The Prisoner’s Dilemma concept originated in the game theory branch of mathematics as a way of evaluating a person’s choice between two actions that involve another person. It was first published in 1950 by scientists working for RAND corporation, at that time a Cold War think tank. Another famous contributor to the concept was mathematician John von Neumann.
The name comes from a situation in which two prisoners are held for questioning. Either of the two can win big, being freed, by ratting on the other if the other doesn’t rat. If both rat out, neither gets much, just early parole after sentencing on a major charge. If both hold their tongue, they both get token sentences on lesser charges.
The concept caught on and there are now many variations on the original concept. It is powerful because it can predict how a person will act when faced with a choice that has social ramifications in how they are perceived by other community members.
The Hollywood Version of a Prisoner’s Dilemma situation is as follows:
You are watching a crime show. Two characters have decided to trade drugs for money. The price and quantity have been set earlier, and they have agreed to come alone to a park. A vigilant policeman is patrolling it, so the two have to exchange their bags without examining the goods on the spot. Each has two bags: Person A, a bag of money and a bag of cut-up newspaper; Person B, a bag of drugs and a bag of cornstarch.
The background music signals dramatic tension … close-ups of hands picking up one bag each … the two people walk towards each other … the handoff is made …
The Prisoner’s Dilemma question is: Which bag does each person bring to the transaction?
Some definitions of terms for this concept:
So if Person A brings money and Person B brings drugs, then each is a cooperator and the transaction is mutual cooperation. If A brings newspaper or B brings cornstarch, they are a defector, not doing as promised. If only one defects, we have a betrayal transaction—and, especially if defection has earned a really big prize, a chase-and-revenge story for the rest of the show.
In mutual defection, something not commonly seen in drama, neither side gains and both point accusing fingers.
In real life, mutual defection is frequently avoided by setting up expensive protections. In the “Let’s Make a Drug Deal” case above, mutual cooperation is ensured by carrying out the transaction in an abandoned warehouse, with both parties backed by armed-to-the-teeth gangs, and the bags being examined on the spot. It’s a lot more expensive in effort, a lot less flexible in arranging, but the parties still have some profit.
Entertainment stories are full of transactions where one party cooperates and one party defects.
In real life, most arrangements that humans make with other familiar and friendly humans have a very high probability of mutual cooperation. I call these Double-Cooperator transactions.
When life feels good and life is easy, it’s because most of the human-human interactions a person is experiencing are of that variety. We exchange the bulk of our hard-earned cabbage for cabbages, cab-rides, kitchen cabinets, and other legal goods and services from various people and institutions, and we plan on doing similar deals with them in the future, perhaps many more times. It is very likely that we will all do as promised, and everyone benefits without very much attention to protecting themselves.
Another environment where double cooperator is beneficial is when a group of people are working on a project that is brand new and innovative, such as designing a personal computer in the late 1970s (before IBM jumped into the game).
Low-protection Double-Cooperator parameters work best in environments where most of the following are true:
Not that there are no safeguards at all on such deals—in some places you have to show ID to use your debit/credit card, just like you used to with a check, there’s probably an established merchandise returns policy, and the like—but they aren’t major. At least, not as long as the stakes are low compared to the likelihood of mutual cooperation.
When that balance changes, so the chances of defection are higher or the potential losses in case of defection are greater, then the protections become more elaborate and expensive. In real life, when parties sign contracts, that’s an added expense for protection. Building more protection into a transaction—greater incentives for each party not to defect—is the symptom of a Double-Defector transaction. This lowers its profit, flexibility, and speed, but doesn’t make it hopelessly bad. Particularly compared to a low-protection Double-Cooperator situation in which one or both parties do defect.
Highly protective Double-Defector parameters work best when most of the following are true:
This last point explains why bankruptcies and divorces often get messy. In end-of-the-line deals there is a lot more temptation to defect and “you did so first” accusations are easy to make.
If too much protection is applied to a transaction, it becomes frustrating and low profit. A suspicious father having his daughter’s dates sign contracts before taking her out is farce, not real life.
If too little protection is applied it invites betrayal. For example, when a first-time tourist comes to Times Square, New York City, to “score”, and he or she encounters a long-time Times Square con-artist who preys on tourists, the chances of Person A getting good drugs from Person B are infinitesimal.
“What evil lurks in the hearts of men? The Shadow knows!”
So went the opening lines of a radio mystery series of the 1930s. The good/evil matrix has been around a long time as a way of explaining people’s actions. The Bible story of Cain and Abel is based on it. So … how does the cooperation/defection matrix concept compare as a way of predicting the pattern of action that will be taken by a person when they are given a choice?
As I stated at the end of the Introduction to this book, the good/evil matrix works pretty well in anticipating what fictional characters will do, but only hindsight lets us apply it comfortably to real people.
In real life, “evil” is the label put upon someone who’s scared “good” people a lot before s/he finally lost against them, like Adolph Hitler. It is also put on those perceived to have betrayed “good” people, like “the evil bankers” who caused the economic misery of the late 2000s’ recession.
These are after-the-fact labels, without predictive value. Hitler was voted into power to save Germany from even deeper ravages of the Great Depression. He was made Time’s 1938 Man of the Year for those efforts, not because he promised to put up concentration camps. Likewise, no more people were calling bankers evil in the late 1990s or early 2000s than in the previous decade. “Good” and “evil” are easy to apply after the fact.
On the other hand, we can use the cooperator/defector matrix to examine real world risk and reward, and predict: “What is my risk for taking a certain action? What is my reward? What is the risk and reward for the other person involved?” Answering these questions is much more often useful for predicting human behavior.
It’s 2005. I have a house. I have a mortgage. I happen on hard times. I can continue paying on the mortgage (cooperate) or I can kiss it goodbye (defect).
Since housing prices are steadily and consistently rising, the banker—the other party in this transaction—will want to refinance (cooperate) and sees little risk in doing so. I don’t mind because my hard times are likely to be transitory, and there’s excess value in the house. I’m happy to cooperate, times are good, and the banker is a nice guy for helping me through this rough spot.
It’s now 2008. I have the same house. I have a new mortgage. I happen on hard times again. This time … things are different. Will the outcome be the same?
Using the Prisoner’s Dilemma concept, it’s easy to predict that the banker won’t refinance and that I will likely walk. (With or without that matrix, it’s not hard to predict I will call the banker evil.)
The Prisoner’s Dilemma concept can explain transactions between people and groups of people. It has predictive value.
We’ll use it to examine more real life transactions in Book Two.