Would you like to play a game?
YOU ARE A CONFECTIONER!
You are advertising your delicious brand of sweet, sweet candy, "Bananamazing", in your home town of Rythingeve.
HOWEVER, another firm in the same town has their own brand of candy: delicious "Kiwincredible". You both want to maximise profit. But how?
Advertising will increase profit, but advertising also costs money... by which I mean, XP.
If neither Bananamazing nor Kiwincredible do any advertising, both will sell in the same amounts, and, in addition, the companies get to keep all that cash they had earmarked for advertising costs - great! You both profit by 3XP.
On the other hand, if both Bananamazing and Kiwincredible buy advertising, they will still both sell the same, but both have paid money to do so. Obviously not preferable! You both profit by 1XP.
BUT! If one company advertises, but the other one DOESN'T, the company that advertised gets ALL the sales! And the other guys get zip! The guy who advertised gets 4XP. The other guy gets nothing!
Now, you can talk to the other company, sure... but you have no guarantee that they will tell you the truth! You can't believe a word they say. You might as well be working in the dark. And you can only act once - once you decide, you're committed to your decision for... let's say, the next month. No take-backs.
So, in order to maximise profit, what do you do? Do you:
- Cooperate, and buy no advertising, hoping your competitor cooperates too? Or do you:
- Defect, and buy advertising, hoping your competitor is a cooperative sucker?
To put things more clearly:
| They | They
| defect | cooperate
You | 1XP each | 4XP for you
defect | | 0XP for them
You | 0XP for you | 3XP each
cooperate | 4XP for them |
Or, more succinctly:
| D C
D | (1,1) (4,0)
C | (0,4) (3,3)
"Fine," I hear you cry, "a game for two players. What if there're more people?"
Well, scaling this up to multiple players is very simple:
- Everybody simultaneously picks an action, C or D.
- Everybody is played against everybody else, once only.
If, for example, we had 20 players altogether, and they all answered D, then each individual player has chosen to defect in 19 different games - one game against each of the other players. As a result, each player receives 19XP.
On the other hand, if all 20 players answer C, then that's 3XP per game multiplied by 19 games per player, which equals 57XP per player!
Got all that?
I can see gears churning in your head already.
Background information to this game
The Prisoner's Dilemma of the title is much better explained in that node than I could manage here. You will note there are no prisoners in this candy-marketing scenario, but that's intentional, because in the traditional PD, all the payoffs are negative. And that would make for a lousy game, since there would be no incentive for anybody to join! However, the situations in this candy-marketing scenario are analogous with respect to the various forces coercing you to cooperate or defect.
The idea of playing a one-shot PD game with many participants comes from Douglas R. Hofstadter's book, Metamagical Themas, in which he plays the game with 20 of his associates for the very same stakes, except with real US dollars.
And why am I holding the game at all? Because it seems like it'd be fun! And I think we could get some interesting responses once it's all over. And I have some ideas for follow-up games...
27 noders elected to participate. Their names were kept anonymous at this stage. I did not participate myself so as to remain impartial.
Participants were encouraged to 1) try to identify other participants and talk amongst themselves and sway each other's votes and 2) assume that all other noders were lying, backstabbing, cheaters, wanting nothing more but to maximise their XP at all costs, who might not even be in the game.
"May I make a tentative prediction? I would like to put it on record before the results are in that I predict %100 of participants will cooperate, or failing that, all but one."
The final vote was secretly /msged to me on August 24, 2007. All choices were final.
The following 14 confectionery brands elected to cooperate:
Each cooperator received (3XP for each of their 13 fellow cooperators) + (0XP for each of the 13 defectors), making 39XP in total.
The following 13 confectionery brands elected to defect:
Each defector received (4XP for each of the 14 cooperators) + (1XP for each of their 12 fellow defectors), making 68XP in total.
"The mathematics behind this game is fairly simple. Regardless of what everybody else decides to do, you are guaranteed more XP if you defect than if you cooperate. However, if you cooperate, everybody else gains more XP than if you defect. This game, therefore, is a measure of how altruistic E2 is as a community. The result? Pretty altruistic."
"I'm generally a player of the "always-generous" strategy -- which I fully realize is not always a strong strategy in non-repeat scenarios. Howeever, I have faith in my fellow E2ers."
"I'm defecting because I'm a selfish bastard."
"I will defect, because I'm a very bad person."
"Everything2 has a fine community of upstanding, model citizens, full of progressive social spirit, who all wish to succeed *together*. 'D', of course."
"If I really wanted the XP I'd play as a Defector because it's a one-shot vs a lot of unknown untrusted strategies, Tit For Tat not applicable, but I'm up for the interesting POV."
"I choose DEFECT. Yay capitalism."
"wah wah wah... I'm cooperating because I'm a bleeding heart, granola chomping, chai drinking, friendly, low xp-noder like AspieDad. Ha! I don't trust him for a moment. The cold war was won on paranoia, clandestine operations and key defections from the soviet leadership and research echelons. Give me a D!"
"There were a couple of reasons: I am unaware to the other prisoner's intentions, I am curious to experience new feelings such as being free of my personality which is a reason why many games are made with character choices e.g. the good and dark side, I acted as a character, there was no detrimental effect that I am aware of apart from the use of experience points."
"I was really hoping you were going to publish the names of the confections prior to your opening the ballots. Recruitment seemed an important part of the game, because, any additional participants, provided that they didn't sway others votes, automatically increased the potential of XP at no cost to others. Now, if those participants were prone to defection, they would, of course, make others prone to defection as well. However, it is in the interest of all involved parties to recruit people who will cooperate -- because a cooperating person not only makes cooperating more attractive, but also is more profitable whether you are a coooperator or a defector. As such, if it looked, based on the candy names, that there were a lot of defectors, it would encourage participants to recruit more players to increase the odds of cooperators -- and potentially sway the behavior of others. Anyhow, my second intent here was just good old irony. Defectilicious will choose to cooperate."
"What can I say? I did the math and I like to win."
"I defect! Because I'm greedy, and because it's the only way to guarantee a result."
"I'll defect. this way, I am guaranteed at least some XP."
"How is this for an answer: I would cooperate because 1) advertising and salespeople irritate me, and 2) if the other person does advertise, I have lots of candy I don't sell, which means I get more for myself to eat. (But I wonder if you should include admin people separately, as they could give themselves as much XP as wanted.) (I also wonder if there is a correlation between how much XP someone has vs. what they vote.)"
"I would like to enter, choosing option C, because the world would be a better place with less advertising."
"DEFECT!!! I will buy advertising at laugh at those who don't. Scoff even."