Diplo game thought experiment

[Krill]

> The whole concept of not learning as diplomacy is carried, that there is a separation between reputation, trustworthiness and risk are contradictory.

So are you saying that if this game were tried, NAPs and other agreements would be broken as often as they currently are (i.e. basically never) because everyone would still be playing for their reputation and future games? Even if there were enough players who explicitly agreed in the beginning to make an effort to treat the game as a one-off, in terms of reputation?

It's a testable claim, at least.

Pretty much.

[flugauto]

What would happen if a PB game were played with the following "rules"?

1. There is full diplo. Anyone can say anything to anyone.

2. No agreements are ever binding. Any moves allowed in game can be made opportunistically as if it were AI diplo. (Double move etiquette still applies, obviously.)

3. Everyone must agree to make a good faith effort not to judge a player negatively for any action taken, either for any game other than this one, or as a reflection of character. This includes lying, going back on your word, deliberate deception, exploiting trust, and anything else within the game framework that can be deviously thought up.


Now, this really doesn't make a hard change to anything about a full diplo game, because #3 is only a cultural paradigm shift. It's always been really interesting to me how the culture we have for full diplo games stays how it is. (For instance, how NAPs are sacred, and someone who breaks one is actually judged as a person, and not someone who is playing a game for an advantage inconsequential to anything else.) I'm convinced this goes beyond the pragmatism of seeking better relationships in future games.

So what would happen if #3 were made explicit? Would it change much? Would it be possible? Would it even be worth trying to talk to other players? (I think the answer is clearly "yes".) Is human nature too strong for people to detach themselves in this way?

The reason for this post is I've been thinking about the ideal game I'd like to play months later when PB11 winds down, and I'm wondering if it wouldn't be fun to have a game more like an AI diplo game in gameplay, but where we could actually talk about the game too.

How about a diplo game without NAPs, without any deals? This way you don't have to break anything, but can talk freely.

[suttree]

I think Jowy is getting closer to the heart of things here. It's not a matter of people forgetting anything - its introducing the same sense of psychological distance you get from a no diplo game. You "ought" not to trust anyone and you are "good" because you act in your own self-interest.

After all, the meta in no diplo is the same, and all the players know the meta is the same. And yet in a diplo game they play differently because someone sends them an e-mail that, for all they know, could be a complete lie.

Imagine an alien watching us playing civ4 () In one game we play in a certain way (no diplo) with limited information. In another we have access to the same information but perform a strange ritual involving pixels on a screen (diplo) completely divorced from the game and so behave differently. Weird!

The different behaviour in diplo and no diplo has nothing to do with "the game" and everything to do with the culture that surrounds the game (people who break agreements are bad, information shared via e-mail can be trusted) a culture that is reinforced by the bad feelings and resentment that WillLP refers to.

Will wants to play a diplo game with a no diplo mindset - something like Werewolf civ4. Since most players are cultured to play in the current style, this would take some work on the part of the players (ie. make a conscious effort to frequently lie for advantage) and perhaps some ingame changes to reduce trust and inject a sense of danger. But it seems interesting, or at least worth talking about.

[Jowy]

...Except you can't tell people that their opponents are oath-breakers and cowards when people know that they aren't to various degrees. Just as you can't force people to attack without pause in Always War.

Your sentence still doesn't make much sense to me, Krill.

You know the saying "fool me once, shame on your, fool me twice, shame on me"?

Basically, that. That's the distillation of what the players are supposed to just forget, right?

Would that be the player drawing the wrong conclusions though? I'd imagine that usually when agreements are broken, it's because the player puts himself in a really vulnerable situation that makes it too tempting for the other guy to break the agreement. So the lesson to be learned would be to not blindly trust someone just because you have an agreement with them, rather than the lesson being that you can never trust that guy on anything or that you should never play with that guy.

So to apply your point to NAPs: Player X signs an NAP with 2 players. Player A breaks it, Player B doesn't. Player A could benefit or not from breaking the agreement, that doesn't matter.

Who does the Player X sign future agreements with?

He wouldn't sign agreements with Player A if he fails to realize why he was betrayed. Most of the time he wasn't betrayed because that betrayer is an evil son of a bitch who just wants to watch the world burn. He was betrayed because he left himself too vulnerable. I think agreements are very useful, but you have to have a good mix of agreements and natural deterrents. You can't just leave your cities empty and hope that your agreement keeps you safe. Right now if we were to play a game, I would imagine that most players will not realize this, and will just rely on their NAP's, but eventually the meta game would evolve to a state where what I described is the default way to treat agreements.

[WilliamLP]

Will wants to play a diplo game with a no diplo mindset - something like Werewolf civ4. Since most players are cultured to play in the current style, this would take some work on the part of the players (ie. make a conscious effort to frequently lie for advantage) and perhaps some ingame changes to reduce trust and inject a sense of danger. But it seems interesting, or at least worth talking about.

I think that's accurate - and your prison description and quote from Ellimist were great!

It might also be only a small subset of players capable of playing like this - able to play for months and still treat it as just a computer game manipulating bits around for no other purpose.

I take trust extremely seriously in real life, and I also would never break a NAP in a normal diplo game here (and would probably never play one), but I'm very sure I'd have no problem being "that guy" in a game where everyone was willing to say upfront that anything goes.

[Krill]

He wouldn't sign agreements with Player A if he fails to realize why he was betrayed. Most of the time he wasn't betrayed because that betrayer is an evil son of a bitch who just wants to watch the world burn. He was betrayed because he left himself too vulnerable. I think agreements are very useful, but you have to have a good mix of agreements and natural deterrents. You can't just leave your cities empty and hope that your agreement keeps you safe. Right now if we were to play a game, I would imagine that most players will not realize this, and will just rely on their NAP's, but eventually the meta game would evolve to a state where what I described is the default way to treat agreements.

Not how it works. Understanding why someone will betray you does not mean that is the only reason someone would betray you. Only that they will betray you and that they may betray you again. However, Player B, due to the trust that accumulates can also leave his border empty and so Players B and X are both benefiting. Noble is 100% correct with his description of the problem, quoted below.

One reason why people don't break NAPs is that it's essentially an iterated prisoner's dilemma. If you know the game will end at the current iteration, the optimal choice is to defect. Thus if I decide that I'm never coming back to RB, I'll break all my NAPs and win my last game and then disappear. But nobody decides that, and thus NAPs are upheld. (The other reason is that people think it's immoral or unethical to break a NAP.)

Likewise, breaking a NAP within the game will cause opponents to view you as more likely to break NAPs in the future within the game. Thus that is part of the consideration in deciding to break a NAP. Thus the earlier in the game it is, the less likely you are to break a NAP. (This may carry over to future games and thus there may be no effective change, I haven't thought enough about this to decide whether that's the case.) You can't just instruct players in the game to consider the likelihood of player A breaking a NAP as equally likely as player B. It's a judgment on each player's character, same as a player gauging player A as more aggressive than player B.

[WilliamLP]

Right now if we were to play a game, I would imagine that most players will not realize this, and will just rely on their NAP's, but eventually the meta game would evolve to a state where what I described is the default way to treat agreements.

I wonder that too, if all it would take is a few people willing to treat the whole experience (including diplo) as just a game, and if that alone could actually start to shift the whole culture a little bit.

A person could even reason that they're doing a greater good? Because the actual game may be more interesting for all! (In the way that most seem to prefer the on-board gameplay of AI diplo.)

I think that long agreements being unbreakable actually leads to a lot of the poison in full diplo games. If you know you're going to attack me on turn 160, and would now, if you could, and it's now turn 120, then we have a really awkward relationship that just stews and builds tension over a long time.

[suttree]

Step 1: mod the game to remove the power graph
[edit] heck remove the graphs all together - no objective measure of strength in game

[Krill]

He wouldn't sign agreements with Player A if he fails to realize why he was betrayed. Most of the time he wasn't betrayed because that betrayer is an evil son of a bitch who just wants to watch the world burn. He was betrayed because he left himself too vulnerable. I think agreements are very useful, but you have to have a good mix of agreements and natural deterrents. You can't just leave your cities empty and hope that your agreement keeps you safe. Right now if we were to play a game, I would imagine that most players will not realize this, and will just rely on their NAP's, but eventually the meta game would evolve to a state where what I described is the default way to treat agreements.

Not how it works. Understanding why someone will betray you does not mean that is the only reason someone would use for betraying you. Only that they will betray you and that they may betray you again. However, Player B, due to the trust that accumulates can also leave his border empty and so Players B and X are both benefitting. Noble is 100% correct with his description of the problem, quoted below.

One reason why people don't break NAPs is that it's essentially an iterated prisoner's dilemma. If you know the game will end at the current iteration, the optimal choice is to defect. Thus if I decide that I'm never coming back to RB, I'll break all my NAPs and win my last game and then disappear. But nobody decides that, and thus NAPs are upheld. (The other reason is that people think it's immoral or unethical to break a NAP.)

Likewise, breaking a NAP within the game will cause opponents to view you as more likely to break NAPs in the future within the game. Thus that is part of the consideration in deciding to break a NAP. Thus the earlier in the game it is, the less likely you are to break a NAP. (This may carry over to future games and thus there may be no effective change, I haven't thought enough about this to decide whether that's the case.) You can't just instruct players in the game to consider the likelihood of player A breaking a NAP as equally likely as player B. It's a judgment on each player's character, same as a player gauging player A as more aggressive than player B.

[suttree]

Step 2: all teams start in contact but open borders agreements are banned.