess so at least).

I don't like having everything laid out for me so I can see what the AI is thinking. Diplomacy, unfortunately won't be a whole lot different anyways; perhaps they will do something with it in an expansion or DLC pack. They went for graphics instead of gameplay in this portion of the game. There will be some cool things still; they can't do everything, so no big deal.

But if they do make a big diplo change, a semi-dynamic model would be incredible and bring much more to the game than hexes or 1UPT could ever do. Although the sales pitch may not be as big, it would make game immersion ten-fold to what it is now.

* Knowing you have a trustworthy ally that you can count on and having trades and agreements go accordingly (as long as you do the same).
* Seeing a country/or group of countries evolve over time from a neutral stance to a war-mongering stance; in which you can choose to join, or go against.
* If country is getting close to nukes, the world can either try to stop it through diplomacy or war (if the country is seen as having bad intentions), or let them do as they wish.

There are hordes of incredibly interesting dynamic things that could happen with such a system. It may be difficult, but it would be absolutely an amazing feat. Plus, modders could continue to add to it as they wish.

I don't understand why they don't see (or maybe they haven't let us know) what a great thing they are missing with this. After all, diplomacy is the biggest thing civilizations have had in respect to everything throughout history. It should be a primary focal point of development.

 

[cccv]

I want the AI to be like the player and care naught for friends, enemies, past insults or past treasures, only about who it needs to love or hate to win the game.

I disagree with that one. I want civ to feel more like I'm building an empire alongside other empire builders, and less like we're trying to win a game. I know that's me and we can differ on that opinion, but I'd just like it to be a little more true to life-- two countries will be allies and know they can count on each other to a point if needed, and not fear that, say, Canada is suddenly going to try to invade the US for a last minute "domination victory" >.> So more true life-ish (the "ish" because no, I don't think civ needs to be 100% realistic, I don't give a flying crap if it takes a unit 10 years to move from one city to another or what the crap ever), and less civs trying to win a game. Honestly-- and I may be in the minority for all I know-- I will not suddenly attack someone I'm friendly with just because I realized I could use his cities for my victory. It's just a computer game, but damn it, they're my friends! I honor our computer friendship. ok, I'm pathetic >.>

 

[cccv]

Civ is not being dumbed down... that really isn't the right word. It is headed away from micromanagement and towards a macro style. This is good in many ways, as it lets you do everything much easier and without the excessive clicking. Although my perference is somewhere in the middle (leaning toward micro side); I fear they may go to far with the macro to get the casual gaming audience. But doing so would be good for them to expand their audience.

What they're saying-- and you can believe them or not-- is that the micromanagement is still very much possible. They just made it so the average person can play the game without having to go into the city screen, but rather leave it all up to the governors, but once that average joe gets into the game they'll learn more of the subtlety of it and then will start going in and micromanaging more. So it's not so much that they're moving toward macro-management as it is that they're letting people get away with doing just that, but slowly leading them into the more complex stuff which is still there. Anyway, that's if you believe them and I'm not sure you will.

I don't like having everything laid out for me so I can see what the AI is thinking. Diplomacy, unfortunately won't be a whole lot different anyways; perhaps they will do something with it in an expansion or DLC pack. They went for graphics instead of gameplay in this portion of the game. There will be some cool things still; they can't do everything, so no big deal.

Actually, they keep claiming that diplomacy is the big thing they concentrated on. They won't shut up about it. It sounds like everything they did-- from city states to the removal of religion-- was done specifically with diplomacy in mind. So I'm hoping that some of my concerns will be put to rest when the game comes out. Because everything else you mentioned sounded great-- dynamic diplomacy, all of those bullet points you mentioned-- hopefully some or all of it will happen. I don't know how much. I suppose that's why I made the thread. I wonder just how much of this will happen, and how much is even possible, or if we're just going to have a somewhat more sophisticated version of IV but with a lot of the same annoyances. The same sense of entitlement from the AI, marching into your palace saying: "I'm king of the world and screw you," same bot-like behavior, even if to a lesser degree, and will we have the ability to "read" them or not? Will we know when it's ok to grab another tile in their direction?