SuperCooperators. Roger Highfield
her survived. In a strange way, Karl and I felt bereft after reading these moving tributes to her lost beauty. My melancholy was a faint reflection of Caroline’s radiant glow that had illuminated Vienna long ago, a testament to her reputation.
FROM REPUTATION TO COOPERATION
The most incomprehensible thing about the universe is that it is comprehensible.
—Albert Einstein
By the time my flash of inspiration came in the Wienerwald, my confidence in my ability to crack problems was growing. Deep in my brain a geyser bubbled and sent a torrent of thoughts skyward. I knew I had to work fast. Nearby was my parents’ house, on the northern slope of the Kahlenberg. In my little bedroom, which I have used since I was eight years old, I sat down and began my research on indirect reciprocity.
Usually when you begin a new project you immediately run into difficulties. There’s one unforeseen problem, then another. Often there are many. You need time to wrestle with them and, only if you are very lucky, resolve them. Usually you fail. Not this time, however. Everything I attempted worked, and first time too. After three weeks, I had an almost complete story, a mathematical picture of indirect reciprocity and, most important, how it helps cooperation to bloom. I was driven on by the excitement of trying and succeeding at something new. I was proud of the lightning speed with which I managed to knit my intuitions into a mathematical theory.
After three weeks I saw Karl again to discuss my findings. Once again, we met in the forest. This time, the weather was grey, the air damp and raw. We had arranged to meet in a little inn and, once we had sat down together at a wooden table, I presented my results to him. Even though we were friends, I felt nervous as though I were revealing a secret for the first time. Karl liked the approach and immediately saw the implications.
I began with a computer model that described a population of people. In that population, any one encounter involves two people. One of them is offered a choice—whether or not to help the other. When a Good Samaritan does something nice for someone else, this altruistic act confers a benefit on the recipient at a cost to the Samaritan. That could be when you have to sacrifice your time to help another, whether giving a hand to that doddery old lady to help get her safely across a street or taking a moment to point out the nearest car park to a motorist.
If the cost is smaller than the benefit then the act of charity, once returned, leaves both individuals better off. This puts us in very familiar territory. One can think of this setup as a simplified version of the Prisoner’s Dilemma that we explored in chapter 0. Cooperation means paying a cost for the other person to receive a benefit. Defection means doing nothing. If you think of one person as the donor and the other as the recipient, then it adds up to half of that problem, a demi Dilemma.
As we saw in the case of the Prisoner’s Dilemma, it is rational to defect. But only in a single encounter. If our players see each other again and again, cooperation can emerge because rational players must weigh the benefit of exploiting the other player in the first round against the cost of forfeiting collaboration in future rounds. But, of course, repeated encounters between the same two players leads to direct reciprocity. I now wanted to study the evolution of cooperation in a more general, indirect setting.
I arranged the game so that each player can take part in many rounds, but typically not with the same partner twice. Thus a defector—someone who does nothing to help—could not be held to account by an earlier victim. But defection can be detected nonetheless, as a result of players building up a reputation: a player’s reputation score (Karl and I called it “image” in our paper) is 0 at birth and rises whenever that player helps others. Equally, it falls whenever the player withholds help. This is an important ingredient of the game. It meant that we did not divide our players up into goodies or baddies. Instead, we graded the image of each player so it could be more nuanced and change as the game evolved.
There were also unconditional cooperators and unbending defectors. I built one more feature into the model, to add a touch more realism. Just as only a certain group of people are privy to gossip, so the outcome of any given encounter between players is only revealed to a subset of people in the population. As a result of this, different people hold different views about the reputation of the same person.
Karl and I found that if the cost-to-benefit ratio of cooperation is sufficiently low, and the amount of information about the co-player’s past sufficiently high, cooperation based on discrimination—favoring good reputations—can emerge. Instead of relying exclusively on my direct experience with someone (as is the case for direct reciprocity), I can now also benefit from the experience of others. Now my behavior toward you not only depends on what you have done to me, but also on what you have done to others.
The bottom line of these evolving populations was the following: if there is enough transfer of information about who did what to whom from person to person, then natural selection favors strategies that base their decision to cooperate (or defect) on the reputation of the recipient. If good reputations spread quickly enough, they can increase the chances of cooperation taking hold in a society. And, as one would expect, Bad Samaritans with a poor reputation receive less help.
We were not the first to argue that reputation is possibly an important factor for altruistic behavior. A form of the concept had been articulated—in a verbal outline, rather than mathematically—in a book by Richard Alexander at the University of Michigan, an expert on crickets, katydids, and cicadas. It was Alexander who, in The Biology of Moral Systems (1987), had first coined the term “indirect reciprocity.” Alexander had posed tough questions such as, What is moral? or How do we start to crystallize our ideas about what is good and bad? He argued that the answer lay in reputation. We are always scanning the impressions left by others and are more likely to give to somebody who has a good reputation, someone who has in her or his past given help to others—not necessarily to me, though, but just to somebody. Indirect reciprocity “involves reputation and status, and results in everyone in the group continually being assessed and reassessed.” This, he said, plays an essential role in human societies.
The idea appears in the work of economist and philosopher Robert Sugden of the University of East Anglia. He put forward the concept of “standing” in The Economics of Rights, Co-Operation, and Welfare (1986). The idea goes like this: If you defect against somebody who is in good standing, you move into bad standing. But if you defect against somebody who is in bad standing, you remain in good standing. A mathematical depiction of social norms had also been explored by the Japanese economist Michihiro Kandori. The fact that our new theory of indirect reciprocity had reputable forerunners gave it further credibility.
Karl also had plenty of anecdotes to underline why indirect reciprocity is just as relevant to everyday life as the direct version of reciprocity. He pointed out that the Rothschild family had protected the investments of their English clients during the Napoleonic Wars. They were under intense pressure to give them up, but they kept the interests of their English clients at heart. Afterward, of course, the Rothschild family became extraordinarily rich. Their fortune was a direct result of the power of indirect reciprocity: because they had behaved impeccably, everybody now knew that they could be trusted.
Then there was the story of the American baseball player Yogi Berra, who was famous for his pithy comments and witticisms known as Yogi-isms. One of them was a perfect summary of indirect reciprocity: “Always go to other people’s funerals, otherwise they won’t come to yours.” Berra was counting on the fact that his acts of kindness would not be returned by the recipients, but by third parties who were moved by his public mourning.
The idea is also wonderfully summarized in musical form by Tom Lehrer, the American singer-songwriter, satirist, pianist, and mathematician. In “Be Prepared,” Lehrer’s spiky salute to the Boy Scouts, he sings: “Be careful not to do / Your good deeds when there’s no one watching you.” German speakers even have a saying with the same gist, Milinski notes: Tue Gutes und rede darüber. (“Do good and talk about it.”) The converse was of course true too. This may all sound obvious. But without a mathematical model, we would have no quantitative understanding of how this mechanism really works.