Sid Meier's Civilization Series (Main Titles)

What exactly would the point of telling the AI you're angry be? I mean, you, as the player, are presumably angry and your playstyle will reflect this: ie you declare war on them or whatever. The AI doesn't listen to you unless they're scared of your army, and if they're too scared of your army they'll attack you in self-defense anyway. They most like other players who are too far away to interfere with their own interests.

In other words, they act exactly like you. I very much doubt that you "respect" any of the other civs while playing, you just treat it as a game that you're trying to win. Other civs are evaluated based on how useful or threatening they are to you, full stop. It's fair. Any kind of more complicated diplomacy system would actually be "cheating" in a way, since you would be forcing the AI to act against their own interests. It would skew the power way too far towards the player.

edited 24th Apr '14 10:24:53 AM by Clarste

Yeah, gonna throw my hat on the "AI should be playing the game too" pile. Civ is far too much of a game (as opposed to a historical sim/strategy thing like Europa Universalis or Crusder Kings) to let the AI act based on realistic diplomacy. The player would just be able to bully them into submission.

It would probably have the same effect as it has when the AI does against the player - with the World Congress, unless the player is overwhelmingly powerful at that point, the player should be paying close attention to who might be angered by their decisions and their proposals, and react accordingly: sabotage an enemy, placate a potential ally, try not to antagonize a giant that's already pissed at you, etc. The only situation where you wouldn't and would just do whatever you want is if you had absolutely nothing to fear from any foreign aggression, which isn't always the case.

As far as I know, the AI gets diplomatic penalties from each other in regards to World Congress proposals but not from the player, and thus don't consider the player at all - not making that distinction gives the player a penalty that the AI is not subject to, and thus the AI does not - at least in that situation - act like the player at all, because they do not act as though their actions have consequences.

Not necessarily. It's worth noting that Civ IV gives the player much more diplomatic options with the AI, and it's not particularly easy to bully your opponents into submission there. If anything, it's arguably easier to get a diplomatic advantage here - not that that's necessarily a bad thing.

edited 25th Apr '14 9:01:36 AM by KnownUnknown

Frankly, I've never gotten the AI in Civ V to submit to a demand, ever, even if my military is 10 times stronger than theirs.

But this is the point I'm talking about. The AI gets mad at you for proselytizing to their citizens; why can't you dissuade them from proselytizing to your citizens under the same logic? Why do they get mad at you for voting against a World Congress proposal that harms you, yet receive no diplomatic penalty for proposing one?

Why can they beg for gifts from you, but you can't beg for gifts from them?

The AI and player operate by different principles.

edited 25th Apr '14 9:22:14 AM by Fighteer

They do receive a diplomatic penalty. You, the player, reevaluate their threat/usefulness ratio and act accordingly. That's all a diplomatic penalty is or ever could be. The computer's attitudes can be reduced to a number because of the way computer games work, but your attitude can't. That's the only difference.

You can also demand things from them: click on the pop-up message saying that they spied on you or converted your city and it'll open up a diplomatic window. Offer an unfair trade and they'll probably reject it, just like you would.

That's not a diplomatic penalty, that's a reaction. Both the AI and player (should) take into account possible diplomatic penalties before taking actions, instead of (in this case) never.

Basically, that's after the fact and not involved in the decision making itself. That's a key difference - the AI has no concept of judging the consequences of their congress decisions in regards to the player, because they don't take that into account, whereas the player does and plans accordingly. It's particularly noticeable, because the AI does take into account factors about the player's empire and tendencies when deciding whether to take other actions.

edited 25th Apr '14 9:43:19 AM by KnownUnknown

Not at all players do. I for one never care about what the AI will think: I just do whatever I want and apologize later. The AI being stupid doesn't mean they're not playing by the same rules.

edited 25th Apr '14 9:42:32 AM by Clarste

So? It's still a mechanic that penalizes the player and causes the AI to act in an illogical way. If you don't mind it, that's fine, but the fact that some players play in such a way that they wouldn't have to care doesn't make it any less of what it is.

edited 25th Apr '14 9:47:19 AM by KnownUnknown

I don't see how it penalizes the player. The fact that you can plan ahead is an advantage. If the AI doesn't consider who it's going to piss off before it does something, it just means it pisses someone off and gets into a war it could have avoided. Maybe with the player. If you get really sick of the AI sending Prophets into your territory all the time and decide to stop it through military force, then the AI who was too stupid to stop sending prophets only has itself to blame. And if you don't have enough military force to stop it, then hey, turns out they were right all along.

I seriously do not understand what your complaint is supposed to be. It really sounds to me like you want to be able to bully the AI even more.

edited 25th Apr '14 9:52:33 AM by Clarste

I'd like the AI to anticipate the likelihood of me disliking what they do, on the same basis as they anticipate the effects of their actions on each other. That is, they maintain a value for how I ought to feel about them, given the same priorities as they themselves use.

To address your concern , I'd like the AI to try to make friends with me, instead of passively deciding whether to be friendly on the basis of my actions.

edited 25th Apr '14 9:53:54 AM by Fighteer

For one, it means that the player incurs penalties for the things the computer does not - by proposing congress bills, you incur diplomatic penalties and aggression from other races, which the AI cannot perceive in regards to you. As a result, the AI can and will propose and vote on things that penalize you without taking into account how it effects you, thus essentially removing the player from having an impact on other races diplomatically except from things that they push forward themselves.

And if you can't take action against it, you pretty much have to let it happen. For a game that boasts a wide range of options besides military, including diplomacy, cultural influence, etc, the idea that the fix for a situation is just to war them into submission (which is a workaround, not a reason it's not a problem) doesn't reflect well. If I wanted that, I would've stuck with Total War or Endless Space.

I feel like a general fix would be to have there be a little more interaction between races - maybe by giving diplomats some more options.

On the flipside, it also means that the AI cannot be discouraged from taking certain actions on pain of foreign aggression can, and neither will they prioritize actions that benefit both themselves and their allies if that ally happens to be human. It negatively impacts the

edited 25th Apr '14 10:01:04 AM by KnownUnknown

I feel like diplomatic penalties are their attempt to simulate how a human player reacts to an opponent that's pissing them off, so having an AI get a diplomatic penalty as well as pissing off the human player seems redundant, and would also further hobble an AI that struggles to play the game competently. I guess they could make the AI play by the same "rules" as the player, but do you really want the AI to need even more cheats just to provide a challenge to players? I have no problem that many mechanics are not symmetric.

It's perfectly possible to win the game peacefully: just don't piss off your neighbors. Don't expand aggressively, accept unfavorable deals, give in to their demands. Make yourself more valuable to them as an ally than the risk of conquering you. I win like this all the time. If you feel like you need to have everything go your way every single time, then you need to be a military superpower. That's all there is to it.

True. Not entirely bearing on what we were talking about, though.

edited 25th Apr '14 10:25:14 AM by KnownUnknown

It is exactly what we're talking about. Diplomacy in this game boils down entirely to risk-reward structures. As mentioned, it is a game before a simulation. Every player is trying to win, even the AI. Diplomacy means taking into account the partner's material interests. There are no friends in Civilization, only two players who are of value to each other. Giving people what they want makes them like you. You seem to want them to like you without giving them anything back, which is frankly just cheating.

You can also trade for votes in proposals, so I really don't understand your complaint.

edited 25th Apr '14 10:28:18 AM by Clarste

No, it isn't. You're taking a conversation about a very specific flaw in the diplomacy system, and how it effects the players using that system, and trying to turn it into a conversation about the universal mechanics of the game as a whole. I'm assuming you're trying to say that since the game isn't unplayable as a whole, the individual flaws don't matter - that sort of reasoning has less bearing than you might think.

Yes, you can play the game in a myriad of different ways, but again, the issue of the player having to deal with diplomatic penalties while the AI doesn't causes problems with the Congress - as well as a few mid-late game issues (for example, the AI won't think to not do certain things until asked to, which you can't do in some cases until they do it, whereas the player can be penalized regardless).

edited 25th Apr '14 10:32:26 AM by KnownUnknown

You consider it a flaw that the AI always places its own interests ahead of the player's. I consider it a feature. I think you misunderstand the role of diplomacy in this game on a very basic level.

The main issue with that line of reasoning is that it totally ignores the key issue - even outside the concept of diplomacy, this creates an unequal playing field. One which not only penalizes the human, but causes issues with the AI.

This isn't a feature for AI behavior, but rather it makes the AI blind to the player's actions and thus causes the AI to ignore the player. As I've noted several times, it causes the AI to act illogically, not progressively, because and I've given at least one example where it can do things that shoot itself in the foot (it can't always take into account what might hurt its allies) if done. That is hurts that player moreso than the AI is obvious, but it's still something that causes strange behavior.

You're already aware of this, because you tried to claim earlier that the issue doesn't matter because it gives the player an advantage in some ways:

I'm interested that now you're claiming that the player advantage to a system that you claim is there to give the computer a human-like advantage is correct - especially if that system causes them to act, in your own words - stupidly. Or, as you claim, causes them to act more like a player, who also by your own words plans ahead for their actions, and thus more adequately utilize diplomacy, especially if (as you say) it means they're unable to adequately discern what might incite a war or not.

Lastly, aren't you the one who's been saying that the way you play is different, and that the way you play makes these things less relevant - and thus that's why this doesn't matter to you? Now you're saying that the way you play is how it's supposed to be played, and everyone else just doesn't understand the game like you do?

edited 25th Apr '14 10:54:02 AM by KnownUnknown

I'm saying you can play in a way that makes the AI happy, or you can use military force to make them do what you want. Presumably there are middle paths where you take elements of both approaches. That's what diplomacy is in this game: a balance of power and self-interest. That's all it is. The AI is not obligated to "respect you" in any way. You don't respect it either. You have not once respected the AI. You are disrespecting the AI as we speak. You are asking for an asymmetrical situation where the AI respects you, while you do everything in your power to exploit its risk-evaluation algorithms. Diplomacy is not and has never been intended as a way to exploit the AI's flaws.

You think it's a "penalty" that the player is smart enough to try to appease its neighbors while the AI is not. That's completely ridiculous. It's only a penalty for the AI who couldn't maintain your friendship. If you don't want to go to war over them banning citrus, then don't freaking go to war. You have the choice to acquiesce to an unfavorable deal. You always have that choice. You also had the choice of giving them so much citrus that they would never even consider banning it. You seem to think you don't have that choice: that every diplomatic action must benefit you perfectly. That is nothing but self-serving, and more disrespectful of what diplomacy is in this game than anything the AI could ever do.

edited 25th Apr '14 11:03:40 AM by Clarste

Your arguments still aren't lining up well. Again, if it's a system that causes the AI to be unable to make smart decisions and sabotage their own friendships, why is it fair for the AI? And why is it unfair to the AI to allow the AI to be aware of the same things the players are?

Hell, even if the post I just quotes, your basic tone is that it doesn't matter because the AI being unable to gauge player reaction makes it easier to declare war on and thus defeat them, which, in a sense, would be an exploitation of the system. Especially since your reasoning there -

- implies that if it allows the player to win said conflict, it doesn't matter.

edited 25th Apr '14 11:10:56 AM by KnownUnknown

I'm saying you, the player, can allow the AI to win in an exchange. You can allow them to do something that you don't like. You are completely denying that possibility. Actually, more than denying you seem to not understand it. The AI is allowed to do things diplomatically that you would prefer them not to do. You frame this as a "penalty" to the player, when in fact it is the most advantageous thing the AI could do. As the player though, you are completely free to choose how to respond to this. That's what diplomacy means.

Is the AI sometimes stupid? Sure. But you are asking them to always be stupid, to never take risks, to never ask you for more than you're willing to give. To not seek their own benefit at the expense of yours, to never call your bluff. You want an ally to be a slave, who bows down to your desires at the expense of their own. You want to deny give and take as a part of diplomacy. Because that's what it means.

The player is free to declare war on whoever they want. They can have the best ally who always gives them everything and still declare war on them at a whim. The AI cannot obtain a guarantee of fair dealing from you. Unlike the player, the AI is constrained by its decision-making algorithms. The AI doesn't have "whims" and never wishes to play a game out in a certain way as a change of pace. Maybe they have random elements, I don't know, but those are not accurate representations of human decision-making. The AI can never be human, and that's exactly why in order to make them better players we have to sometimes make them stupider. It's the lesser of two evils.

edited 25th Apr '14 11:22:48 AM by Clarste

You're mixing the freedom on the reaction with the freedom of the action. In that situation, the AI is free to act against (or for) the player however it sees fit for itself (that is, it doesn't itself consider the player in its actions, thus causing it to react less intelligently in regards to the player), and the player is free to react.

However, the flipside of this is that the player is less free to act whereas the AI is as free to react as the plaer - or more accurately, the player is more aware of how restricted their actions really are. They cannot act without incurring precise diplomatic responses that the AI is not privy too, and while they can plan around it. This tends to lead to the AI forcing the player's hand more often than it should, because it literally cannot conceive of the player having a problem with its actions.

I kind of get what you're saying, but saying it lines up on a conceptual level ignores the fact that it doesn't line up on a mechanics level, and - as noted - causes problems it shouldn't. It's basically all about game balance - and as I elaborate on further down, this game generally tries to program the AI to act as a player would, with information balanced with what the player has. Without that information, it causes issues.

That's a very extreme assumption, and one that doesn't fall in line with what we know about how the computer works.

Keep in mind that the computer should have the diplomacy values for other AI, and thus still acts in interest for or against them despite, and even because of, it. That's because computers are programmed not only to - that's how they function for most aspects of the game, and it's the basis for computer intelligence in 4X games in particular. Civ V is particularly notable in this regard, because the computer is otherwise very quick to respond to new data in general.

Note that for most other aspects of the game, though higher difficulty tips the odds in favor of the AI in terms of default resources and techs, the basic mechanics of the game generally subjects all players to the same rules - they use the same tools to do the same things.

This issue, however, causes the AI to act without knowledge of certain variables and thus not act as intelligently as they should to compensate for them - I've noticed this in regards to a few other aspects of the game as well: whereas they are aware of various elements of the other AI's empires and react accordingly. This is very important, because it means that if given some kind of estimate system for player tendencies, they would continue to act in a similar manner - but would now be informed about it.

Note that the AI actually has a stat for how bold they are, and in such situations they are judged by that. How much information they have probably wouldn't have much effect on their boldness, just in how they choose to go about doing it. For example, it would allow the AI to use the Congress system more offensively than they already do (since the extent of that now is merely embargoing people they don't like and abolishing resources they don't have).

But you're claiming that this instead causes the AI to inherently not take risks because they have data, even though that's exactly what they're programmed to do? Exactly what they've been doing for several incarnations of this series - as I pointed out before, in IV the AI deals with more diplomatic options and as far as I know doesn't have any deals it can take that aren't possible with the player, and is able to react accordingly just fine.

The AI, in those situations, is perfectly capable of gauging risks and making reasonable actions based on those risks, and comes out the smarter for it because they know all the variables - this is how a computer pushes its advantage in a 4X, and is one of the basic tenets for how game design works.

In short: a system that gives the AI similar information that the player does and allows them to strategize around it in no way unfair to the AI - and in fact works out better for both parties. Unless there was a serious programming error it would certainly not make the AI any less intelligent.

edited 25th Apr '14 11:53:52 AM by KnownUnknown

The player's actions are never constrained. You can choose to lose the game in a "fun" way. The AI cannot make this choice, it will always be constrained by what the programmers thought was the intelligent thing to do.

What we're asking for, in this particular instance, is for the AI to predict the reactions of the player. This is something they literally cannot do, because the actions of the player are never constrained. The player can choose to build only workers and make them dance in formation. More importantly, the player can decide to conquer their allies just because they think fighting wars is more fun than building up culture or tech. The player is not "constrained" by the diplomatic penalties imposed by being a warmonger, since the player can choose to ignore them completely.

The computer can not choose to ignore certain variables. The computer cannot choose. That is the primary and only difference between the computer and the player. You can create the most complicated and in-depth system for the computer to evaluate the player's hypothetical diplomatic penalty, but it will be wrong 100% of the time because each player is a human being with human agency. We can see this in the conversation we're having right now: you think the AI accidentally hurting you in diplomacy is an unforgiveable sin that mars the game, while I think it's fine and to be expected. That is literally our difference in reaction to the AI's decision. That difference in reaction absolutely cannot be taken into account by the AI. It doesn't know how desperate you are, it doesn't know how self-righteous you are, it doesn't know how bored you are.

It can't know any of these things, and therefore any system that tries to guestimate them will always be more cautious than one that doesn't. And an AI that's more cautious is less likely to win, because the game always rewards taking risks. To win the game you need to push yourself ahead of the pack in some way, be that through size of territory, winning the wonder races, or whatever. A computer that takes less risks is a computer that is less threatening, and since interacting with the player is 100% unpredictable and risky in the first place, crippling them through predicted caution can only hurt them.

edited 25th Apr '14 11:57:59 AM by Clarste

The computer actually already does this to an extent, because it can tell whether or not an AI leader would be angered by Congress proposals and give you that information. Things like "this leader uses a lot of X resource, so abolishing it would make them angry" and "this leader has a large standing army, so instituting a standing army tax would make them angry" or even "this leader trades a lot with this other leader, even though they're not friends, so embargoing them would make them angry."

These are simple conclusions that the AI can very much know about (it has access to all that information about the player, it just doesn't utilize it) and would be able to use: it can't know how the player would react for sure, but it could certainly make assumptions and conclusions based off of what it knows about the player. That's how computers strategize in games like this - Galactic Civilizations in particular is acclaimed for it. The AI does not currently do it here because it for some reason isn't allowed to access that information and instead acts as though it doesn't exist.

edited 25th Apr '14 12:05:19 PM by KnownUnknown

I posit that such information does not in fact exist. That is my premise. You will have to convince me that it does. The fact that you, personally, feel "restricted" by considerations of the AI's reactions (as directly forecasted by the UN interface) is itself evidence in favor of my position.