The Predictioneer’s Game: Using the Logic of Brazen Self-Interest to See and Shape the Future

by Bruce Bueno De Mesquita

  They would have to repeat this message convincingly through months of negotiations between the U.S. attorney and their attorneys. Their attorneys would have to keep going to meetings with the U.S. attorney in which they repeated the message that they were unable to persuade the board to show some flexibility, followed by a plea that the U.S. attorney must give them the ammunition to convince their board that the matter could be settled. That ammunition was for the U.S. attorney to stand up to the hard-liner government attorneys, shooting down their arguments in front of the defendant’s attorneys when they all sat in meetings together. That was the way for the U.S. attorney to flip the hard-liners’ gambit on its head. To achieve this end, the board had to go along with the idea of their lawyers insisting on going to meetings with the hard-liners in attendance rather than trying to have private meetings with the U.S. attorney as they actually preferred.

  Just imagine the board’s reaction. Their first thought was “Who is this guy with the chutzpah [not a word they would have used] to tell us what to do when our entire business is on the line?” Indeed, when I proposed the optimal monetary offer to settle the case, an offer way below what they thought should be put on the table, they thought I was nuts. When I suggested that their attorneys should meet with the hard-liners as well as the U.S. attorney, they thought I was beyond nuts.

  The model found that the optimal offer and settlement price was about a third of what the board thought they should make as their opening offer, and it found that this offer was best made in the presence of both the U.S. attorney and the hard-liners. The board members were sure the U.S. attorney would just get up and walk out if they did what the model was recommending. Fortunately, some of the other people involved in the case had worked with me, my partner, and the model before. They thought that the board should listen, and that they would listen if they just had an opportunity to confront the arguments head-on. With that in mind, the lead in-house attorney set out to arrange a meeting for me with a very senior executive from the firm. That senior representative would put the board’s case before me and see if I could make convincing arguments based on the model’s results—that is, could I persuade this executive that the strategy developed through the game was not nuts?

  Mind you, I have faced this sort of situation many times. Anyone who works on behalf of my firm on a project is instructed never—let me repeat that word, never—to argue for or against an approach to a problem except based on results that can be pointed to directly in the model’s output. We have no place for personal opinion. We are not experts on the substance of the problems we analyze, and generally we know little even about the industry involved, so there is no reason for anyone to take our personal opinions seriously. It is the very fact that we can show that the positions, tactics, and strategies we recommend come out of the model’s logic and the client’s data and not out of our heads that sells the client on the independence and the credibility of the view we express. They can argue with the model’s logic—that’s a conversation we love to have—and sometimes they do, but they always know that they will have to disagree with the logic or with the data inputs derived from our expert interviews (usually with their own experts), and not with us. Logic and evidence, not anyone’s personal opinion, are the focus of our presentations, briefings, and discussions and are, in the end, the basis on which people should decide whether to try an approach contrary to their initial intuition.

  No one should blindly follow a model. It is, after all, just a bundle of equations. But neither should people dismiss a model’s results out of hand just because its implications and their personal opinions differ. To repeat myself, the model’s greatest value is that it provides clients with a different way of thinking about their problems. That is an important part of the power that game theory, strategic thinking, brings to the table. Fortunately, as it turns out, clients generally find that the logic makes sense to them, and the data, after all, are theirs and can be adjusted and rerun to see how robust the findings are. So in the end, when they listen to us it is because they are sold on the integrity of the process.

  Of course, not all clients “check their intuition at the door.” When they don’t follow the model’s advice, the model’s prediction for what is likely to follow in that case tends also to be accurate. They end up with an outcome that is typically a lot less favorable. In the case at hand, the board’s senior management representative was sold on the model-based advice after examining me and the results for about eight grueling hours. The board agreed to go ahead with the approach we recommended.

  To communicate the message credibly, the firm’s general counsel agreed to meet with the U.S. attorney, and requested, even insisted, that the hardliners be present. Naturally, this insistence came as a surprise to the U.S. attorney. It was an even bigger surprise and a deep disappointment to the hard-liners. According to the model, they were eager to be backstabbers behind the scenes, sheltered from the light of day. They wanted to take a shot at influencing, maybe even cajoling or coercing, the U.S. attorney after he/she met privately with the firm’s general counsel. They tried to get out of attending a meeting with the general counsel, claiming scheduling conflicts. The firm’s general counsel deflated that maneuver by stipulating that he/she would meet whenever it was convenient for them. Their hopes for executing their gambit were crushed. By our surprise move, they lost that opportunity.

  The general counsel conveyed the board’s (sincerely held) conviction that they had done nothing that warranted criminal charges beyond misdemeanors. The general counsel went on to argue that the firm’s agreeing to what the U.S. attorney was demanding was tantamount to giving up the affected, important part of their business. The message was that the board would not even contemplate any sort of deal that involved pleading guilty to felonies.

  Of course, we knew that if push came to shove, the firm would have caved in and accepted one or more counts of the serious felony even if doing so would destroy an important part of their operations. They would have done so to bring the process to a speedy end so that they could get on with the rest of their business. They believed that even if they prevailed in court—recognizing that this is always uncertain even in the best of circumstances—the political, social, and economic costs of a prolonged trial would be unbearable. It was better to settle on a plea agreement and suffer the consequences. Remember Arthur Andersen (not the firm involved in this case). The company fought the charges against it resulting from its audits of Enron and was found guilty in court, only to have that judgment overturned by the Supreme Court after it was essentially no longer in business. Sometimes it is worse to win in court than to accept a plea agreement even when you are innocent. The judicial process may eventually come to the right answer, but too often the right answer only comes after unendurable costs have been borne by the defendant.

  The model’s maneuvers, in this as in many other situations, are designed to prevent push from coming to shove. The approach recommended through the model’s logic is indeed the very kind of bluffing we were talking about in our exploration of fundamental game theory back in Game Theory 101 and 102. We knew the board would plead to more than misdemeanors if left to its own devices, but the U.S. attorney did not and could not know that.

  After months of discussion, as anticipated, the U.S. attorney responded to the pressure from the board of directors and chose sides in the ongoing war. Instead of capitulating to the hard-liners within the government, the U.S. attorney berated them in the meeting we had insisted they attend. The U.S. attorney seized the chance to reinforce his/her own initial view of what constituted a just agreement. The hard-liners were made to look mean-spirited and unrealistic, even foolish in their demands. The case was settled by the defendant’s pleading guilty to several misdemeanors and one lesser felony. This was the outcome they desired and felt was just. They believed it was out of reach, and it would have been had they gone in asking for this agreement. Going in with the final result would have left them feeling c
ompelled to make more concessions. As it was, they got an outcome vastly better than the defendant’s management or attorneys (save one) thought possible.

  This case typifies the engineering process. The board of directors, the U.S. attorney, the Department of Justice hard-liners—they are, obviously, all different people, with different upbringings, personal experiences, and beliefs about the world. But they all make choices across the same dimensions of human behavior—there are, after all, only so many options to choose from. They can look for compromises; they can try to coerce people into capitulating to them; they can surrender to their adversaries; they can lock themselves into a war against their opponents; and they can bluff doing some of the above. That’s about the full array of choices of action for any problem. The key to this case was isolating the U.S. attorney; he/she was the principal driver of the outcome. What he/she thought was “right”—whether it was or was not in any absolute sense—and what he/she wanted in ego satisfaction: these were the two essentials behind engineering the settlement. The real question of this case was how those two interests could be aligned most favorably for my client.

  If I were to have brought the area-specialist mentality to this engagement, then perhaps I would have read through the thousands of pages of background produced by the armies of lawyers, perhaps I would have delved into volumes of case law, and perhaps I would have ultimately produced a brilliant argument as to why my clients deserved only misdemeanors and a minor felony. (Of course, this is not what my client hired me for, and is in fact exactly what the lawyers would have produced if the matter had landed in court.) But then, in that clash over the merits, however brilliant an argument I could have produced, despite all of my efforts, it probably wouldn’t have meant a darn thing. Because, of course, the talented government attorneys on the other side could have produced just as sparkling a case for severe felonies—and that leaves so very much to chance. No, the path to favorable resolution was in doing the work to produce an accurate understanding of the lay of the land, and then finding ways to work with it, not against it, through sequences of interaction.

  The process of predictioneering does not rely on the recounting of grievances (which all too often only hardens positions). If people are congregated around an impasse, well, it’s indeed unfortunate that there’s a conflict, but the very act of congregating around it suggests that the parties are in search of some dynamic that will yield an outcome—whatever it may be—to break the stasis. Predictioneering provides a complex network of decision-making channels, valves that open and close as actions are taken or passed upon. If I offer options A, B, and C, then doors D, E, and F open, and so on exponentially (again, hence the need for the trusty computer!). Suddenly, as this network routes decisions, grounds shift, positions change, and in this case a U.S. attorney is led to a place where he/she feels both validated in his/her own views and accomplished in forcing a little heavier penalty on my client than it was apparently (but not actually) initially willing to endure.

  This process involves exploiting or altering people’s perceptions of a situation by looking within the model’s round-by-round output to work out who is responsible for shifts in positions and how to counter those shifts if they have bad consequences for the client. The process is no different whether the problem is resolving Iran’s nuclear program, figuring out what al-Qaeda is likely to do, or facilitating the merger of companies. Every one of these situations involves humans who are not all that different from one another, regardless of where they go to sleep at night.

  So in the next chapter I’ll look at a few current problems such as those listed above to see how we might engineer beneficial outcomes. The examples will help illustrate the potential costs of failing to see or to address what may be around the corner.

  7

  FAST-FORWARD THE PRESENT

  ONE OF THE great benefits of being affiliated with Stanford’s Hoover Institution is the opportunity to participate in small seminars with some of the world’s most interesting scholars and policy makers. These seminars are often off the record, which means that there is the chance for frank exchanges of views on important issues of the day. The discussion during one such seminar, on the Israeli-Palestinian dispute, led me to consider how game-theory reasoning might contribute to tackling the seemingly insurmountable obstacles to peace. The approach I thought about is not a solution to the dispute, but it is a potentially useful step toward advancing the real prospect of a lasting peace.

  For all of its limitations, the idea I came up with provides an example of how game-theory reasoning can nudge us in a new direction even under the most seemingly intractable circumstances. If game-theory logic can foster progress on the Israeli-Palestinian dispute, it surely will have contributed to solving one of the most important foreign policy problems of our time. With that in mind, let’s have a fresh look at Israeli-Palestinian relations. And who knows, maybe somebody reading this book can help turn the idea into reality or can point out some fatal flaw in it.

  LET’S MAKE A DEAL

  Land for peace and peace for land are two formulas that are doomed to failure, whether in the Middle East or anywhere else. It’s an idea that sounds sensible, so it attracts lots of attention. Ehud Barak proposed a land-for-peace deal at the July 2000 Camp David summit between him (he was Israel’s prime minister), Yasser Arafat (then president of the Palestinian Authority), and President Bill Clinton. The Oslo Accords in 1993 also were a land-for-peace deal. Barak’s Camp David plan and its later variants failed. The Roadmap for Peace, another land-for-peace arrangement, failed too. All land-for-peace or peace-for-land deals by themselves will do the same. They are no way to end violence, because neither assures either side that the other is making a lasting promise, a credible commitment.

  Each promise—land for peace or peace for land—suffers from what in game theory is sometimes called a time inconsistency problem. That is, one party gives an irreversible benefit to the other party today in the hope that the other will reciprocate tomorrow. Almost certainly instead, the side getting the irreversible benefit exploits it to seek even more gains before delivering on its promises. Giving up land on the promise of peace inevitably leads to demands for more land before peace is granted. Giving peace on the promise of land later has much the same problem. The peace giver lays down its arms to show good faith, but then the land giver is free to renege, feeling no compulsion to follow through with land the opponent can no longer take.1

  Time inconsistency problems arise in many contexts, not just land for peace or peace for land. In fact, we saw an example of this problem earlier when we discussed North Korea. The threat of reneging on promises once an adversary has disarmed is exactly why asking Kim Jong Il to dismantle his nuclear capability won’t work, but negotiating a deal in which he agrees to disable his nuclear program can work. The problem is no less consequential in the Middle East.

  Look at the decision by Israel’s hard-liner prime minister Ariel Sharon to withdraw (unilaterally) from Gaza in August 2005, ceding that territory to the Palestinians. An important part of Sharon’s motivation seems to be that he concluded it was too costly to defend Gaza. So he chose to make Israeli settlers abandon their homes, whose existence was a major flashpoint with Palestinians, in the hope that yielding Gaza would help promote goodwill and peace. The belief that good deeds, whatever their motivation, will elicit a good response reflects optimism about human nature that sometimes is met by reality but all too often is met instead with greed and aggression. As you know, game theory rarely takes an optimistic view of human nature. Sharon’s optimism was predictably wrong. Shortly after the democratically elected government run by Hamas in the Palestinian Authority used force to oust Palestinian president Mahmoud Abbas’s Fatah from Gaza, Hamas increased missile attacks against Israeli towns near the Gaza border. Land, freely given with no strings attached, did not produce peace. It produced a demand for more land and an increase in violence.

  Mind you, this failure on behalf of th
e pursuit of peace is not some particular flaw among Palestinians. The Israelis have done much the same in the past. Having defeated their Arab rivals in the 1967 war and then again in 1973, Israel not only occupied previously controlled Egyptian, Syrian, Jordanian, and Palestinian land but also allowed the spread of settlements presumptively justified by the biblical covenant between Abraham and God. In fact, Israeli settlements almost always occupy the high ground surrounding Palestinian villages, making it all but impossible for Palestinians to enjoy a sense of security within their own homes. And even more troubling, Israel for decades restricted the movement of Palestinians into and out of Israel, just as they had done within Israel to Israeli citizens of Palestinian extraction. The upshot is that the Israeli government prevented Palestinians from following a peaceful road to independence by restricting their freedom of assembly. When Israel had the opportunity to promote peace with Palestinians after the 1967 war, it fell short, just as the Palestinians have fallen short in efforts to promote peace with the Israelis.

  Every land-for-peace and peace-for-land formula I know of has ended in failure. Every such effort, whether unilateral, bilateral, or multilateral, has, if anything, made the situation worse by raising false hopes only to see them dashed. They are always dashed because the peacemakers simply do not pay attention to the time inconsistency in their strategy. They rely on goodwill and building trust when there is neither. Instead, they should leverage progress toward peace on the narrow self-interest of the contending sides. They should consider whether what they propose is a self-enforcing strategy from which no one has an incentive to deviate, rather than looking for an approach that requires mutual cooperation. Remember the prisoner’s dilemma from Chapter 3? Both players in that game are better off if they coordinate with each other to pursue joint cooperation instead of ending up competing with one another. The problem is that each is even better off by not cooperating if the other player chooses to cooperate. The upshot is that neither cooperates, leaving both worse off than they could have been. That, in fact, is the dilemma. Like the Israeli-Palestinian dispute, joint cooperation is not a sustainable solution unless the structure of the game changes first.