torekp comments on Diplomacy as a Game Theory Laboratory - Less Wrong

44 Post author: Yvain 12 November 2010 10:19PM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (93)

You are viewing a single comment's thread. Show more comments above.

Comment author: GuySrinivasan 13 November 2010 01:07:17AM 13 points [-]

Diplomacy rant/warning!

Diplomacy culture is different in different cultures. The exchange between Eero Tuovinen and Valamir (Ralph) in this thread is particularly fascinating, and brings up a very good point: the game of Diplomacy is defined by the people agreeing to play it. If you do not agree beforehand what is within the game then you are playing different games, which is a very weird situation. "Let's play a game!" "Okay, e4." "Um, I rolled doubles so I go again. What's this about pawns?"

If you do agree beforehand, then you're all playing the same game. But two groups could easily choose different games and still call them both "Diplomacy"... here's Valamir's chosen game:

The entire design of the game of Diplomacy is to illustrate the trade off between short term gain and long term plans and the consequences for spending your political capital (aka trust and reputation) frivolously. Making a killer hose move to win a game today is SUPPOSED to have consequences on your ability to win in the future. That's the point...is winning this game really so important to you that you're willing to permanently decrease your odds of ever winning again with this group of people? If yes...then do it. If no...then don't do it and play for the allied victory instead. But don't do it and then whine about the consequences. Living with those consequences is what the game is about.

whereas here's Eero's chosen game:

You'll notice that I disagree very much with the idea that the board strategy in Diplomacy doesn't matter because you can always play the people. Assuming that everybody is playing in good faith and is actually trying to win without metagame considerations, then board strategy very much plays a part. You can only negotiate about board strategy, after all, and your options for alliances and stabs are limited by what happens on the board. Therefore a successful alliance of convenience (distinct from the alliance of honor or psychology or whatnot Ralph advocates) is only possible when you succeed in creating a plan of movement that puts everybody's units in positions where betraying your allies is not to your benefit. To put it simply, an alliance can be solid by investing enough resources in positioning troops in such a manner that betraying the alliance has a negative expected value. This has nothing to do with whether you believe in the other player's promises as promises, and everything to do with whether those promises make tactical and strategic sense for you and him.

These are different games. Eero claims the second is a better game qua game. I tend to believe him. If during the same "game" one person thinks they're playing Valamir's and one Eero's, it's no wonder dysfunction results.

Comment author: torekp 18 November 2010 11:13:19PM 0 points [-]

This point about Diplomacy culture illustrates one of the most important ways in which it's not particularly true that

The conditions of Diplomacy - competition for scarce resources, rational self-interested actors, importance of coalitions, lack of external enforcement mechanisms - mirror the conditions of game theoretic situations

as stated in the OP (emphases added). The players aren't self-interested because, in the usual case, they're playing with friends and acquaintances. (Or even simply because they're typical human beings interacting with other human beings.) And the availability of external enforcement mechanisms has already been pointed out.

Of course, game theory doesn't actually require self-interested actors either. At least not in any sense of "rational self-interest" which goes beyond "rational interest" or, for brevity, "rationality".