Poll: what makes a good diplomacy system?

[Naokaukodem]

Thats a very strange argument. This is a game, so AIs should play in a way that is most fun for the human, who is actually playing the game. The AIs aren't people playing the game, they're there to enhance *my* enjoyment, not there own.
Many people (myself included) feel a total break in history immersion if AIs do things that make them feel like gameplayers, rather than real countries.
For example, if the US and UK had annexed France rather than liberating it after WW2 (play to win!), or if the US troops in Britain before D-day had suddenly marched on London and conquered it (play to win!) or if the US suddenly invaded Canada during the 20th century (play to win!) then that would feel totally wrong.

Your examples fit perfectly the realism bit of my sentence. "if the game mechanics are not aimed towards realism, then nevermind the realism". It means that if the game does not "force" you to give back cities, or does not force you not to invade a weak neighbour, then you shouldn't go for them, no more than the AI.

Although, in Civ4 and other Civs anyway, we are far to see this kind of things you describe. The fact is, that in previous Civs, Ais plays all against the player. This is the result of a "friendly" AI. Contradictory? No. If the AI is there only for flavor, and not winning, then the programming of it is not rational, hence those gangs up against the human player.

Ultimately, AI playing to win would be more fun to me. AIs would be rooted to the mechanics of the game and the geopolitics of the map. Would trigger much more enjoyable games, like those games we can see in multiplayer, that are far more interesting than actual Civs single player, for the diplomacy part at least. As to multiplayer, that isn't that much fun, not because of diplomacy, but because of players quitting. Last game, there was a big war dude with many horse archers that began to invade other players. Instead of resisting, those players quited, and what? The big dude made peace with AIs! And what? He started a war with another player! And what? He quited at his turn! At the end i was the only human player left with the big dude, he had the same army as mine, but he had a score much more high, because he had some cities from various civs. That (and many other things, among them the time lenght of the games, but it's not the topic here), is making the multiplayer broken, not the fact to play to win.

 

[plasmacannon]

This is a good summary of how I feel. I am interested but surprised that many people would not be bothered if someone who they spend the whole game focusing positive diplomacy on (in Civ4 terms, adopting their civics and religion when asked, giving them gifts when asked, not attacking their friends, joining their wars when asked etc.) suddenly turned around and attacked them.

I agree with you.

This has happened in a Civ4 game of mine too. Same religion, civics, joined a battle, etc. +13 relations with genghis Khan, and he decides to attack me? While, there were 3 other empires still left in the game.

Geesh, sometimes you just can't trust these barbarians.

I suppose what I am getting at, is that there might be a few leaders that one could make a case for the opposite, but, for the most part, you are correct.

 

[KMRblue1027]

I like a more hectic and unpredictable game which means the AI should be able to have complete control of all their actions, but they should make all actions lead to a goal. If a close friend has 3 oil and won't share I would attack and so should the AI. Every action performed by ideal AI should contribute to a grand strategy, eventually leading to winning the game.

 

[chongli]

This is a good summary of how I feel. I am interested but surprised that many people would not be bothered if someone who they spend the whole game focusing positive diplomacy on (in Civ4 terms, adopting their civics and religion when asked, giving them gifts when asked, not attacking their friends, joining their wars when asked etc.) suddenly turned around and attacked them.

And what if the civ who happens to be the focus of all this attention simply does not want it? Should they be forced to like you just because you shower them with gifts? Perhaps they don't care about anything you give them because all they want is your land?

You should not be able to distract the AI from what it really wants by heaping meaningless token gifts upon them. If the AI's goal is to achieve the new "conquer all capitals" victory condition, the only way to appease them permanently should be to cede your capital to them.

 

[Zeiter]

I won't argue for the diplomatic system in civ4 since I've played it too little. I wonder, however, how there could be a way to:
1. encourage the player to try and take over the world (like some crossbreed between a mad Bond villain and Hitler),
2. make the player feel good about it and
3. make it happen in a realistic manner.
Civ has always prioritized gameplay before realism. If the two are combinable - great, but here I don't see it happening and I'd gladly settle for 1 and 2.

What you're basically asking is for a game that makes domination and conquest victory, as we've known them, impossible, which they historically and realistically are and should be.

The fact that there is a domination victory condition is encouragement enough from a gameplay perspective to take over the world. There's no problem there.

It's not that I want to feel good about it. It's that, IF I choose to try to take over the world, I WANT to feel like I'm playing the role of a Hitler or Ghenghis Khan type of character. Sometimes human players WANT to roleplay as the evil bad guy and be told that they are being the evil bad guy.

Which means, I want the other AI to hate me and fear me and band together to stop my reign of terror. I think that that's both fun from a realism standpoint, and makes good gameplay sense.

Alpha Centauri simulated this nicely. If you launched an unprovoked planet buster attack in SMAC, all of the world leaders immediately declared war on you and expressed their horror at your actions and their intentions to put an end to your barbarism at all costs. This applied to even one's closest friends. This seems both much more sensible in terms of creating a gameplay challenge, and in terms of realism and flavor, than a simple -1 "You nuked our friend!" from a civ that is for some ungodly reason still friendly with you like in Civ4.

Basically, if human civ players are going to roleplay, then they shouldn't have to do all of the heavy lifting on their own. There should be gameplay mechanisms that help the human roleplay.

So, to take a counter example, if you were buddy-buddy with all of the civs the whole game, you should be appreciated by the other AI like a Ghandi so that:
1. You are rewarded for your sacrifices from a gameplay perspective.
2. You can actually feel like you are Ghandi.
The AI should not only have the programming to do this, but should also have a gameplay incentive to treat you as a Ghandi when you deserve to be treated like a Ghandi (to give them a gameplay incentive not to backstab you), so that the realistic behavior is also the behavior that gets the highest score for the AI. In SMAC, they did this by, if a pactmate won the game, adding 50% of that pactmate's victory bonus to your own score. So even if you didn't win the game, you could rationalize not backstabbing your close friend not only from a realism standpoint, but also from a standpoint of getting the highest score possible.

And likewise, if you go from being a Ghandi to nuking the whole world, the world should be, if anything, even more horrified. Your good actions throughout the game should not compensate for such things—if anything, they should make the AI, who are now your friends, hold you to an even higher standard.

That is to say, AI's that are friends with you should praise you more highly for the good things that you do, and criticize you more strongly for the bad things you do, since they have invested their friendship in you and have hitched their wagon to your horse in that game, so to speak. Whereas the AI should treat an untrustworthy, evil human player more skeptically. So if you do something good, you only get a +1 because the AI assume that it's only a cynical ploy of yours (so it takes a lot more to get back into their good graces where they trust you). And if you do something bad, you only get a -1 because it is hardly surprising to any of the AI in the game that you, the evil Hitler civ, would be doing these bad things.

 

[Sydney Posada]

There seems to be a lot of debating to cross-purposes here. There seem to be two main philosophies on how to enjoy Civilization:

1) It's an immersive, alternate-historical, role-playing experience.
2) It's a strategic board game, like chess.

Obviously most people are a mix of both, but it should be clear that people who follow #2 exclusively would probably like having an AI that knew it was playing a game and tried to win. People who follow #1 would prefer they act more as role-players, acting out their parts as perfectly as possible, even if it means not turning on their friend who's about to win the game.

Why can't they satisfy both parties? Include an option similar to "Aggressive AI"; instead, have it called "Cutthroat AI" or "Game-Player AI". Program this AI to act like a player trying to win a game.

Thus, players can choose for any given game whether or not they want a typical role-playing AI Civ experience, or a pseudo-multi-player experience in which every AI will do anything to win the game. Both aspects of Civ enjoyment can be satisfied. It will probably require a lot of time to actually get the programming right, but it would be well worth it.

 

[KrikkitTwo]

There seems to be a lot of debating to cross-purposes here. There seem to be two main philosophies on how to enjoy Civilization:

1) It's an immersive, alternate-historical, role-playing experience.
2) It's a strategic board game, like chess.

Obviously most people are a mix of both, but it should be clear that people who follow #2 exclusively would probably like having an AI that knew it was playing a game and tried to win. People who follow #1 would prefer they act more as role-players, acting out their parts as perfectly as possible, even if it means not turning on their friend who's about to win the game.

Why can't they satisfy both parties? Include an option similar to "Aggressive AI"; instead, have it called "Cutthroat AI" or "Game-Player AI". Program this AI to act like a player trying to win a game.

The simpler way to satisfy both parties is to set up the rules of the game so that
Achieving the strategic goal in the game (part 2) requires players to play 'alternate historical' roles (part 1)

 

[Loppan Torkel]

The fact that there is a domination victory condition is encouragement enough from a gameplay perspective to take over the world. There's no problem there.

It's not that I want to feel good about it. It's that, IF I choose to try to take over the world, I WANT to feel like I'm playing the role of a Hitler or Ghenghis Khan type of character. Sometimes human players WANT to roleplay as the evil bad guy and be told that they are being the evil bad guy.

I'd argue that most don't, that's why they make the leaders somewhat obnoxious, the average player needs to feel he can win the game by conquer the world without being an ass.

Which means, I want the other AI to hate me and fear me and band together to stop my reign of terror. I think that that's both fun from a realism standpoint, and makes good gameplay sense.

To some degree, yes, but I know there were a lot of complaints to the Total War series, because the AI had a habit of ganging up every time the player got a bit bigger. Realistically there should be no way to conquer the world, here it should be as viable as any other victory condition. Though they have made some adjustment as it now is enough to conquer the capitals of each civ. I like the new approach.

Alpha Centauri simulated this nicely. If you launched an unprovoked planet buster attack in SMAC, all of the world leaders immediately declared war on you and expressed their horror at your actions and their intentions to put an end to your barbarism at all costs. This applied to even one's closest friends. This seems both much more sensible in terms of creating a gameplay challenge, and in terms of realism and flavor, than a simple -1 "You nuked our friend!" from a civ that is for some ungodly reason still friendly with you like in Civ4.

I think they got pretty angry at you in civ3 if you did this, maybe they just got the balance wrong in civ4. It seems reasonable to have more serious reactions.

Basically, if human civ players are going to roleplay, then they shouldn't have to do all of the heavy lifting on their own. There should be gameplay mechanisms that help the human roleplay.

So, to take a counter example, if you were buddy-buddy with all of the civs the whole game, you should be appreciated by the other AI like a Ghandi so that:
1. You are rewarded for your sacrifices from a gameplay perspective.
2. You can actually feel like you are Ghandi.
The AI should not only have the programming to do this, but should also have a gameplay incentive to treat you as a Ghandi when you deserve to be treated like a Ghandi (to give them a gameplay incentive not to backstab you), so that the realistic behavior is also the behavior that gets the highest score for the AI. In SMAC, they did this by, if a pactmate won the game, adding 50% of that pactmate's victory bonus to your own score. So even if you didn't win the game, you could rationalize not backstabbing your close friend not only from a realism standpoint, but also from a standpoint of getting the highest score possible.

Maybe they could add a pactmate score bonus, but I think the rest sounds better than it would be ingame.
You want to be rewarded for gameplay sacrifices and want the AI to reward you for it?! Either you make the system obvious which makes this another gameplay mechanic to take advantage of, or you make it non-obvious, leaving you to play the game like a prima donna with the AI-players sitting around and only reacting to your gamestyle.

I just don't see this turning out good if taken as far as you'd like. Say you're playing against Gandhi and he has played peacefully throughout the game and is closing victory, by culture or space-race. Your only chance is to attack him, the only problem is that if you do, you'll be dog-piled by the other AI-civs because they like Gandhi.

I think there should be balance in this. If you play nice you should receive some benefits from it, but you should always have to be suspicious toward the other civs, not knowing they won't attack you because you're such good friends with them. This should only make them think twice before attacking you due to a greater diplomacy-hit rather than some AI-morality that needs to be adjusted to the player each game.

And likewise, if you go from being a Ghandi to nuking the whole world, the world should be, if anything, even more horrified. Your good actions throughout the game should not compensate for such things—if anything, they should make the AI, who are now your friends, hold you to an even higher standard.

That is to say, AI's that are friends with you should praise you more highly for the good things that you do, and criticize you more strongly for the bad things you do, since they have invested their friendship in you and have hitched their wagon to your horse in that game, so to speak. Whereas the AI should treat an untrustworthy, evil human player more skeptically. So if you do something good, you only get a +1 because the AI assume that it's only a cynical ploy of yours (so it takes a lot more to get back into their good graces where they trust you). And if you do something bad, you only get a -1 because it is hardly surprising to any of the AI in the game that you, the evil Hitler civ, would be doing these bad things.

Couldn't the AI-civs encourage Hitler by giving him +2 when he does good instead, or see between the fingers when Gandhi nukes someone because he's been so good this far and perhaps does it for the greater good?

The simpler way to satisfy both parties is to set up the rules of the game so that
Achieving the strategic goal in the game (part 2) requires players to play 'alternate historical' roles (part 1)

What if those rules make the game more boring?

 

[Sydney Posada]

The simpler way to satisfy both parties is to set up the rules of the game so that
Achieving the strategic goal in the game (part 2) requires players to play 'alternate historical' roles (part 1)

That's a good game design idea to keep in mind, but they're never going to get the goals of having pseudo-historical role-playing AIs and playing-to-win AIs coincide perfectly. In fact, it's doubtful they can even come close.

One basic stumbling block is that the rules of Civ just don't tend to make both true friendships and playing to be the sole victor to make sense in the same game. If they can somehow pull it off, more power to them.

However, realistically, if they can't do that, I would love to have the option (in Custom Game) for Cutthroat AIs who actually play the game to win, in addition to the normal role-play-y AI. I would definitely play both AI options a lot, and I suspect that a ton of other people would, too.

 

[AfterShafter]

Personally, I'd like to see more types of binding agreements, varying degrees of peace, free transit, trade routes, etc etc. Would be tricky to include without making it either too convoluted or too easy to exploit, but it would go a long way to making diplomacy a tool you can't just ignore with a large enough military.

 

[Naokaukodem]

It is simple; if AIs are programmed in order to play a role, they will not tend towards the player in a satifying maneer, because they will act irrationally. I mean, they WILL surely have unfair advantages, as well as unfair disadvantages.

It's inevitable. So we would end up again, despite all the efforts of the programmers (no more tech trading...) in a situation where the upper difficulty would be doable only with exploits. Using exploits is not fun to me, particularly when those exploits are demanding (example: tech trading; you have to know what techs the AIs will not go for and prompt them every turn. Come on it's not a grand strategy game! It's mean peny pitching).

We would play a game of riddles rather than a game of war and management: a totally different game from what i expect!

 

[chongli]

Personally, I'd like to see more types of binding agreements, varying degrees of peace, free transit, trade routes, etc etc. Would be tricky to include without making it either too convoluted or too easy to exploit, but it would go a long way to making diplomacy a tool you can't just ignore with a large enough military.

How exactly does a binding agreement work on an international level? I thought binding agreements required an overarching authority to enforce. No such authority exists in the case of nations as each nation holds sovereign power in its territory.

 

[RD-BH]

How exactly does a binding agreement work on an international level? I thought binding agreements required an overarching authority to enforce. No such authority exists in the case of nations as each nation holds sovereign power in its territory.

The Developers and the game engine are "overarching authority",
--- binding agreements can easily be enforced.

 

[chongli]

The Developers and the game engine are "overarching authority",
--- binding agreements can easily be enforced.

Putting binding agreements into the game mechanics is a fatal mistake. Think back to the absolute garbage of "You are overruled by the senate!" that plagued past Civs.

Heck, even Civ 4 had the incredibly annoying Nuclear Non-Proliferation Treaty which was impossible to defy no matter how powerful your state was.

 

[RD-BH]

Putting binding agreements into the game mechanics is a fatal mistake. Think back to the absolute garbage of "You are overruled by the senate!" that plagued past Civs.

Heck, even Civ 4 had the incredibly annoying Nuclear Non-Proliferation Treaty which was impossible to defy no matter how powerful your state was.

Civ IV currently has binding agreements.
The 10 turn peace treaty comes to mind.

Regardless:
... I still hope for a system
... ... with risk
... ... followed by a chance
... ... for short and long term gain.

 

[chongli]

Civ IV currently has binding agreements.
The 10 turn peace treaty comes to mind.

Yeah and I hope all of them are removed from Civ V.

Regardless:
... I still hope for a system
... ... with risk
... ... followed by a chance
... ... for short and long term gain.

What?

 

[Jopo]

I think LOTS of OPTIONS make a good diplomatic system, lots of different actions that you can do.
I think the AI should be EXTREMELY AGRESSIVE, using dirty tricks and espionage to WIN, I want a REALLY tough challenge.

Cheers from Finland!

 

[RD-BH]

Yeah and I hope all of them are removed from Civ V.

Good point.
As a sovereign nation, you should be able to break treaties at will.
The penalty would be in diplomatic relations (ie "-2 You stopped trading with us!").

RE: risk vs. short and long term gain

Example:
As you build relations, each increase opens options to negotiate for ever greater rewards.
... (ie trading additional resources, research pacts, embargos against common enemies, defensive pacts, etc)
... building relations could initially be as simple as tribute.
There could be a Diplomatic version of a tech tree.
You are risking an upfront cost to "research" (build relations) the Diplomatic "tech".
... "research" could be your investment of tribute or even turns of unbroken treaties.
The Diplomatic "tech" opens new options for trade, borders, research, etc. (the short term gains).
The long term gains come from moving deeper into the Diplomatic "tech" tree.
... (ie increasing interdependency with that AI resulting in ever larger rewards)
The risk is always the breaking of relations, pacts, trades, etc for the AIs overall game goals.
... (this should knock off Diplomatic "tech" tree levels along the appropriate branch)
... Just because you open up a new level, doesn't mean the AI will sign a new treaty.
... You can break treaties for personal gains just like the AI can.
AI personalities could include % risks for breaking agreements along certain Diplomatic "tech" lines.

The idea is to have a more robust/lively diplomacy system that involves risk and gain.

 

[Jamuka]

If there is a huge discrepancy between the two options that represents a flaw in the system. When there is such a discrepancy, the AI should play to win.

 

[Aro]

That's basically Krikkitone's suggestion, only he went all specific and in-depth on it. I like this idea that diplomatic actions incur penalties/bonuses to your civilization in some way. If this is boosted based on inter-civilization relations, I believe we could have an interesting diplomacy mechanic that even works for both the AI and for MP.

..........................

Agreed... But I would like to may turn this 'on' or 'off', like many features in Civ IV when you play a custom game. In fact, I would like and approve any improvement made on the game to make it harder and more realistic and more complex... But only if you can turn the whole thing off, one by one, and play a game as simple as you want, like our fathers used to do...

You know, some days you just want to slaughter a little, without domestic problems in your way… I’m married, I know it.