I’m sure most of you are familiar with the Prisoner’s Dilemma. Briefly, two agents have the choice of either cooperating with the other or cheating him. If both cooperate, they win a certain amount. If one cooperates but the other cheats, the cheater gets a lot and the other gets nothing. If both cheat on each other, they both get very little.
Here’s an interesting fact you may not know. There have been programming competitions to determine what algorithm best deals with the Dilemma. What they do is, they pit the algorithms against each other, in one-on-one games, until they’ve all played each other. The winner is the one that ends up with the most units at the end of all the games.
In these competitions, there’s one simple algorithm that always wins, despite its very simplicity: it’s called tit-for-tat. It can be described simply as “always cooperate unless the other cheats, then retaliate, but be quick to forgive.”
Why does this simple strategy, that takes only a few lines of code, beat all others? Because it tries to establish cooperation, and mutual cooperation is more rewarding than mutual cheating (in accordance with the gain values of the Prisoner’s Dilemma). In more commonplace terms, we can express the tit-for-tat attitude like this: “start by trusting everyone, if they try to mess with you then stop them any way you can, but keep the possibility of trust open.”
The element I haven’t mentioned here is that the algorithms can also store plays of the past and learn from them, if they are programmed to do so, like human beings. If you combine this tit-for-tat strategy with the possibility of learning, it seems inevitably that cooperation will spread and that conflict will lessen over time, because cooperation rewards the individual agents more than conflict.
Birgir T. Runolfsson Solvason made a little study that is very much related to this matter, in which he discusses the theoretical growth of cooperation as well as its constraints, and applies his theory to the real-life case of early Iceland, which was a stateless society. What he found was that, while there is a limit in the number of people that can coexist in some types of organizations, there is no limit to the potential cooperation that can exist in any given society, no matter the size.
How do we interpret these results? I think the framework of love and control that I’ve been discussing provides us with a valid interpretation. To “cheat” in interpersonal or social relations means to control others, be it by war, physical violence, by stealing or defrauding, by asserting authority and diverting their energies for your own ends, by manipulating them, and so on. All forms of control can be expressed as trying to gain more than you would by “cooperating” by hurting the other fellow. To “cooperate” means to produce and trade freely, to aid each other, to act in accordance with the law of love.
Note that assembling in a big group (such as our geo-political entities) or in a hierarchy (such as a government or corporation) does not equal “cooperation.” People even define the existence of such groups as being a form of sophisticated cooperation, that the bigger and more complex groups in a society are, the more sophisticated the society.
But groups are the cheater’s way to social cohesion, because they establish purpose by controlling people instead of doing so by dialogue. The bigger a group or organization is, the more bureaucracy it has, the bigger the pool of money that can be manipulated, the more incentive people have to control each other, and so on. In a hobby group, this will turn out pretty benign: you can’t really implant people with identities based on their love for skeet shooting or magic tricks. But we know how badly this turns out in other areas.
So if this tit-for-tat theory is correct in substance, at least as far as it goes, then why isn’t it uniformly demonstrated on this planet?
There is one vital factor that is not included in the Prisoner’s Dilemma games: inequality. The fact that some people are stronger or richer than others makes it possible for them to reap more rewards by using control than by treating others as their equal. Just think how much more a government gets by taxing people than it would by donations and pay-for-services (assuming, of course, that alternatives were available and not just made illegal like today).
Looking at the history of the State, think how much more you’d get by taking over an area and imposing a tribute. So there is an incentive for the stronger to use violence in order to reap greater rewards, defeating tit-for-tat. Theft exists and flourishes (whether in individual or bureaucratic forms) because the gains expected through control are in fact higher. It’s to the advantage of the powerful to exploit rather than cooperate.
This problem did not arise in Iceland for a long time because the people there started off as roughly equal. The existence of cooperation also has a general effect of fending off inequality, as it is harder to pick on one weaker fellow if he’s allied with a hundred more. But the degeneration of cooperation into groupthink can create hierarchies, which generate inequalities of power and overturn the whole system.
So how can tit-for-tat, ergo equality, be maintained? By using “sustainable” processes, which will not degenerate into tools of control. This means total decentralization, small communities instead of expanding communities, and the wide dispersion of all powers of aggression. The first two especially seem to be recurring points on every concept that I have written about lately: individual-centric decision-making and maintaining extremely small groups (in the low hundreds or less) seems to be absolutely necessary in order to have any lasting freedom.