Poll: what makes a good diplomacy system?

[KrikkitTwo]

This seems overall reasonable. I'm not sure about the gradual increase and decline though.
The effect would be clearer to the player if it was instant.

This would balance war weariness nicely. War weariness gradually gets worse and instantly goes away, this effect would instantly happen and then gradually go away.

Gradual increase wouldn't be on the Happiness
Gradual increase would be for parts of the "Relation Value" like "we have been at peace/had a trade treaty" to make the impact of peace or a trade relation better with time (A treaty I signed ten turns ago has less impact than one I signed 100 turns ago). (Harder to attack my long time friends than my short term friends)

The general idea is to
1. allow the AI player to play like a human
2. give the human a reason to play like a 'normal country'

 

[Kazimierz Wielk]

I do like the idea of penalties as related to population preference, especially in government types that are representative. Even if you are President of the US, you can't unilaterally declare war on Canada (for it's Prairie Provinces Oil Reserves). Wars can be, and have been, unpopular in various countries. If you go against your people's will, there should be an impact. If you choose to do that as a human player, you can, but you deal with the penalties.

I know some people have complained that they don't want a "The Senate Does Not Allow Your Declaration Of War" from previous Civ's. To be honest, in a democracy and republic, it is totally realistic. Autocratic leaders don't need approval for their actions, but still must deal with their population's attitudes and reactions. Such are the makings of revolutions. Such are the makings of regicide (many Roman Emperors are examples).

The "I don't want to be directed by my people while I am a human player" argument isn't that strong, because we are confined by our people in Civ4 to a certain degree. We have cities of different cultures, which we must work on by building buildings. We have cities with different religions which we try to alter by sending different missionaries and choosing which temples to build. We are still, to a certain degree, constrained by our populace and we mould that populace over time to better align with our "human player" desires.

 

[Akka]

"playing to win" is a purely meta-gaming point of view.
If the AI is supposed to represent a civilization, it should act like a civilization, NOT like a player trying to win a game.

Of course, the best is when the goals of the game organically derive from the principle of what the game is about ; for example, an AI "playing to win" in a basketball game is totally okay - that's the POINT of basketball.
But when the principle of what the game is about conflict with a meta-gaming view on the action, then the first should trumps the second ; for example, a far weaker civilization declaring war on you because you're about to "win the game" by launching your spaceship is just stupid.

 

[Che Guava]

"playing to win" is a purely meta-gaming point of view.
If the AI is supposed to represent a civilization, it should act like a civilization, NOT like a player trying to win a game.

Of course, the best is when the goals of the game organically derive from the principle of what the game is about ; for example, an AI "playing to win" in a basketball game is totally okay - that's the POINT of basketball.
But when the principle of what the game is about conflict with a meta-gaming view on the action, then the first should trumps the second ; for example, a far weaker civilization declaring war on you because you're about to "win the game" by launching your spaceship is just stupid.

Well, there's inevitably some late-game silliness when it comes to the space race, but I think think that's pretty much unavoidable if you want to have competitive opponents, but on the larger issue, I think it should be approached from the other way around: you should have a better chance of winning if you act like a real civilization, rather than the 'world domination' focus that seems to occur. If there were more downsides to large empires and war-mongering, I think you'd have a much more complex game where civs would probably stick closer to home and use diplomacy and trade more.

I know some people have complained that they don't want a "The Senate Does Not Allow Your Declaration Of War" from previous Civ's. To be honest, in a democracy and republic, it is totally realistic. Autocratic leaders don't need approval for their actions, but still must deal with their population's attitudes and reactions. Such are the makings of revolutions. Such are the makings of regicide (many Roman Emperors are examples).

The "I don't want to be directed by my people while I am a human player" argument isn't that strong, because we are confined by our people in Civ4 to a certain degree. We have cities of different cultures, which we must work on by building buildings. We have cities with different religions which we try to alter by sending different missionaries and choosing which temples to build. We are still, to a certain degree, constrained by our populace and we mould that populace over time to better align with our "human player" desires.

I, for one, would welcome the return of the senate

 

[GhostWriter16]

I'd be happy to see a general modifier that applies to all relations you or the AI has with all other civs that relates to 'trustworthiness', i.e. if you backstab a close ally, ALL AI civs are less likely to have good relations with you, which would be a good deterent to underhandedness.

Apologies if that actually does exist, I'm not all too familiar with how the modifiers work.

Well, it doesn't exist. It ought to though. Instead, in Civ 4 you're forced to keep your word. Honestly, you should even be able to sign open borders and walk in and then kill them because that's the way it really is.

 

[Akka]

I think it should be approached from the other way around: you should have a better chance of winning if you act like a real civilization

That's precisely what is implied in the "Of course, the best is when the goals of the game organically derive from the principle of what the game is about" sentence
It's up to the designers to make it so that game goals are closer to "what the game is about" goals.

 

[Dale]

What I prefer is a system mostly like Civ4, where the human player can declare war whenever they want, but AIs are much more likely to declare war on enemies than on friends.

Is this an advantage for the human player? Sure. But it is simple and makes the game more fun. It makes the AIs feel more like countries than Deathmatch Bots, but still allows for full player agency, where the player doesn't have to fight against the will of their own populace (a la Civ1 Democracy declaration of war failures).
This is how its nearly always worked in Civ (early exception for democracy/republic), and in the vast majority of strategy games with diplomacy engines.

So my preference is for the AI to play like a limited AI, and then crank up the difficulty level so you are still challenged.

I completely disagree with this. The AI in a game should NOT be designed to play like that. The AI in a game should be designed to play to challenge you. To have an AI the way you describe would make anything past mid-game mind-numbing.

"How so?" I hear you ask.

Ok take the Space Ship example used a couple times already. You start building the space ship. You know the space ship wins the game for you. You know that the Ahriman AI will not declare war on you if friends, so you bury the AI's in money gifts to keep them friends and sail to victory.

Or how about culture. You know you need three cities to have extreme culture (Civ4). You know that wins the game for you. You know that the Ahriman AI will not declare war on you if friends, so you bury the AI's in money gifts to keep them friends and sail to victory.

There's even conquest/domination. You know that you have to defeat the other Civs in war. You know that the Ahriman AI will not declare war on you if friends, so you bury the AI's in money gifts to keep them friends, pick them off one by one by backstabbing them and sail to victory.

I'd rather an AI that notices what you're doing and acts to challenge you. If I'm nearing a space ship victory, I'd like some of the AI's to stop trading with me, demand me to stop building the space ship, even try to take over the cities I'm building the space ship in. Or if I've got three cities nearing culture victory I'd like some AI's to challenge my ownership of them, or to put 100% of their effort into beating me with their own three cultural cities. If I'm taking over civ after civ, I want some of the AI's to fear that, form a block and counter me.

I don't want a lie down, stand on AI, which is exactly what you're proposing. Keep them friends, and there's no challenge since you completely 100% control what happens.

 

[Che Guava]

That's precisely what is implied in the "Of course, the best is when the goals of the game organically derive from the principle of what the game is about" sentence
It's up to the designers to make it so that game goals are closer to "what the game is about" goals.

And what I meant by my post is 'I agree, nothing more to add to this'

Sorry, my reading comprehension is atrocious this evening....!

 

[Aussie_Lurker]

So Ahriman, lets see if I understand you correctly-you want the human (you) to have free agency to do whatever you want-with *no* penalties-whilst the AI is confined to pre-determined behavior *and* you want foreknowledge of exactly what those confines are. Sorry, but that sounds totally lame to me, kind of like saying "I'll take on Deep Blue at chess, but only if it can't move its queen, & only if I can know its next 4 moves in advance". Sure you might win, but it'll be a hollow victory. I say give everyone "free agency", but with the knowledge that there are penalties for totally illogical/irrational behaviour. Obviously the degree of penalty suffered would depend on your government type-less representative governments will attract fewer penalties domestically for attacking a long-time friend, but then less representative civs should always be less trusted by players & AI alike-especially AI's/players who have more representative government types. In a representative government, attacking a civ that has a lot of their culture in some of your cities should come at a cost in those cities (just as was the case with religion in CivIV-"we are angry that you're attacking our brothers in the faith"). Trading with a civ that is diametrically opposed to yours in terms of government type should also draw ire from your people-as should trading with civs of different cultures, under certain civic types (nationalist, protectionist). By contrast, trading with civs with whom you share cultural/diplomatic bonds-or shared political outlook-should gain you extra happiness in your cities. Attacking a civ with whom you've had bad relations with in the past (either continuous or on & off) should draw no penalty domestically, & attacking a civ that is diametrically opposed to your own in any of the various "civics categories"-to use a Civ4 parlance-should actually give you a "Cassus Belli" happiness bonus (& diplomatic bonuses with like-minded civs), particularly if you're running representative civics. Many, if not all, of the examples I've provided have at least some basis in real history, so the only reason I can see to oppose their implementation is just because you want a game with no real challenge. If so, then might I suggest that a "Churn & Burn" RTS is probably much more up your alley!

Aussie.

 

[Earthling]

The first choice over the second, but I voted the question lacked complete phrasing/options, because it didn't even begin to indicate the importance of simply have an overall intelligent AI and lack of overboard handicaps/bonuses.

Also, Kazimierz Wielk reflects a lot of my sentiments/things I'd agree with anyway so I'd really like to have reasonable systems restraining a human leader/reflecting treatment and wishes of citizens. Of course they've cut religion and so on entirely so I guess there's at least not going to be the silly downsides there (much potential was lost with the AP and mechanics in general),

 

[KrikkitTwo]

"playing to win" is a purely meta-gaming point of view.
If the AI is supposed to represent a civilization, it should act like a civilization, NOT like a player trying to win a game.

Of course, the best is when the goals of the game organically derive from the principle of what the game is about ; for example, an AI "playing to win" in a basketball game is totally okay - that's the POINT of basketball.
But when the principle of what the game is about conflict with a meta-gaming view on the action, then the first should trumps the second ; for example, a far weaker civilization declaring war on you because you're about to "win the game" by launching your spaceship is just stupid.

The issue is that ALL players (human and AI) are supposed to represent civilizations

So the game needs to be designed so that ALL players (AI and Human) are encouraged to act like it.

I completely disagree with this. The AI in a game should NOT be designed to play like that. The AI in a game should be designed to play to challenge you. To have an AI the way you describe would make anything past mid-game mind-numbing.

"How so?" I hear you ask.

You know that the Ahriman AI will not declare war on you if friends, so you bury the AI's in money gifts to keep them friends and sail to victory.

I don't want a lie down, stand on AI, which is exactly what you're proposing. Keep them friends, and there's no challenge since you completely 100% control what happens.

That does not apply if
1. the AI can do the same thing to you (I want to win the conquest game but the AI keeps giving me gift techs, and now I can't attack them
or
2. if they stop accepting gifts when they realize they might need to hate you more (redded out.. your cash is no good here)
or
3. if gifts become less and less significant as a win condition approaches (-25: we are envious of your success)


In each case the GAME has been designed to keep the players acting like civs... but the AI manipulates the game rules to attempt to get to a goal

 

[Dale]

In each case the GAME has been designed to keep the players acting like civs... but the AI manipulates the game rules to attempt to get to a goal

This would make the AI play more like a human adapting to the situation in the game, and able to "play and win the game". This is what I'm hoping for in Civ5.

 

[Osama Bin Davis]

While I dont normally agree with Ahriman and find him quite negative on alot of issues. I have to agree with his sentiment.

I find the diplomacy system as it is currently a useless pursuit, it feels too random and makes no sense. I feel like if I have spend time and resources to cultivate a good relationship with a civ it should be rewarded with good relations not a random attack that serves no purpose.

I am all for opputunist attacks, but only if they make sense. If I want to play vs random idiots that play with a win at any cost attitude I can play multiplayer. I want to actually have strong allies with some ai players and have bitter enemies, but it needs to make sense.

I should be heavily penalised for attacking a strong enemy out of the blue. My people should become unhappy, other civ's wont trust me as much etc. Trade should play a more important part, if we have a mutual beneficial partnership the ai shouldnt just turn around for no reason and stab me in the back.

 

[Charles Martel]

While I dont normally agree with Ahriman and find him quite negative on alot of issues. I have to agree with his sentiment.

Same here. I often find myself disagreeing with Ahirman, but not in this instance. Ahriman is absolutely right in saying that there is a fundamental asymmetry between AI and human behaviour. No amount of hiding the diplo info and coding a “play-to-win” AI will make that asymmetry go away.

Some posters downplay the asymmetry, and insist that diplomacy should not constraint AI’s behaviour more than it constraints the human player’s free choice (as Aussi Lurker puts it: “The AI should have the same Free Agency as Humans”). Frankly, I don’t know what to make of these claims. The AI is just a relatively simple algorithm and a bunch of variables, it’s not sufficiently complex to credibly simulate the behaviour of a free agent. (Side note: there is one exception, the AI can credibly simulate human behaviour that is deliberately irrational. If am asked to guess whether my opponent is either Monty’s AI or else a human playing irrationally, then I would struggle to tell the difference). One day (in a distant future) the AI might become so sophisticated to be able to simulate decently a multi-player environment. Until then, it doesn’t make much sense to argue that single-player’s diplomacy should be as close as possible to multi-player’s.

By the way, there are multi-player games where people don’t play to win at all costs, but just have fun role-playing. If the AI is really to enjoy free agency, then it should have the freedom to choose between “play-to-win”, “role-play” and even “play-badly-just-to-frustrate-the-humans”

 

[Aussie_Lurker]

Yes Charles, they're called Diplomacy Games, & I'm a regular player of them. That said, there are winners of these games-its just that victory is based on more than the victory conditions of Civ4. You are awarded points on the basis of role-playing & storytelling as well. Don't think it prevents back-stabbing & other shenanigans, though-you just need to have a very good in-game reason or else you'll get penalized. Also, though, as you're playing other humans, you never know if treachery will come your way from some unexpected quarter.
So, whilst diplo games don't have a "win at all costs" mentality, they are still incredibly cut-throat in many ways. All I'm saying is that we shouldn't have the AI strait-jacketed by its diplo ratings towards you if you're not similarly constrained. Any constraints should come via your people & how highly you rely on their good opinion. So though I know I probably should be able to trust that Police State that I've been trading coal to for the last 30 turns, the fact that he's running a police state should make him much less trust-worthy in my eyes-especially if I'm running a democracy. Does that make sense?

 

[McLMan]

It would be nice if the AI was smart enough to realize it was in a winnable situation, and then do what it takes (attack allies) to win at that point. Otherwise good relations should mean something.

 

[Akka]

The issue is that ALL players (human and AI) are supposed to represent civilizations

So the game needs to be designed so that ALL players (AI and Human) are encouraged to act like it.

Yes.
The aforementionned "population opinion", making waging wars on friend induce severe penalties from your civil unrest, would be a good way to achieve that.

 

[Ahriman]

I completely disagree with this. The AI in a game should NOT be designed to play like that. The AI in a game should be designed to play to challenge you. To have an AI the way you describe would make anything past mid-game mind-numbing.

Do you find this to be the case in Civ4?

Civ4 has an AI that plays like a civilization, based on its diplomacy modifiers, not plays to win.

Its not a great AI for a number of reasons, but its one that is fun to play against.

Each leader has a AI parameters that determine the best relations they can have with another faction where they will still attack them, or be brought in a war against them.
Most AIs will *never* attack a player with whom it is friendly, some will never attack you if they are pleased.

Now, there is all kinds of scope for things like "you have a vast empire, -2 diplomacy modifier" or "you're building a spaceship, -2 diplomacy modifier". You can incorporate all those sorts of things into the existing system, that make it harder to maintain good relations.

[DuneWars mod has a terraforming victory that requires you to control and turn into grassland a % of the planet, as you get closer to this goal you get increasing diplomacy penalties with non-terraformers.]
I also don't mind a Master of Magic/FFH2 style "you are building the final wonder that wins you the game, I will attack you now!" mechanic for a final endgame war.

But for most of the game, diplomacy modifiers need to actually mean something. I need to know that I don't have to guard my border with my good friend in case he backstabs me and takes the cities in the heart of my empire. That's the whole payoff from diplomacy that keeps him a good friend; I can move those military units elsewhere.

Oh, and Aussie, please use paragraphs more. Its very hard to read those textblobs.

 

[Loppan Torkel]

The issue is that ALL players (human and AI) are supposed to represent civilizations

So the game needs to be designed so that ALL players (AI and Human) are encouraged to act like it.



That does not apply if
1. the AI can do the same thing to you (I want to win the conquest game but the AI keeps giving me gift techs, and now I can't attack them
or
2. if they stop accepting gifts when they realize they might need to hate you more (redded out.. your cash is no good here)
or
3. if gifts become less and less significant as a win condition approaches (-25: we are envious of your success)


In each case the GAME has been designed to keep the players acting like civs... but the AI manipulates the game rules to attempt to get to a goal

Won't this just become another meta-game and something that will reduce the players' freedom?

I don't think this looks like a good idea. Again, something based on what Meier himself brought up: The AI-players shouldn't be too sympathetic, they also shouldn't come begging for mercy or, in the end, act like real-life politicians, because the players shouldn't feel bad when they try to win the game.

The AI-players should act like ruthless players in a game for world domination. They should role-play a caricatured version of their real life counterparts and play to win within limits. Alliances should be broken when needed.

I voted the poll-options weren't good enough.

 

[KrikkitTwo]

Won't this just become another meta-game and something that will reduce the players' freedom?

I don't think this looks like a good idea. Again, something based on what Meier himself brought up: The AI-players shouldn't be too sympathetic, they also shouldn't come begging for mercy or, in the end, act like real-life politicians, because the players shouldn't feel bad when they try to win the game.

The AI-players should act like ruthless players in a game for world domination. They should role-play a caricatured version of their real life counterparts and play to win within limits. Alliances should be broken when needed.

I voted the poll-options weren't good enough.

That's what the players should Act like. But the game should mean that winning involves acting like a civilization.

So not a 'senate stops you from declaring war' but...
The AI has offered me a gift, if I take it, it will cause more unhappiness for me to declare war on them later... I'll decline it.

Then the AI can use the same logic... (the human player is offereing me a gift.... etc.)

If attacking my 3000 year ally has Massive happiness penalties then there is a Benefit to having a 3000 year ally (because they get those happiness penalties if they attack me).

If making a trade treaty with someone I was just in a massive war with causes happiness penalties, then I will wait until we have been at peace a little bit longer to propose those treaties (and so will my AI previous enemy)