This joke has been internetting for the past week. (Karakul kick: Noam Nissan)
Here’s the game theorists’ version: Three game theorists with identical preferences but asymmetric information walk into a bar. The server asks “Does everyone want a beer?” They respond in sequence:
- Game Theorist #1: “Yes!”
- Game Theorist #2: “Yes!”
- Game Theorist #3 “I don’t know.”
15 comments
Comments feed for this article
September 26, 2011 at 7:16 pm
kerokan
I don’t get either of those jokes 😦 care to explain?
September 30, 2011 at 11:21 am
j-conn
Kerokan: The question in both is does EVERYONE want a beer: the first logitician wants a beer (otherwise she would have said no), but doesn’t know if the other two do. The 2nd wants a beer and knows the 1st does too, but does not know about the third. The third now knows that the other two want beer and so does he, so he says yes.
The game theorists have the same preferences (they all want beer), but asymetrical information (the first two already know that everyone wants beer, the third is not aware of atleast one of the other two’s preferences.
September 30, 2011 at 11:26 am
j-conn
Actually, I’m wrong about the game theory one (didn’t see common value) and just noticed other people have answered. See Jeff’s for a proper explaination.
September 26, 2011 at 8:34 pm
H
winners curse?
September 26, 2011 at 10:15 pm
PR
Kerokan: for the logicians, the first two guys don’t know if the third guy wants beer, so they have to say “don’t know”, the third guy knows that the first two guys want beer, otherwise they would just so “no”. the third guy knows he wants to have beer too, so he says “yes”.
but i don’t get the game theorist’s joke either.
September 26, 2011 at 10:38 pm
Evan
yeah, the game theory one took me a while (and by a while, I mean I had dinner in between reading it and getting it).
The game is asymmetric information and common value: all three guys value the beer at some common but unknown amount, and each has a private signal that provides an unbiased ‘estimate’ of that common value.
The first two guys have a private signal that is above the cost of the beer, so they want beer (and assume everyone else will too). The third guy has a signal that is beneath the cost of the beer. If he were to answer first he would have said no, but because he infers from the answers of the first two guys that they have positive signals, he is now unsure whether to trust his signal or not.
I don’t think it works as well as the logician one. I might have to try it out in my advanced topics game theory class tomorrow.
September 27, 2011 at 7:17 am
Rajiv
Shouldn’t the game theorists be saying probably, probably, probably not?
September 27, 2011 at 1:50 pm
jeff
rajiv, yes that works too, maybe better. i was trying to stick to the same language.
September 27, 2011 at 1:38 pm
jeff
Think herding but with a unanimity twist. (“does everybody want beer?”)
The common value aspect is the key. If they had private values then they are back in the logicians joke. In the logicians joke, each guy wants to say “I want beer but I don’t know about these other guys.” They can effectively send that message by saying “I don’t know.” (If you didn’t want beer you would have just answered “no.”) The third guy learns from the answers that the first two want beer, he knows he wants beer too, so he can conclude that yes, they all want beer.
In the common value joke, each individual is only partially informed about the quality of the beer. So no individual knows for sure that *he* wants beer, but *he* wants beer if and only if they all want beer.
Each guy wants to say “if the other two know that the beer is good here then we all want beer.” For the first two guys this means they have to say “yes” no matter what they individually know about the beer. Saying no means no beer, saying yes means they get beer if the other two say yes.
But now the third guy is stuck because he knows that the first two say yes no matter what they know about the beer. The third guy learns nothing about the quality of the beer from the first two answers so he has to say “i don’t know.”
September 28, 2011 at 8:51 pm
patrick
Isn’t the common value the sequence by which the question has been answered? It’s true to respond that you don’t know if ‘everyone’ wants a beer, if you are the first and second respondent but not true if you are the third respondent – hence the positive response. The third (last) respondee has taken their cue from the person who asked the question, indicating that ‘everyone’ is in fact the question poser and the affirmative respondent, with the ‘don’t knows’, (logically) correct in their response. The joke is that logicians shouldn’t be equivocal in a bar (else they wouldn’t be there) and that if they are in a bar, it’s no different to being anywhere else where they wouldn’t know what ‘everyone’ wants. The question poser knows this, but equally the joke is on him/her because he would expect the correct answer is ‘I don’t know’ and the wrong answer ‘yes’ is from the fool at the end.
September 28, 2011 at 9:14 pm
patrick
Isn’t the common value the sequence by which the question has been answered? It’s true to respond that you don’t know if ‘everyone’ wants a beer, if you are the first and second respondent but not true if you are the third respondent – hence the positive response. The third (last) respondee has taken their cue from the person who asked the question, indicating that ‘everyone’ is in fact the question poser and the affirmative respondent, with the ‘don’t knows’, (logically) correct in their response. The joke is that logicians shouldn’t be equivocal in a bar (else they wouldn’t be there) and that if they are in a bar, it’s no different to being anywhere else where they wouldn’t know what ‘everyone’ wants. The question poser (CORRECTION) DOESN’T know this, but equally the joke is on him/her because the correct answer is ‘I don’t know’ and the wrong answer, ‘yes’, is from the fool at the end.
September 29, 2011 at 3:48 pm
Mild Speculation
This is too much fun.
I’m sure this could be improved but as a first shot, how about 3 computational complexity theorists?
The server asks “How long will it take you to decide if you all want beer?”
The answers are “I don’t know,” “I don’t know,” and “HALT.”
Unfortunately the outcome is not well-defined (the last guy’s preference is unknown). Maybe we could have infinitely many complexity theorists and someone says “HALT,” but that would mean no beer, which is a depressing ending.
September 29, 2011 at 10:16 pm
Anonymous
Think SPNE
September 29, 2011 at 10:18 pm
Bob
Think SPNE
June 27, 2013 at 12:01 am
Three Mind Your Decisions Readers Walk into a Bar… - Mind Your Decisions
[…] are some funny variations of this, like a game theorist version at Cheap Talk and a coffee version at (x, […]