An intuitive way to think of social control is by taking examples from our every day world. We have a opinions about numerous things, opinions which help us know how to act in different situations. Often these opinions are formed through social mechanisms.
Ex: People who live in small villages sometimes complain about feeling watched by their neighbours. They feel that if they don't behave according to the social norm, they will ridiculed and talked about behind their back. Still, they don't want to move to a big city since there are a lot of crazy people there, and they find it awful that people can live for years door to door to a complete stranger.
Some effects can be good. In a small village, life is governed by the social control of the local community. This is why people can leave their bike unlocked, being sure no-one steals it. In a big city the size of the crowd reduces the social control. If a person buys a bad car, she can't prevent other people from going to the car dealer, and in a big enough city a dishonest dealer can can continue his business regardless of the clients opinion about him. In a small town he would soon loose his clientele if he tried that.
Some effects can also be bad. The social pressure to conform can hinder people who want to do different things. An artist or a writer might be considered weird and odd by the locals. A person who once has been excluded from the society (perhaps after having been hospitalised for some mental illness) might never be accepted again. In a big city it is possible to start over, make new friends who base their feelings for the person on his current behaviour.
Ex: Another social mechanism is that different groups develop different tastes and preferences. Music is a good example of this. There is no objective scale that can measure the quality of all music. Instead, when we want to buy a new record, we use the advice of people who have approximately the same taste as us (friends, music reviewers etc.). Classical music and hop hop can coexist, each of them developing its own quality scale. In fact, often many competing scales develop within one genre, which is seen in the plethora of reviewing magazines that exist for (for example) pop music.
In summary, social control is a group behaviour that indirectly forces the group members to behave in a particular way.
Internet is as an open system more like ``a big city'' than ``a small village'', in that it is possible to act in any possible way without anyone being able to stop it. If this is permitted in electronic markets there will be a large amount of fraud and con men doing business there. In a big city you can't know who you are dealing with if you meet for the first time.
How is it considered possible to negotiate, cooperate and perform commerce in an open electronic system if there is now way of formally knowing the intentions of the other actors? By looking at how the market does commerce today, we find that there exists many informal mechanism and ideas that are absolutely necessary for commerce to work. There are notions such as reputation, a stability over time, and estimated risk of being conned, all of which help one company to choose with whom to do business.
But first, let's introduce a game of two prisoners. This will cast new light on some issues in the security of large electronic markets.
The prisoner's dilemma (name coined by A. Tucker and made famous by R. Axelrod in [1]) is a non-cooperative game with two prisoners. The prisoner's dilemma shows that if chances that the prisoners will ever meet again are low, then the best strategy is to try to cheat on the other prisoner. But, interestingly, if the game is iterated, then the best choice of action is to ``cooperate while the other prisoner cooperates.'' The strategy tit-for-tat is found to be an evolutionary stable strategy, ESS. Such a strategy is the best strategy for a player provided the other players play by the same strategy.
Figure 2: Should Row and Column confess or deny to minimize their
prison time? Since they gain most if they confess, independent
of what the other does they will confess, even though they only
get one year if both deny.
The prisoner's dilemma can tell us a lot about automatic electronic markets. A transaction between two parties is most certainly a one shot deal. Probability is low that the two will do business again. This means that the agents find themselves in a game similar to that of the prisoner's dilemma, and that the best game theoretic move is to try to cheat the other agent. What's missing is a means to bring the system back to an iterated prisoner's dilemma-like game.
If the actors on an electronic market are dependent on their reputation to act, their earlier actions are important for how other agents will treat them, and this is exactly what makes the iterated prisoner's dilemma support cooperation. A form of social control, or put in other words, a distributed locally implemented mechanism, can help to achieve cooperation in an open system which would otherwise be prone to extremely malicious agents.
Social control is a way for the population of actors to avoid unwanted actors. Let's consider some cooperation mechanisms used by non-cooperative agents. That is, mechanisms that promote cooperation even though they are free not to cooperate if one agent find that to be the best action.
The Clarke Tax (as described in [3]) is a one-shot voting-by-bid mechanism for group decision making that eliminates the advantage of trying to manipulate the result. There is no risk that a rational agent chooses not to cooperate (manipulate) since this doesn't maximise the expected utility. Reducing agent communication before the vote, as in Clark Tax, can therefore lead to systems in which cooperation is the most efficient action.
But there must be some way to assure that the agents act rational and that non-rational agents are removed. In computational economies it is common to introduce a fictitious currency to do elimination of bad alternatives. If agents are paid according to their rationality and have to pay a small fee to stay on the market, then non-rational agents will run out of money and will be removed from the market. Without this cost of staying on the market there will be room for all sorts of actors on the markets, some of them non-cooperating, malicious, some of them just wasting system resources.
If the system is open there is no way of making sure that there is no prior communication between the agents before the Clarke Tax voting is made. If some of the agents form a collaborating subgroup (conspiracy) they can undermine the fairness of the voting. Although very important, these mechanisms alone might not be sufficient to assure a desired behaviour of an open system.
Actors on a market must be able to refuse to cooperate with others that are found to be disobeying the rules of the game. There are two reasons why someone might be doing this. The first one is that they are trying to fraud the other party. The second is that they could be acting ``non-rationally'' because they have incomplete information, because of a mistake by the programmer (a bug) or some fault in the system. It is not safe to trust all security to the game theoretic mechanisms and assume that all actors are completely rational. They must be completed with a way to dismiss non-rational (possibly malicious) agents. And a way to do this without introducing global authority mechanism is through social control among the actors.
All this raises a number of questions. How can it be possible to introduce new, reputationless, agents into the system? Can an agent build up a good reputation and use this to commit crimes? What happens if someone wrongly gets a bad reputation and is hindered from doing business? Is any of this even legal? We will try to address some of this below, but first we'll look at what mechanisms of social control are there that make open agent systems not promoting cheating agents?
Different behaviours of the participants will give rise to different emergent effects, and this can be used to advise implementors of such systems how they can go about to avoid creating systems that are prone to explosions of black markets and other criminal behaviour in the actors
___________________________________________________
Lars Rasmusson