Chapter 1 Jailbreaking the Prisoner's Dilemma I (Lieberman) was 10 years old at the time of the Cuban Missile Crisis, in 1962. It really scared me, because it seemed like there was a real possibility that a nuclear war might bring the world to an end. Russia had set up nuclear missiles in Cuba, 90 miles from Florida. US President Kennedy issued an ultimatum that they be removed. Or else. Was Kennedy really prepared to start a nuclear war to back up his threat? Would Russian President Khrushchev and Cuban President Castro launch the missiles preemptively, or in retaliation? Tension was high. Many were seriously expecting a nuclear war, because the normal behavior of political and military leaders is: once you make a threat, you have to be willing to back it up. Most of all, though, everybody feared the fragility of the situation. Nobody thought Kennedy nor Khrushchev had any real intention of starting a nuclear war. But we realized that if they went through the normal political and military processes, one misstep, and nuclear war would be the result. In school, I remember going through air-raid drills, where weÕd be told to crawl under the desks in the event of sirens announcing impending war. But I grew up in New York. Even as a child, I knew enough to realize that if the Russians dropped even a single bomb on the Empire State Building, considering where we were, weÕd all be vaporized. The air-raid drills were useless. It was at that moment that I came to the realization: adults didnÕt know what the heck they were doing. The discussions on TV about the situation sounded insane. Nobody proposed any sensible way out. The first job of politicians and military leaders was supposed to be to assure that citizens of their countries were safe. But at that moment, it seemed like the leaders themselves were the biggest imaginable threat to the worldÕs citizens. There must have been something wrong with all the political and military processes that were supposed to protect us, if they led us to the absurdity of that point. What was wrong was that the leaders did not recognize that their interactions formed a particular kind of pattern that was heading to disaster. WeÕre going to tell you about that pattern: The PrisonerÕs Dilemma. Introducing the PrisonerÕs Dilemma In this book, weÕre going to talk about a wide variety of social and economic problems and propose some solutions. But thereÕs a unifying themeÑthe tradeoff between cooperation and competition. The thesis of this book is that the root cause of many societal problems is getting this tradeoff wrong. The political Right extols the virtues of competition; the Left extols the virtues of cooperation. Neither side acknowledges that sometimes itÕs better to cooperate, and other times, itÕs better to compete. So admitting that fact is the first step. But once we get that far, thereÕs still the problem of how to determine whether to cooperate or compete in a particular situation. In this chapter, were going to give you some tools for how to think about the issue. WeÕll show you an important concept from mathematics, but donÕt worry if you donÕt have a strong math background, itÕs not difficult. We will walk you through it step-by-step, and there are only a couple of very simple formulas. There are also a number of very good online videos that use animation to teach you the basics. A good one is from Scientific American [Moyer 2012], and a simple search on YouTube for ÒPrisonerÕs DilemmaÓ will find several others. Even mathematicians, though, donÕt fully appreciate the extent to which this pattern really does describe many real-world situations. They donÕt always connect the dots between what they learn from the mathematics, and its implications for society. ThatÕs our job in this book. If youÕre already familiar with the PrisonerÕs Dilemma, you might be able to skip ahead to the section WeÕre All Prisoners of Reality, below, but a refresher certainly wonÕt hurt. Now, letÕs start. The reason itÕs called the PrisonerÕs Dilemma is because of the following story, used in the literature to present the problem. Two suspects, Bonnie and Clyde, are arrested and held in prison by the police. The police have insufficient evidence for a conviction, and, having separated both prisoners, visit each of them to offer the same deal: If one testifies for the prosecution against the other and the other remains silent, the betrayer goes free and the silent accomplice receives the full 5-year sentence. If both stay silent, the police can sentence both prisoners to only one year in jail for a minor charge, like possession of a weapon. If both betray each other, they both will receive 3-year sentences. Each prisoner must make the choice of whether to betray the other or to remain silent. However, neither prisoner knows for sure what choice the other prisoner will make. So the question this dilemma poses is: What will happen? How will the prisoners act? Suppose youÕre Bonnie. The choice you have is either to stay silent, or to betray your partner, Clyde. What do you do? Of course, emotionally, your fellow suspect is probably your friend, and you have feelings of loyalty to him. You donÕt want to play the snitch. The cop is offering you the deal in the hopes that your selfishness will overcome your feelings of loyalty. But the point of this story isnÕt to depict an emotional struggle between loyalty and selfishness. For the moment, weÕll put aside thinking about the emotional impact of the situation. WeÕll focus on the question of what might actually be in your self-interest, in the narrow sense of which choice is more likely to result in less jail time for you. Then, weÕll look at how the situation shapes up for both you and your partner. We figure it out using a branch of mathematics called game theory, because it applies both to games like chess and poker, and also to decision-making situations in real life. Keep in mind that game theory is a way of abstracting the situation by pointing out the mathematical pattern in the story. That way, we can apply the lessons learned to many other situations, whether or not prisoners are involved. The pattern is solely about whether the sentences the prisoner gets are better or worse, depending upon the choice they make. ItÕs the pattern that counts, not the story itself. The story is only there to motivate, to help you think about it. So you canÕt get the prisoners out of their dilemma by suggesting, e.g. that they bribe the cops, hire a good lawyer, accuse a third party, have their friend bake a cake with a file in it, tunnel out like El Chapo, etc. The PrisonerÕs Dilemma, one step at a time WeÕll figure it out, step by step, by thinking about each combination of choices by both prisoners, then comparing them. With a little patience, youÕll learn about one of the most remarkable paradoxes in all of mathematics! Imagine youÕre Bonnie. The choice you have to make is whether to stay silent (cooperate with your partner Clyde, to foil the prosecution); or to betray Clyde, in the hopes of getting a lighter sentence. Mathematical game theory uses the term cooperate for choices like staying silent (cooperating with your partner and not cooperating with the cops). It uses the term defect for a betrayal, or any kind of choice that is competitive (or uncooperative) with your partner. To make it easy to follow, weÕll draw a set of diagrams, each representing a combination of choices in the shape of a ÒVÓ. The tops of the two arms of the V represent the choices by each of the prisoners. The point at the bottom of the V represents the outcome, the number of years in jail each prisoner gets. The two arms of the V ÒcauseÓ the result at the bottom point. We give each particular situation a name, which appears in the middle of the V. The best way to follow the explanation is to look at the picture, read the text, then go back and look at the picture again. HereÕs our first example. Suppose both Bonnie and Clyde stay silent, that is, they both refuse to accuse the other one of committing the crime, (while maintaining their own innocence). In this case, according to the original PrisonerÕs Dilemma story, they both get one year in jail. If neither confesses, the best the prosecution can do is convict them on a lesser charge, like possession of weapons. This is the best the both of them can do, as we shall see (itÕs not possible to get both of them off completely). We call this situation the Reward. Now, as Bonnie, you think, ÒSuppose, instead, I betray Clyde?Ó, accepting the prosecutorÕs deal. Remember, weÕre still assuming, for the moment, that Clyde is staying silent. That would lead to the following situation: In this case, you get off scot free, but poor Clyde is left holding the 5-year bag. Because you are tempted to betray your partner by the promise of freedom from jail, weÕll call this situation the Temptation. Now, letÕs look at both the Reward and Temptation situations together. In this diagram, we connect BonnieÕs choices (silence or betrayal) in a light yellow (in color versions of this book). A darker orange is used to connect the prison sentence Bonnie gets in each case. In green, we show the Òteam scoreÓ that adds up both Bonnie and ClydeÕs prison sentences. That way, itÕs easy to visually compare them across the VÕs. Comparing the two situations, which one is better for you? If you stay silent, one year in jail for you, as shown in the upper left V, the Reward. If you betray, you get zero years. ThatÕs obviously better than one year in jail. So, (if you can get past your guilt in ratting our your partner), you give in to temptation, and betray him. But remember, all this is assuming you trust Clyde to stay silent. But suppose, instead, he betrays you? Then what? The next diagram introduces two new situations. In both of these, as shown at the upper right of each V, Clyde betrays Bonnie. At the upper left of each V, as before, we have BonnieÕs choices. You can choose to stay silent. In that case, you will be the Sucker who has to serve hard time, the 5 year sentence, while Clyde, the lucky devil, walks. The betrayer serves no jail time. Or, you can both betray each other, and then you both get the Punishment of the intermediate sentence. Three years is worse than 0 or 1, but still not as bad as 5 years. So, again, unfortunately, you decide to betray Clyde. LetÕs step back and look at the overall situation, with all four cases displayed. Your task, as Bonnie, is to choose between the top row and the bottom row. In the top row, you stay silent in both the Reward and SuckerÕs Loss situations. In the bottom row, you betray Clyde in both the Temptation and the Punishment situations. Now, we compare prison sentences vertically in the grid. On the left side, weÕre comparing ÒBonnie gets 1 yearÓ at the top right square, to ÒBonnie gets 0 yearsÓ on the bottom right. Fewer years is better, so the bottom row, betrayal, is preferable. On the right side, we compare ÒBonnie gets 5 yearsÓ at the top right to ÒBonnie gets 3 yearsÓ at the bottom right. Again, betrayal wins out. In both cases, the bottom row choice is better than the top row choice. So, no matter what Clyde chooses, if you, Bonnie, act selfishly, you will betray him. Tough luck, Clyde. But wait a secondÑisnÕt Clyde going to go through exactly the same sort of reasoning? WonÕt Clyde come to the conclusion that he should betray you, by exactly the same reasoning? Then, what weÕre left with, is the Punishment situation. Inevitably you both betray each other, and wind up with the 3 years in jail, apiece. But notice that this outcome is actually the worst situation for the team, with 6 years in the clink, total. The worst outcome! If only both of you had decided to stay silent, youÕd both have gotten off with only a year apiece, the best situation for the both of you. What a shame! ThatÕs the PrisonerÕs Dilemma. ItÕs a situation where, if you consider your options selfishly, you would choose not to cooperate with your partner, because it would seem better for you, regardless of what your partner does. He makes the same choice, and the result is that both of you defect. But if you had both cooperated, the situation would have been better, collectively, for both of you! It may seem paradoxical, but itÕs something that often happens in the tradeoff between competition and cooperation. What went wrong? If you just consider the situation from a local point of view (your own), you come up with an answer that doesnÕt achieve the best team score. But if you consider the global situation (both your own and that of your partner), you can do the best for everybody. And, as weÕll see later, cooperation is your best strategy in the long run. Take one for the team. What goes wrong, when both of you choose, selfishly, to betray each other? We can summarize this situation as: Think locally, fail globally. The obvious cure, (as one of our favorite slogans puts it) is: Think globally, act locally. ThatÕs where we are with the choice between competition and cooperation in politics, economics, and society. Too many individuals and organizations consider a situation from their own local point of view, and make a bad choice to compete with others they interact with. If they considered the more global situation (how their choices affect themselves and others), and, of course, if they could recognize in their situation the pattern of the PrisonerÕs Dilemma, theyÕd choose to cooperate. Things would be better for humanity as a whole. ThatÕs the lesson of the PrisonerÕs Dilemma. LetÕs run the numbers You may notice that the cops had to be really clever about choosing the number of years they offered the prisoners in the plea bargain, in order to trick them into the Punishment scenario. If there were different numbers, things would turn out differently. Prison sentences are a bad thing. To compare the desirability of various situations, we have to keep in mind that longer sentences are less desirable. So weÕll turn the numbers around, using negative numbers of years to describe the prison sentence. You could think of it as the number of years you lose from your life. WeÕll show the desirability by the length of the bar next to the prison sentence. If we stack up the prison sentences in the order of their desirability for you, Bonnie, it would look like this: To help you remember the order, weÕre going to give this formula a mnemonic name: The TRaPS Inequality. The TRaPS Inequality: Temptation > Reward > Punishment > SuckerÕs Loss T>R>P>S The PrisonerÕs Dilemma ÒtrapsÓ you into defecting when you should be cooperating. By definition, the term PrisonerÕs Dilemma refers only to those situations that satisfy the TRaPS Inequality. Later, weÕll see other kinds of situations that can occur. This is really the only math formula you have to know to understand the PrisonerÕs Dilemma2. We told you the math wouldnÕt be that hard! The fact that game theory recommends that each partner defect against (betray) the other in a single PrisonerÕs Dilemma game is a bummer. It might lead you to a pessimistic view of human nature, that it is inevitable that people will prioritize their own welfare over that of others, and the world would descend into a cesspool of selfishness. Indeed, that really happens in geopolitical situations, as weÕll see in our analysis of the Cuban Missile Crisis. But thereÕs hope for humanity in the story of the PrisonerÕs Dilemma. In the 1980Õs, evolutionary biologists found another way of thinking about the problem that leads to a much more optimistic viewpoint. Stick with us. From MADness to cooperation The PrisonersÕ Dilemma has a fascinating history. It was first discovered in 1950 by Merrill Flood and Melvin Dresher [Flood 1952], who worked at the Rand Corporation. They were working for the US Defense Department, studying scenarios of the possibility of nuclear war, as this was the height of the Cold War. As weÕll see in our chapters on War, this scenario does indeed apply to many situations involving potential and actual war. Though their original intent was to describe a theoretical problem, they soon realized the relevance of their work to the political situation at the time. It showed the folly of the policy of Mutually Assured Destruction promoted by the military at the time. The aptly-named MAD was the idea that merely the threat of nuclear war would be sufficiently abhorrent as to prevent nuclear war. Some even went so far as to think that nobody would dare start a conventional war because of the risk theyÕd provoke a nuclear war. Flood and DresherÕs work pointed out that if things ever managed to satisfy the TRaPS Inequality, weÕd risk winding up in a situation where nuclear war and total destruction would become almost inevitable. Ironically, Flood and DresherÕs argument was completely ignored by the Cold War military and politicians for whom they were ostensibly working. On one hand, you could say that MAD ÒworkedÓ because, fortunately, we managed to (just barely) avoid the nuclear disaster that Flood and Dresher warned about. On the other hand, though, we came pretty damn close. Recent historical investigation has uncovered just how close we really were. KhrushchevÕs memoir said that Castro had asked him to launch a nuclear attack on the US, a request that he denied because he disdained Castro personally, as ÒhotheadedÓ [Pear 1990]. LetÕs return to the Cuban Missile Crisis, with which we started this chapter, and see how it plays out with a PrisonerÕs Dilemma analysis. In this situation, either country could cooperate (by removing missiles threatening the other), or defect (by launching (or even threatening) a nuclear strike. Fear of being a victim of a first-strike nuclear war (SuckerÕs Loss) would seem worse than even the possibility of both sides being destroyed by each other (Punishment). The satisfaction of avoiding war (Reward) might not be as compelling as the Temptation to launch the first strike. A first strike would potentially knock out the retaliatory capability of the other side. The inevitable result, though, would be the Punishment of mutual destruction. Eventually, though, it seems Kennedy made a secret deal that removed missiles in Eastern Europe and elsewhere that were threatening Russia, in exchange for removing the Cuban missiles. On a more black-humorous note, the paradox was later satirized in popular movies, for example, Dr. Strangelove and War Games. From war to evolutionary biology The PrisonerÕs Dilemma remained a mathematical curiosity, until it was picked up in the 1980s, amazingly, by evolutionary biologists. The biologists were trying to understand the following puzzle: Evolution is based on competition, as at least a cursory reading of Darwin might indicate, encapsulated by the slogan, ÒSurvival of the fittestÓ. So, how is it possible for cooperation to evolve in human societies? Until that time, conventional evolutionary theories had a hard time squaring the competitive nature of Òsurvival of the fittestÓ with the observed tendency of organisms to (at least, sometimes) cooperate. Multi-celled organisms grew out of cooperation between single cell organisms. Mitochondria started out as independent cells, then became the energy source for larger cells. There are many examples in nature of symbiosis between plants and animals. Modern discoveries such as the intestinal microbiome continue to astonish at the complex nature of the cooperation between organisms. By the 1980s, computer simulations were being used by evolutionary biologists, so Robert Axelrod of the University of Michigan decided to run a tournament of simulated agents, so he could investigate the behavior of the PrisonerÕs Dilemma in a population of agents. One novel aspect is that the game is iterated. Players could play again and again, with the same partners or different ones, and have a memory of what happened in the past and use it to affect their strategies (though they werenÕt allow to communicate with each other, to simplify the analysis). This Iterated PrisonerÕs Dilemma (IPD) is the really important case. Each iteration of the simulation was a round-robin tournament of pairs of agents playing two-player PrisonerÕs Dilemma with each other. Each agent was controlled by a program that made the decision to cooperate or defect. Some agents had random strategies, but he also invited researchers to submit their own strategies, which would participate in the game. Agents who had successful strategies could ÒreproduceÓ within the simulation, so that more agents using that strategy would appear in the next round. Scientists could then see how the population of agents would evolve. Would the most ruthless, competitive agents win, or would the population of agents learn how to cooperate with one another? This led to AxelrodÕs 1984 book, The Evolution of Cooperation [Axelrod 1984], which formed the basis of the modern understanding of the problem. The results were startling. The upshot: When the game is repeated, the best strategy changes from defection to cooperation! So, indeed, cooperation can evolve. WeÕre not stuck with the pessimism of game theory telling us to defect in the single-shot case. What a relief! Why is that true? Well, in the single shot case, you may indeed gain an advantage by defecting. But that advantage comes at a longer-term costÑit will affect how that partner plays with you next time. Strategies that do best in the long term are those who choose to cooperate with a fellow cooperator. So if you defect, it makes you a less desirable partner, and makes it more likely that youÕll be playing with someone who will defect against you. ThatÕll hurt you. This speaks to a short-term vs. long-term tradeoff, and weÕll consider many such tradeoffs later the book. As long as you Òcare enoughÓ about the future, youÕll cooperate. There were many, many variations of strategies tested, and work continues to this day investigating all the possibilities. In AxelrodÕs initial work, a very simple strategy, submitted by Anatol Rappaport, turned out to be surprisingly successful in a wide variety of situations (but not all): Tit for Tat (TfT). TfT always chose to cooperate on the first round. From then on, it simply mimicked what its partner did on the previous round. Surprisingly, it won over more complex strategies. What can we learn from the success of Tit for Tat? TfT succeeds, not by defeating its partners (since it imitates its partners, it wonÕt do much better or worse), but by eliciting cooperation from its partners. On one hand, you might think of TfT as the reciprocity rule that most religions have, to encourage followers to have empathy for others. On the other hand, you might think of it as the equally Biblical Òan eye for an eye, a tooth for a toothÓ since it Òtakes revengeÓ whenever the opponent defects. But the implications of the revenge view were well understood by the game theorist Tevye, a character from the play Fiddler on the Roof: ÒThatÕll leave the world blind and toothlessÓ. Modern interpretations of the Biblical phrase suggest that the Hebrew wording is better read as Òonly an eye for an eyeÓ, counseling against disproportionate punishment. Modern variants of TfT insert additional forgiveness, even randomly, to avoid endless cycles of defection. This makes the system more robust in the presence of uncertainties. In a Justice system, the cost of administering justice is often borne by everyone in society, so everyone loses when one person commits a crime. This is unfair, but worth it, considering the inefficiency of lawlessness. Attempts to reduce the cost of Justice, say, by having a dictator decide, often lead to overall worse outcomes for society. Cooperation is not free, itÕs just better than the alternatives. More generally, the properties that make TfT and other successful strategies work, are that they are nice (never the first to defect), forgiving (will sometimes return to cooperation after a defection), and provokable (wonÕt always cooperate in the face of continual defection). Understanding principles like this gives us hope that we can develop strategies for PrisonerÕs Dilemma situations that better lead to cooperation. Astonishingly, it took until 2013 for somebody to get the idea of actually trying the PrisonerÕs Dilemma experiment with real prisoners [Khadjavi 2013]. And it turns out that real prisoners are smarter than game theory mathematiciansÑthey choose cooperation, even in one-shot cases, far more often than the mathematics would predict. No doubt this was because the emotional aspects of the situation (which, youÕll recall, we said we were going to put aside when we first told the story) kicked in. Perhaps, also, it was significant that this particular experiment was conducted with only female prisoners. Emotions like loyalty and empathy may well serve an under-appreciated functionÑto get us out of making poor choices in PrisonerÕs Dilemma situations! When competition beats cooperation Everything also has its ßip side. ItÕs important to note that, if the TRaPS Inequality doesnÕt hold, we donÕt have the PrisonerÕs Dilemma. We can even get the opposite situation, where cooperation fails and competition succeeds. For example, imagine that our cops think that the two prisoners have Mafia connections. They say, ÒWeÕre going to set up a competition between you two for who can give us the best evidence to convict your Mafia Godfather. If you both stay silent, weÕll throw the book at both of you, regardless of your guilt or innocence. The guy who gives us the best evidence that results in convicting the Godfather will go scot-free, and the other guy will take the heat. Whaddya say?Ó Here, the TRaPS Inequality doesnÕt hold. WeÕre not allowing ties, so there isnÕt any Punishment for mutual defection. The so-called Reward for mutual cooperation results in the worst team situation, both individually, and for the team. In that case, the best strategy for the prisoner is to deliver the best evidence against the Godfather that he can, even if it implicates his partner. Cooperation (and here we mean cooperation with your partner, not cooperation with the prosecution) loses, competition wins. Garrett Hardin [Hardin 1968] wrote about the Tragedy of the Commons, a classic exposition of the kind of situation where cooperation around a shared resource can fail. He described a situation where many people held a resource in common, for example, a pasture to graze animals. Each animal owner gets the full benefit of grazing his own animal on the common. If the common is big enough to handle all the animals, everything is fine. But if the common pasture isnÕt enough to feed all the animals (a situation of scarcity), the inevitable result will be that the shared resource will be exhausted. The attempt by the whole group to share a resource cooperatively with one another results in failure, unlessÑby goodwill, persuasion, or some mechanism that enforces itÑeverybody grazes only their fair share. WeÕll return to the issue of scarcity later, where weÕll see that scarcity both causes cooperation to fail, and makes cooperation more difficult. HardinÕs situation was essentially a multi-player PrisonerÕs Dilemma. But he wrote in 1968, before the full impact of the PrisonerÕs Dilemma was really widely known, and he didnÕt reference the original work. When leftists promote the virtues of collective cooperation, Libertarians are quick to counter with the Tragedy of the Commons, to remind them that shared resources donÕt always work. They recommend a regime of private property, which (at least, in a relatively even wealth distribution) allocates resources amongst participants to avoid exhaustion. So, if the numbers can go both ways, either favoring cooperation or favoring competition, then which is it? That, of course, depends on the details of the situation. In the upcoming chapters weÕll explore some of the variables that affect the outcome (weÕve already mentioned scarcity). But our general argument is that, due to technological change, these variables are moving in the direction of favoring cooperation over competition. You may have heard the business term, coopetition, a mix of cooperation and competition [Brandenburger 1997]. It refers to situations where, even though businesses may compete with each other, they might still be able to find particular issues on which they can cooperate. Google and Facebook may be competitors, but they can cooperate on things like supporting visas for immigrant engineers. We should encourage companies to be more cooperative, resisting their habitual instinct to be competitive in all things. But we really donÕt like the way this term is normally used in the business literature. Cooperation and competition arenÕt considered equally the way they are in the PrisonerÕs Dilemma analysis; cooperation will only be tolerated in the service of larger, competitive goals. So we wonÕt use this term. Another difficulty you may notice in applying these lessons to the real world, is that in many cases, weÕre talking about stuff thatÕs difficult to quantify. If weÕre talking about peopleÕs happiness, or loss due to death in a war, etc. these are intangible quantities and canÕt be as easily or accurately measured as years in jail. One response is to try to develop measures that stand in as a proxy for the quantity of interest. The government of Bhutan, noting that the traditional economic measure of the Gross Domestic Product (GDP) ignores many factors, committed itself to trying to maximize Gross National Happiness, by aggregating a variety of quality-of-life measures [Bhutan 2016]. In the case of the PrisonerÕs Dilemma, itÕs important to note that the central formula is an inequality, so it is relatively insensitive to the exact numbers for each of the factors. ItÕs their relationship that counts. WeÕre all prisoners of reality LetÕs look at how some of this plays out in some real world situations. War One of the most obvious, the most destructive, and the most absurd behaviors, is war. Why do nations go to war? Nobody will admit that they want war. So how can war possibly happen? If both sides cooperate with each other by eschewing war, both will benefit by having more resources to meet their citizensÕ needs. Neither really ÒwantsÓ a war that will result in death, injury, and waste of material resources on both sides. But they become more afraid of the possibility of being defeated militarily by an unscathed conquering foe, than they are of the situation where they get attacked, but they can attack back. Even if it causes equivalent loss for their own side. Of course, warmongers donÕt ever think about the loss of human life and cost to the other side. The official story is usually that the threat of retaliation will deter attackers. But the overwhelming historical evidence is: even when thereÕs a credible threat of retaliation, wars occur anyway. Overconfident adversaries simply dismiss the possibility that they might be defeated. So deterrence rarely works. To see this, simply ask the warmonger: If there were a significant chance of being defeated, would that deter you from proposing war? First, youÕd have to get past their conviction that defeat was impossible. Then theyÕd say that, even if defeat were possible, it would preserve their honor to Ògo down fightingÓ. ItÕs even worse than thatÑdeterrence canÕt possibly work, because military people take an oath that they will persist in trying to destroy the enemy even in the face of their own destruction. That is, both sides have agreed in advance that deterrence is out of the question. If deterrence doesnÕt work for them, how could they expect it would deter the other side? Often, just the possibility of war engenders such fear in people, that they believe they have to build up an army and weapons in ÒdefenseÓ. The military (and its suppliers and supporters) constitute a powerful force with an incentive for exaggerating threats, so that theyÕll feel needed and valued by the society, and they can profit from military preparations. If that occurs on both sides, then voilˆ, youÕve got war. WeÕve already alluded to the reasoning in our discussion of Flood and DresherÕs original work on nuclear war. But letÕs break it down here and think about the general situation. Each side always has a choice about whether or not to attack (or to take revenge for an attack, which here, amounts to the same thing). Suppose the Temptation to achieve military victory (or removing the threat) is perceived to exceed the Reward of a peaceful and prosperous society; and the prospect of the Punishment of both sides being damaged by a war, bad as it is, is at least considered better than the SuckerÕs Loss of being attacked while not having a military response. Then weÕve got Temptation > Reward > Punishment > SuckerÕs Loss. WeÕve got the PrisonerÕs Dilemma. It doesnÕt matter Òwho started itÓ (the favored excuse of the playground bully). Off to war we go! LetÕs review. If both sides cooperate by refraining from war, the result is peace and prosperity. If they both defect (attack), war breaks out and destruction happens on both sides. But neither side knows what the other will decide, so we have to investigate both possibilities. Since itÕs always possible the other side will attack (or the other side has already attacked), military action may seem more desirable than the status quo, since, if it is victorious, it will deal with the (perceived or actual) threat. A war is bad for both sides, of course, but it might still seem better than being unilaterally defeated by the opponent. At least we fought back, theyÕll say. So, for the nation that has to choose either the top row (donÕt attack) or the bottom row (attack), the bottom row seems to be better in both cases. So it chooses to attack. So does the other side, which is faced with exactly the same choice. Now you know why we have war: Fear of war causes war. We canÕt emphasize this enough, so weÕll say it again, in bold italics. Fear of war causes war. Franklin Roosevelt said, ÒThe only thing we have to fear is fear itselfÓ. Because the PrisonerÕs Dilemma hadnÕt yet been discovered, he couldnÕt have known how right he was. And if the public had understood how right he was, we probably could have avoided all the wars that followed World War II. Pretend wars and real wars ItÕs not always obvious, in a real situation, who the prisoners actually are. Another way to view the situation of war is that thereÕs a pretend war between the two nations that are ostensibly fighting each other. But the real war is between the military-industrial complex and the citizens, on both sides. So there are actually two real wars. US President Eisenhower warned of this in the speech where he coined the term military-industrial complex. This too, amounts to a PrisonerÕs Dilemma. The Temptation, for a military, is to enhance its prestige with its citizens, to recruit, to increase the military budget, increase opportunities for heroism, etc. It can do this by exaggerating the potential threat, thwarting attempts to negotiate treaties with the other side, making demands for resources, stirring up fear. Military contractors make large contributions to political candidates likely to support war, especially in districts where these contractors hold economic power. Military people feel most useful and engaged when actually fighting threats, what they signed up to do. This gives them a cognitive bias towards considering threats more likely than the average person. Military commanders stir up recruitsÕ anger and aggression, to make it more likely theyÕll take action instinctively should the need arise. This becomes a self-fulfilling prophecy. The SuckerÕs Loss for the military-industrial complex is to be considered useless by the population, and have its budget and personnel slashed. That would, of course, be better for the citizens, leaving more resources and reducing death and injury. So the military is actually defecting against the citizens. It implicitly cooperates with the so-called ÒenemyÕsÓ military, whose citizens are faced with exactly the same choices. The story told to Americans to justify the Cold War was that Communism was aggressively hell-bent on taking over the world, and it had to be stopped before it was too late: The Domino Theory. America was helping out third-world battleground nations by spreading democracy and free markets. When the Cold War finally ended, the West could finally communicate openly with people in the former Communist bloc. We could ask them, what were they thinking at the time? They werenÕt discussing strategy for taking over the world. They feared that American Capitalism was aggressively taking over the world, and had to be stopped before it was too late [Vorobiev 2016]. The very same Domino Theory. You can imagine how they might get that impression: a Soviet official visiting Angola and seeing people there drinking Coca-Cola, a sight he hadnÕt seen the previous year. Even if they werenÕt trying to take over the world themselves, of course they believed that Communism was a superior system to Capitalism. Russia was just ÒhelpingÓ innocent third-world nations by promoting Communism. Russia had to defend itself against America, the only country in the world who had used nuclear weapons, a proof of the insanity of its leaders. It was the RussiansÕ (paranoid or not) fear of America that led to behavior easily perceived by the American government as aggression. So much for war. Capitalism Now, letÕs look at commercial competition in our economy. Actually, many business people use the language of war to describe business competition, so the concepts arenÕt actually all that different. We live in a Capitalist society, and one of the founding principles of capitalism is competition between companies and competition between products. ÒCompetition brings out the bestÓ, capitalists say, supposedly to consumersÕ benefit. But that doesnÕt consider that you might lose out on the opportunity that would have happened had you cooperated, both with the customers and other companies. Neither it is considering the cost of duplicating effort also performed by the other company. So what weÕve got here is a cooperation/competition tradeoff. What better way to analyze such a tradeoff than the PrisonerÕs Dilemma? Commercial competition: Value proposition Companies are always faced with a choice. They can provide products that provide good value for their customers, charge fair prices and treat their customers well. Or they can provide products that are poor value, overpriced, and fail at customer service. The latter is often more profitable. Business people say they want to provide fair value and treat customers well, as captured by the business slogan, ÒThe customer comes firstÓ. But the Temptation for higher short-term profits, driven relentlessly by Wall Street, can sometimes overwhelm their sense of fairness. This is coupled with the fear of the SuckerÕs Loss of being defeated by competitors, or the stock price tanking as a result of low sales. Fear of other companies and Wall St. is the cause of poor value products. Commercial competition: Advertising LetÕs take the example of advertising. Some of the purpose of advertising is to make customers familiar with a product, or to grow the demand for that kind of product, not just the vendorÕs brand. But in the case of established companies in mature markets, like Coke and Pepsi, consumers already know the brands, and overall cola demand is unlikely to change much. So itÕs all about market share. To give you some perspective, the budget of the US National Science Foundation is roughly $7 billion. The advertising budgets of Coke and Pepsi combine to around the same $7 billion. Which does more good for the world? And, year in and year out, the market shares of Coke and Pepsi donÕt change very much, so their Òreturn on investmentÓ in advertising is essentially zero. So how does an ad agency convince Coca-Cola to spend on advertising? If both companies donÕt advertise (cooperate), they each have more money. But each company is Tempted by the prospect of success of their ad campaign, and fearful of the SuckerÕs Loss of the success of the otherÕs companyÕs ad campaign, so they are each stuck with their bloated ad budgets, and the rest of us have to waste our time watching the ridiculous, annoying, lying ads. Why do we have deceptive and annoying advertising that doesnÕt do any good for either vendors or consumers? Why do we have expensive sugar water promoting diabetes, almost identically from both Coke and Pepsi? Why did $7 billion go down the drain? Fear of competing companies is the cause of useless advertising. Again, we can take another point of view on the situation. The pretend war is between competing companies like Coke and Pepsi. But the real war is between both companies on one side, and their consumers on the other side. Another real war that takes place is between Coke and Pepsi on one side, and on the other, smaller competitors: beverage companies with less market power, startups, supermarket brands, etc. Environmental pollution Climate change, pollution, and other environmental crises can also be explained with the concept of the PrisonerÕs Dilemma. A polluter, like an oil or gas company, gets profit from the pollution, but makes an incremental contribution to global warming. If thereÕs too much of that, ecological collapse is the result. Why do we have unending pollution and no Carbon Tax, which a consensus of scientists and economists agree is a rational response to climate change? Fear of lower profits is the cause of pollution. Political gridlock LetÕs explain the gridlock in todayÕs US Congress. If both parties cooperated to solve national problems, it would be better for the public, whom they are ostensibly supposed to serve. But the competitive election system, and the organized bribery system, called ÒlobbyingÓ, put such fear in the hearts of elected officials that we get Why political gridlock? Fear of losing elections causes political gridlock. Racism, sexism and other forms of discrimination Even phenomena like racism can be modeled with the PrisonerÕs Dilemma. [Axelrod & Hammond 03] reported a simulation where a majority group cooperated with each other and defected with the minority group. It has nothing to do with the actual characteristics of the group, only that it is apparent which group a given agent belongs to. It formed a stable pattern. But of course, it led to suboptimal results for everybody. The Temptation of establishing privilege for a majority group is that that group thinks it might get some advantage from oppressing the minority. ItÕs fearful of being taken for Suckers by the minority, a fear that is usually baseless. Those who advocate the Temptation need that fear in order to convince the majority that the discrimination is necessary. But theyÕll miss out on the Reward of a diverse and harmonious society, which is better for productivity and certainly ethically preferable. So we wind up with the Punishment of perpetual strife between the factions. Similarly, any kind of nationalism, tribalism, sexism, homophobia, religious discrimination, etc. etc. is a loss for everybody. So why does it take place? Fear of losing out to a minority causes discrimination. ThereÕs also another pretend war going on here, as you might have guessed. The leaders of the majority use fear of the minority to gain power over the majority citizens. Reducing recidivism in the PrisonersÕ Dilemma In describing these very diverse problems in similar terms, we hope you can start to see the pattern. And if the problems all have the same pattern, the solutions to these diverse problems also have a pattern. Basically, in each case, we have to try to make sure that greed for the Temptation doesnÕt overwhelm the hope for the Reward; and the fear of the SuckerÕs Loss doesnÕt overwhelm motivation to avoid the Punishment. Then we can make the choice to cooperate. In war, we have to realize that the situation of peace and prosperity, the Reward, is the best thing that could possibly happen. We have to resist the Temptation of military victory promoted by the warmongers. Our security is better served by cooperation with the other side, promoting common interests to achieve peace, rather than fear of the SuckerÕs Loss of defeat. In commerce, the Temptation to produce shoddier products or raise prices might gain them some short-term profitability advantage or favor with Wall Street, but in the long run causes them to miss out on the Reward of customer satisfaction, well-being and loyalty. Fighting with other companies for market share at no benefit to the consumer is similarly pointless. Environmental crises will be just as bad for the polluter as they will be for everyone else. Companies should actually welcome regulation, since if everybody is forced to curb pollution, it can remove the fear of losing competitive advantage. Car companiesÕ profits didnÕt decline after they were forced to introduce seatbelts (and, as a bonus, they had more live customers!). Politicians should realize that cooperating across the aisle would be their best bet of being able to achieve concrete results they could brag about to their constituency. The Temptation of negative ads is that they may cause temporary declines in polling numbers for their opponent. And nobodyÕs fooled, as public approval ratings for Congress descend into the single digits. (Or perhaps everybodyÕs fooled, since re-election rates are high.) Cooperation across the aisle would better achieve the Reward of meaningful public service. And, as civil rights movements expand and succeed, people are gradually realizing that the benefits of racial, ethnic, and gender diversity far outweigh whatever benefits a dominant group feels it might get from maintaining discrimination. We wonÕt arrive at a single strategy, since it is clear that PrisonerÕs Dilemma simulations are extraordinarily sensitive to the population of agents, their history, and communication. But we are likely to get increasingly valuable insight into what makes cooperation and/or competition work, and under what circumstances. Understanding PrisonerÕs Dilemma situations is humanityÕs get-out-of-jail-free card. Recap This has been a lot, so letÕs review the main points. ¥ The tradeoff between cooperation and competition can be modeled by a table that shows the results of each possible choice for each participant. ¥ In some situations, if each participant considers only whatÕs best for themselves, theyÕll choose not to cooperate, or to compete (called defecting). ¥ But if you look at the overall situation, it would be better if everybody decided to cooperate. ¥ This paradoxical situation is called the PrisonerÕs Dilemma. ¥ The PrisonerÕs Dilemma only happens when the numbers in the table satisfy a certain mathematical formula, the TRaPS Inequality. Temptation > Reward > Punishment > SuckerÕs Loss This often happens in real life. It can explain situations of ÐWar Ð Failures of economic systems ÐPolitical gridlock and dysfunction ÐRacial, ethnic, and other kinds of discrimination ¥ We havenÕt yet found a general solution to the PrisonerÕs Dilemma, but we know several factors (like scarcity) that make it more or less likely. ¥ The key to solving many of societyÕs problems is recognizing when this pattern appears, and trying to influence the factors that give rise to it.