Vote Now on the Prisoner’s Dilemma Contest

We recently posted a contest, asking readers to choose the one question they’d ask if picking a partner to play the Prisoner’s Dilemma.

I did not expect this contest to generate more than 350 replies. Picking the single best out of 350 seemed impossible, so I thought we should winnow it down to the Top 5 and ask you to vote on the best.

But I happened to be out at the University of Chicago last week, and ran into someone who I realized would do a much better winnowing job than me. In fact, he’s probably more qualified to do so than just about anyone else in the world.

It’s our friend John List, a top-drawer experimental economist, whom you’ve read about on the blog before. I think you’ll agree that John’s explanations for why he chose what he chose are fascinating and illuminating. Thanks to all of you for participating and especially to John for his contribution. And don’t forget to vote.

What’s the Best Question to Select Your Prisoner’s Dilemma Partner
A Guest Post by John List

When Stephen recruited me for this chore, the marching orders he gave were simple: select the 5 best responses to his Prisoner’s Dilemma (PD) query.

Sounds simple, but actually making the choices was far from easy. The group here is a sophisticated lot that covered a great deal of ground and provided an array of novel responses. Nevertheless, at the cost of leaving out several worthy posts, I have narrowed the list to my Top 5. I attempted to include queries within the realm of “what is the best question to parse individual types,” to “what is the best way to coerce people,” to just outright fun queries.

A first rule in my family is that “If mama ain’t happy, nobody is happy,” so I included my wife Jen List‘s question, No. 114, in the Top 5:

“How old are you?”

Beyond familial ties, I like this response because it shows evidence of being a good consumer of the economics literature. For instance, in my own research I have found that age is importantly correlated to play in this class of games.

When analyzing outcomes from a game show called Friend or Foe, where players are playing the two-person PD game live while staring into each other’s eyes, I found that mature contestants tend to cooperate much more than younger ones. This is the case regardless of their partner’s age.

In a second paper, “Young, Selfish, and Male: Field Evidence of Social Preferences,” I conduct a field experiment that uses an anonymous n-player PD game.

The results on age and cooperation were quite similar and have been replicated in more recent economic experiments as well. For instance, Gächter, et al. (2004) found that older people are much more likely to perceive others to be helpful and fair, and, as a result, they are more likely to act cooperatively themselves.

Post #288, jrrd, gets at something similar as does post #343, Sarah, so I will work to convince Da Freaks to hand out extra schwag if this entry wins!

The second finalist provides a query in a similar spirit, but involves a type of “treatment” variable more in the spirit of indoctrination — post #342, G. Owen Schaefer:

“What is the number of ethics courses you’ve taken, minus the number of economics courses you’ve taken?”

Owen provides supporting documentation here.

One will readily learn from this literature that much of the research that examines differences in cooperativeness between economists and non-economists suffers from selection issues. That is, people who chose to become economists may do so because they were born or raised with a slant towards self-interestedness (another interesting question in and of itself!).

Some studies, however, have produced tentative evidence that economics training itself has effects, by comparing behavior among freshman and senior economics students (here, for example).

I should note that many others on the blog were also tuned into this type of query (e.g., #11, Craig, #129, Allison, #159, Cyrus).

The third finalist — post #268, Joe — goes about the differentiation in a unique manner by asking about the famous example used in the original Beautiful Mind movie about Nash‘s life:

“Given that you are in a bar, would you prefer to pursue the most attractive person in the bar, or would your efforts focus on someone less attractive?”

This seemingly superficial question gets at the following dilemma: if a group of young men are sitting in a bar and a group of women walk in, where one of them is particularly attractive — is it an optimal strategy for all of the young men to pursue the most beautiful woman? As Nash’s character reasons in the movie:

“If we all go after the blond we block each other. Not a single one of us is going to get her. So then we go for her friends. They will all give us the cold shoulder because nobody likes to be second choice. What if no one goes for the blond and we don’t insult the other girls? That’s the only way we win.”

To be fair, this is not exactly what the PD game is about but I was persuaded by the post’s originality. In fact, the one-shot PD game discussed in this post has what is called a “dominant strategy” to testify.

That is, regardless of what the other person does, it is always in your best interest to defect, or in this example, testify.

To see this most clearly, make a 2×2 matrix with your choices as the row entries (testify or remain silent) and the other person’s as the column entries. Insert the payoffs in terms of years in jail and you can see in a one-shot PD game that it is always in your best interest to testify. This is of course the crux of dilemma.

In fact, this famous example from the movie did not even provide a non-cooperative equilibrium to the game that Nash played that night (I fully suspect that night at the pub was made up altogether, but that is for a different blog post).

In the movie, all of the guys chose the marginal women as Nash directed, leaving the most physically attractive woman to seemingly go home alone. That certainly is not an equilibrium — if Nash played his own non-cooperative equilibrium concept, he would have changed his mind in this case and gone home with the most attractive woman!

In this manner, the movie was off in that it missed the essence of a non-cooperative equilibrium. Disappointing in and of itself, but enough of that, let’s move on.

The fourth entry is of the coercive spirit and is provided by post #257, Charles:

“What is the name and address of your most cherished family member?”

There were several entries of this form — posters noting mafia connections (#21, #48), contract killers (#316), rapists (#124), passing soap (#233), sexual molestation (#229), people’s mothers (#195); even Klondike Bars (#240) were mentioned.

But this particular entry induced the most cooperative spirit from me. Maybe it’s the family thing coming back to haunt me.

Finally, the fifth entry was given first in post #147, Colin:

“Do you read Freakonomics?”

This, of course, can serve as a useful question to parse individual types, but also might be important in its own right. Who wants to go to prison with someone who has not read Freakonomics? Recall that if you both testify you will be sent to jail for 5 years, so why not spend it with a fellow Freakonomics reader? I can think of much less desirable company.

So there you have it. Apologies to those I left off the list, and congrats to those who made it. As they say here in Chicago, vote early and vote often — and don’t be afraid to bring a few of your dead friends and relatives to the polls.

So here are your five choices:

1. “How old are you?”

2. “What is the number of ethics courses you’ve taken, minus the number of economics courses you’ve taken?”

3. “Given that you are in a bar, would you prefer to pursue the most attractive person in the bar, or would your efforts focus on someone less attractive?”

4. “What is the name and address of your most cherished family member?”

5. “Have you read Freakonomics?”

Whichever choice gets the most votes in the comments section within 48 hours of this posting (barring obvious fraud) will receive her/his choice of Freakonomics schwag.


I believe it all depends on your own interests. Do you wanna establish the (coöperative) behaviour of the other person? Or do you want to convince him to cooperate? If cooperation is your first choice by preference, then you would like to know what the other guy would do. Will he cooperate or defect? Asking for his age is the best option in that respective. However, when you know you are going to defect, then you want to convince the other guy to cooperate. Question number 4 would be the best option in that case.




#4 is my insured bet too- we all need it. But then are we not assuming that the addressee can be found. Try calling for a phone number these days. Here is one instance where competition has made matters so much worse- you get old numbers, wrong numbers, fax numbers, no answer, 1800 non working numbers, dial a horoscope numbers i.e., all the wrong numbers- might as well go on line and look- takes longer, but I eventually get what I need. The peter principle at work- in organizations that think that improving technology and getting rid of the human touch increases profits do not know that not only employees, but employers rise to the level of incompetence when it comes to what works and what doesn't.


I would vote for the #4, because out of all the 5 it gives the best chance for you to choose the person to coerce with...

Although #5 is attractive, its the most dangerous question to be put forth with...

Shannon Prestridge



#2. Self-interested "rational" behavior is socialized rather than innate, so knowing something about a person's socialization is a better way to forecast their future actions in the game than, say, knowing about their age (the correlation that #1 is going for is just a correlation, not a causal mechanism) or their reading habits as #5 requests (they might read and dislike Freakonomics). #4 only makes a difference to the game if the implied threat is credible, and I'll disagree with Dr. List and argue that #3 tells us nothing useful about gaming behavior because of the idiosyncrasies of standards of beauty -- a PD game with clearly-specified payoffs eliminates that ambiguity.

The question that didn't make the top 5, the one I would really ask, would be something like: "is success doing well for yourself, or doing better than others?" That gets more directly at PD logic, and while it's not a perfect predictor I'd warrant that asking the question just before making a selection of a partner also invokes a social norm that might affect the subsequent course of play (provided that you don't choose the person who answers that success is doing better than other people, of course). In that way the asking of the question might help to alter the social context of the game by getting your potential partner to start thinking of the game in a different way -- in this case, as a more cooperative endeavor.



I vote for # 2

Craig (the real one who proposed the question - which can be verified by checking the email and IP address.)


"In fact, the one-shot PD game discussed in this post has what is called a "dominant strategy" to testify.

That is, regardless of what the other person does, it is always in your best interest to defect, or in this example, testify."

This is not true, the prisoners dilemma does not have a dominant strategy. Both cooperating is better than both defecting.
The prisoners dilemma is a dilemma because the lack of communication forces both players into a mutualy harmful strategy. You don't want to be the sucker that has to sit in prison for 10 years so you choose defect, your partner does the same and you both go to prison for 5 when you could have gone to prison for 6 months.


#3 because it tests their reasoning


#4 (of course)



It seems to me that all these questions suffer from not being answered truthfully. Especially if the person you are asking is smart.
So number 4 comes closest to being a threat wrapped up as a question and therefore the better choice.

Steve L.

I love #4. It makes the indefinite nature of PD much more visceral. Plus, a little fear never hurt.




#4, not only does it make the person invested in the game. It creates the implicit expectation of your own silence. That added benefit is crucial.


I have to say its all be OHT. Too much of it! And I have a negative vote for ALL choices!

-1 to 1.
-1 to 2.
-1 to 3.
-1 to 4.
-1 to 5.

Conor - ireland

#3 definitely...

Uch, i hate freakonomics, I'm beginning to see aptonyms everywhere, the top 5 list gets drawn up by none other that John List... and this week the 100m sprint world record was broken by Usain BOLT... hilarious no?


Surely #3



Allen Baranov

1 gives you a better chance of guessing which way to go but not all the way there. I think this is my second choice.
2 would give you a "huh?" but ethics fall apart when faced with really bad odds.
3 is an interesting way to ask "when we are faced with the dilema, will you sell me out?". Why not just ask the guy straight out and be done with it?4 is my choice!
5 is a nice question but doesn't get a perfect answer.

More information on choice:
My assumption is that the person has to answer correctly. If this is the case then he is quite aware of your intentions and with a simple question - you have really told the person "I intend to plead innocent, I suggest you do the same".

If the guy has the option of lying then no question you ask will help you make a decision - question #1 being the one that will most likely get an honest answer.


I was trying to figure out how the answer to #3 would affect my own decision.

With that question, you're doing two things. You're implying that you know game theory, and you're asking whether they also know game theory. If they do, then they know the dominant strategy, but they also know it's not optimal. So if they DO know game theory, would you assume they went for the dominant strategy even though it's not optimal, or would you assume that if you both know game theory, you'd cooperate and get the optimal solution?

If it were me, I'd have picked a game theory problem not in a blockbuster movie though -- Lots of people would spout off the answer from the movie regardless of what they think or know.

#4 is interesting because it tries to change the results of testifying. Instead of 5 years or scott free, you have potentially killed or potentially killed. It'd probably be the most effective though.

So I'm thinking #4 is the best question (assuming the other has to be honest), but #3 is more interesting.

I wonder what would happen if you went with #4, and the most cherished family member was already dead? Aieee!