The Winner of Our Prisoner’s Dilemma Contest Is …

We ran a contest asking readers to submit the one question they’d ask to help pick a partner for the Prisoner’s Dilemma.

Then we had a special treat: the University of Chicago economist John List (whose writings, by the way, were the inspiration for the contest) agreed to comb through the 350+ entries and choose the Top 5. He did a great job, explaining the logic behind each choice. Here are his Top 5 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?”

Then we asked you to vote on the best answer. (Btw, here’s an interesting blog post dissecting the choices.) The winner was clear: Question No. 4, with 35 percent of the vote.

The runners-up were: No. 2 (24 percent), No. 3 (23 percent), No. 1 (16 percent), and No. 5 (2 percent).

I am surprised that so many people went for No. 4 — the question that implicitly threatened physical harm. Why was this so? It may be that as much as we modernists embrace the notion of “conflict resolution,” what we really want is simply conflict.

So congratulations to Charles, who submitted the question, and will get to pick his favorite piece of Freakonomics schwag.

Mike Zillion

Funny, I didn't read question number four as an implied threat. I read it as a test of the level of intimacy and connection the answerer has with his family. Do you even have a most cherished family member, and can you recall his or her address from memory? That's a telling quality.


"It's better to be feared than loved.", Niccolò Machiavelli


I really don't understand why the question "implicitly threatened physical harm". This is, to me, an odd interpretation, for example could it not be a signal of in-group-out-group biases and therefore be a basis for parochially altruistic behaviour, or reciprocal altruism and therefore indicate whether one should or should not cooperate with the other depending on their answer? This idea of a threat is a bit Hobbesian for me and contrary to the mountains of evidence in which people actually cooperate in PDs. Oh well...


anyway- the answer I liked the best wasn't part of the top 5, it was just some way of trying to ascertain how the person feels about "snitching" since those would be the best people by far to have in this situation.


Terrible winner. There was no stipulation on the question that the person had to answer it at all, let alone correctly. How many people would give that question a correct answer? And if you could guarantee a trustworthy answer, why not ask him for his bank account #'s instead?


George Tenet Fangirl and Erik have this spot on. Far from eliciting cooperation answer 4 (which implicitly threatens the family of the other prisoner) almost guarantees mutual defection.

Also why decide by popular vote? The same dummies who provided all the dumb answers have now chosen a dumb winner.


As several people have pointed out, question 4 is not guaranteed a right answer, but there is another problem with it as well. Question 4 does nothing to help you decide which one of the three people to choose from. Maybe I didn't understand right, but I thought the purpose of asking the one question was to pick your partner. Three names and addresses doesn't help you differentiate between the three. There has to be some sort of useful variation, so that you can pick the person who is most likely to not confess. I'm not sure why this question won. But hey, democracy got W. re-elected, so this isn't its biggest lapse.
Regardless of the question and the answer, the solution to the Prisoner's Dilemma remains the same: confess. You still can't risk the scenario occurring where you don't confess and your partner does.


Put it this way, in situations where you are worried about someone testifying against you do you think that people have tried to find out someone's economic background or the address of family members more?


#11 and #15 are absolutely right. Has no sense the answer must be the truth and were not implicit in the problem as it was enunciated.
Quite strange to understand as "the best solution" something voted by most of the people.
Best politicians are always the winners of the elections? You use democratic methods when you do not have something better at hand.


I'm not so much interested in the answers to these questions, but would like to know if anyone has gathered data on what happens in real-world prisoner's-dilemma cases. How often are crooks able to keep their traps shut and minimize their punishment, and how often do they cave in? In our household, we watch so much crime TV that's probably not very realistic (e.g., Law & Order), that some real-world data would be great.

Rich Wilson

Maybe I'm missing something, but I don't see the point of #4. There's no gaurantee that you'll get a correct response, right? Nobody is going to lie about having read Freakonomics. Someone is very likely to lie about where their family members are if asked by a fellow suspect.


I think you are playing the role of polite host when you say you are surprised. If I really thought you were surprised, I would probably stop reading your journal and your work. As it is, I will go on assuming that you, like everyone else, is less than perfectly honest about some things -- in this case your opinion of the human condition.

It's hard to imagine any educated (non-autistic) adult without living in this world who doesn't recognize that almost every person on this planet is perfectly willing to harm other people severely for personal gain, and that many people even jump at the chance. This is why dictatorship works -- a dictator needs x number of people under his control who are perfectly willing to destroy each other and their own underlings to retain power or gain even a little more. It's easiest when those x people cover a wide range of skills -- enough to run a corporation or government.

There are a lot of dictatorships.



very interesting, age I guess would be a very important factor, but I wouldn't pick any of these questions, now im not sure what kind of question i'd ask but none of these seem that great


2, 3, and 5 are all similar in this list (trying to find out how knowledgable the oponent is) but 4 is the only threatening one. If 2 or 3 was deleted from the choices, the other might have won, or if an instant run off voting system was used. This result is more a statement about the voting set up used rather than which answer is best.

I'm not surprised 4 won, I would have chosen it.

George Tenet Fangirl

While that's definitely not the question I would ask, it IS the question I would most want asked of me by my associate. After all, once I told them about my fictitious Aunt Penny who lives at 3421 B St, I could safely betray them and walk away free.

Seth Burn

This really is fairly obvious: we want tangible leverage. In a traditional prisoner's dilemma defection strictly dominates cooperation. We don't want the person to trust us. We want the person to RATIONALLY act in our best interests. Giving us leverage over the other individual makes it rational for them to act in our best interests. Mutually assured destruction is a lot safer than "Trust me", or, how well do you play complex intellectual games?


Question #4 does not require a truthful response. It is meant to be a threat but, like Jeopardy, the rules of the game stipulated that it be phrased in the form of a question. It is essentially a roundabout way of telling your partner, "if you rat me out, I will have my associates track down and take revenge on your family."

From a rational perspective, in a single round of prisoner's dilemma, there is absolutely no advantage to co-operating. Defecting *always* gives you the better payoff, no matter what the other player chooses. The only way you can expect your partner to cooperate is if you assume that they are irrational (a very risky assumption) or if you give your partner the impression that a defection now will have future consequences that ultimately make it a worse option than cooperation. If you fail to make your partner believe that there will be additional negative consequences to defecting (or additional benefits to cooperating, though this is more difficult to do because of the payoff matrix) you virtually guarantee they will defect.


Michael F. Martin

Maybe we don't know what we want, and when we don't know, we assume that whatever the other person has is what we want. Conflict ensues.


I am not surprised, I voted for #4. This is how organized crime keeps people from talking. Whether or not the respondant answers honestly the idea of just the willingness to inflict physical harm changes the payout matrix and thus the dominate strategy of your fellow prisoner.


The prize should be contingent on finding a SINGLE PERSON in prison who has taken either an economics or ethics course...