The system of rules is quite simple: Everybody defines his own rules.
This sounds nonsensical: Why will somebody restrict himself by some set of rules? The reason is that we all have to cooperate with others if we want to survive in this world. But cooperation is not possible without some sort of contracts between those who cooperate. Thus, to be able to cooperate with others, you have to make contracts with them, accept some restrictions of your own free will.
A rule in the network is not simply some text which describes some restrictions of your behaviour. A rule also contains some other parts:
The rules, therefore, contain everything they need to be enforcable.
But if everybody is in the network, and has his rules written in his own words, all these data would be almost unusable. Even the best information technology would be unable to handle them to make them useful.
If there is only a small number of people around you, you can spend some time to make different contracts with everybody. If there are lots of different people, you would like to have some standard contracts - some offers to everybody else in the world: If you accept condition A in relation to me, I accept condition B in relation to you. Most often, such an offer will be completely symmetric, so that A equals B - in agreement with the Golden Rule.
As a consequence, it is quite useful to accept rules of this type: There will be, automatically, lot's of people who accept some rules relative to you. But this happens only if these other people accept such a rule with the same A. Because of this, it may be practical to develop some recommended standards.
These standards we will call conventions. They also have other advantages: To store the fact that you have accepted some convention takes much less storage - all one has to store is some link, instead of the full text. Moreover, conventions are much easier to handle by programs. One would needs true artificial intelligence to let a program evaluate explicit texts. But to check if a potential customer has accepted a certain list of conventions is easy. So, while conventions restrict our freedom of formulating our own rules, reducing this freedom to simple yes-no answers for a finite list of conventions, the use of conventions has certain advantages. And the restrictions are not really important - they are, essentially, only about some fine print. To fix the really important yes-no questions, conventions will be sufficient.
Another, more flexible way to define rules we will name penal codes. A penal code consist of a list of paragraphs, where each paragraph contains some number, which defines the penalty for violating it. Now, instead of simply accepting or rejecting it, everybody can change the numbers.
The main advantages of conventions remain: To store a penal code, we do not need much memory: A link to the code in general, and a list of numbers. And one does not need true artificial intelligence to handle such a penal code with programs: The meaning is fixed, what changes are only numbers, and numbers are something programs can easily work with.
For example, you may want to be sure that your potential client accepts penalties for things you don't like to do him which are greater than some minimal penalities. This list of minimal penalties you can define once reading the code. After this, all the program has to do is to compare numbers of the penal codes of your potential clients and your minimal list.
The freedom of choice of the ideal set of rules is already much less restricted: You have not only a simple yes-no choice, but can differentiate, modify the numbers in a continuous way.
For a rule to be completely defined, it should be clear what happens if the rule will be broken. One possibility is that one defines explicitly the penalty.
Again, this sounds meaningless, because it doesn't really answer the question. The problem remains on the next stage: What happens if I don't pay the fine I have promised to pay?
The solution is a default penalty: Publication of the fact of the violation of the rule in some black list, which is visible to everybody.
This penalty is quite hard: Now everybody knows that one cannot rely on your promises. It becomes much much harder for you to find somebody who agrees to cooperate with you. It will not be impossible: But the conditions for future cooperations will be much worse: You will have to do almost everything in advance, to reduce the risk for your partner. Or you will receive less money for your job. Or you have to cooperate on equal terms with other rulebreakers.
But who decides that you have broken your rules? This is some arbiter which you have to accept together with your promise. This acceptance means that you sign some statement that he has the right to decide if you have broken the rule. This signed acceptance gives him the right to put the decision against you into the black list.