When win-win games get gamed
[ by Charles Cameron — a paradox I’m currently chewing on, courtesy of Richard Landes]
.
I want to take something that Richard Landes has been saying, abstract it from those conflicts to which Richard applies it, simplify it by removing one technical term that’s part of his detailed breakdown, and present it in as bare-bones a manner as I can manasge. You can see two of Richard’s own versions of the issue below.
**
The basic question, as I understand it, is this:
How does a non-zero-sum move look to a zero-sum player, and vice versa?
Let’s suppose your entire background and upbringing revolves around the idea of zero-sum, you-win-I-lose games. If you see any sign of your adversary gaining an advantage in a negotiation, that’s proof positive that you’ve been bamboozled: somehow, you must have lost. Your job is to make no concessions — to be the winner in a winner-takes-all contest.
Now suppose your background and upbringing have revolved around the idea of non-zero-sum, win-win games, in which both sides of a negotiation make some concessions so that both can emerge as winners. You offer concessions in good faith, and in return you expect similar concessions.
What happens when a zero-sum game player and a non-zero-sum game player meet in play?
It’s Landes’ suggestion that any generosity on the part of a win-win player offering his winner-take-all opponent a concession will be taken as evidence of weakness,. and the opponent, far from making a corresponding compromise, will press on and demand more — making further nmoves which will offer so little that the win-win player will be at a loss to explain why such a promising start to bnegotiations fell apart so badly.
A further supposition: so-called honor-shame societies are geared for zero-sum, winner-takes-all gameplay, while societies which rely on an innocence-guilt reading of human behavior will be no less inclined toeards playing non-zero-sum, win-win games.
That’s the idea expressed in game-theoretic terms, as simply as I can put it. To my mind, these game considerations are worth thinking through in their own right, absent the specifics to which Richard Landes brings them.
**
Here is Landes making the same point, in the context of Israeli relations with the Palestinians, the Arab and Muslim worlds, and European liberalism –all of the above broadly speaking:
And a text version, Cognitive Egocentrism, for those who find words fly home faster in print…
**
Help me think this through.
August 3rd, 2015 at 6:25 am
Charles can you elaborate on what you mean by “innocence-guilt”? I’m trying and failing to read that as differing from what you call “honor-shame” – thanks!
August 3rd, 2015 at 6:34 am
In an honor-shame society, what matters is the perception people have of you. It doesn’t matter whether you’re innocent or guilty, what matters is what you’re believed to have done — so if you’re innocent but believed to have done wrong, you (and your family) are dishonored, while if you did something wrong but got away with it, with nobody suspecting you, your honor is intact and you can hold your head high.
.
In an innocence-guilt society, is doesn’t (“really”) matter what others think of you, if you know you’re innocent you’re innocent — even if a court finds you guilty — and conversely, if you’re guilty, even if nobody knows it, you’re guilty.
.
One is based on perception, the other on actuality.
August 3rd, 2015 at 12:28 pm
A couple of points. The first is a false premise and internal example of cognitive egocentrism from the article:
“… a dogmatic projection of a Western culture which has, by and large, renounced violence…”
Western culture has not renounced violence, it’s the most successful culture in history at violence, but to outsiders, not insiders. A near/far dichotomy clears it up somewhat; vast clouds of pink mist are fine as long as they are across an ocean and can’t be directly seen. The centrism in the statement is from the near perspective and ignores what has been done far away.
The other point is that the game theory framework generally assumes a two-party conflict. Win-win situations are often cooperative, and the more players that cooperate, the better they’ll do against a field of non-cooperators. The effects of that cooperation may seem mysterious to the win-win individual, suddenly appearing as though they lived in Flatland. If they do figure it out, they’re likely to regard the cooperation as cheating and try to force it out of existence.
August 3rd, 2015 at 2:16 pm
I have been pondering this too, in the context of win-lose Vladimir Putin and the win-win West. That’s not to say that all of the West goes for win-win, just that win-win is closer to what the EU has been trying to do since WWII.
August 3rd, 2015 at 10:18 pm
I think I see the contours of the same issue in this para about Murray Gell-Mann and Richard Feynman, from The Jaguar and the Fox:
What happens when a “natural born” cooperator plays a “natural born” competitor in Prisoner’s Dilemma?
August 4th, 2015 at 4:05 pm
“What happens when a “natural born” cooperator plays a “natural born” competitor in Prisoner’s Dilemma?”
One or the other changes the rules.
August 4th, 2015 at 8:47 pm
I wasn’t aware of it before, but T. Greer mentioned the simple hueristic ‘Tit for Tat’ last week on twitter:
https://en.wikipedia.org/wiki/Tit_for_tat#In_game_theory
The cooperator has the advantage
August 4th, 2015 at 8:58 pm
I should clarify that- it isn’t really the cooperator but actually the imitator or emulator.
August 4th, 2015 at 10:15 pm
Ah. My question was about a single round of PD, not the iterated version. Once we get into iteration and the possibility of changing strategies we lose the force of habit & assumption that I was trying to convey with my words “natural born” in each case. There’s an assumption pon one side that we’re all focused on eiwin-win, right, and an assumption on the other that we’re in a winner-take-all contest. It’s the impact of those two primal assumptions, eacb of which views the adversary as conceptually just like itself, that I’m after.
.
Once we get into iteration and implicit flexibility, the whole point I’m working on disappears.
.
For iterated PDs, Axelrod’s The Evolution of Cooperation and The Complexity of Cooperation are basic. Does anyone have other, more recent texts to suggest?
August 5th, 2015 at 12:28 pm
A couple of examples of win-lose advocacy.
.
1. Quotes from Henry Kissinger on a “grand strategy” for the Middle East.
.
2. No Ribbentrops to Putin’s Molotov.
.
I’m not going to get into the questions about grand strategy, just pointing out that Kissinger evidently believes that the US can order nations in the ME about and set up the circumstances, as Putin would like to come to a grand bargain with the other superpower (assuming Russia is a superpower). Definitely win-lose (at least for Ukraine) in Putin’s case, more ambiguous in Kissinger’s.
.
But both depend on worldviews of us vs them, and Kissinger’s is close to win-lose, although it’s not explicit. It’s not clear that any of the views in the ME are win-win, except for the people who would take it as a win to live their lives in something like democracy, without war and with enough to live decently. I assume that they are significant numbers outside the ruling classes and armies – the participants in some of the more peaceful demonstrations of the Arab Spring. Their side of the confrontation hasn’t gone well.
.
Conversely, Putin is trying to engage the US in a M-R sort of game, and the US is not engaging. A little clearer win-lose vs win-win, I think. Must be very frustrating to Putin. Not ended yet. So far it’s a stalemate. It really looks like Putin doesn’t know what to do next.
August 6th, 2015 at 2:48 pm
For multiple rounds, apparently a few years ago Freeman Dyson discovered strategies that beat tit for tat, the extortionate strategy:
http://edge.org/conversation/on-iterated-prisoner-dilemma
It works when the opponent that you are tit for tatting discovers that you’re doing it. They can then manipulate the outcome to change your results. Theoretically there are ways they can dial your results lower than theirs, with the lower you go the higher your opponent goes.
The catch is that their countermeasures can be found out, in which case you then add countermeasures, the advantage then collapses, so it isn’t evolutionary stable without extraordinary measures to conceal the extortionate strategy.
August 6th, 2015 at 5:20 pm
Fascinating, Grurray, and very much in keeping with the ways adversaries in practice learn from one another in accord with the evolutionary perspective.