There are several reasons for expecting that those in power of various states may not like the network.
But simply forbidding the network does not destroy it. One needs effective methods to suppress it. Such methods exist - the network would hardly survive if the state applies totalitarian methods against it: One can, for example, forbid any private ownership of computers or private connections between different computers, and impose ridiculous penalties up to a death penalty with torture to enforce them. Without much doubt, such a radical, totalitarian approach would allow to suppress the network.
One should not think that such radical measures will be impossible in modern, civilized democratic states. All one needs is a sufficiently radical hate campaign against the network participants, which makes them, in the eyes of the democratic majority, as subhuman as jews in Nazi Germany or pedophiles today. This is easy: Pedophiles will probably use the network to exchange pictures of naked children, so it is easy to associate network users with pedophiles. And it is sufficient: Irrational high penalties like life sentences for simple volitional blow jobs given to thirteen-years old boys, and long time imprisonments for simply owning forbidden pictures are reality in modern democracies today. So, why not life sentences for a network connection and twenty years for owning a laptop or a memory stick?
But these radical, totalitarian methods have side effects, and the more radical the methods, the greater the side effects. Without computers in private ownership, a state will loose a lot in economic competition with the rest of the world. It seems not an accident that the crash of communism has happened at a time when computers have become more and more important in everyday life - but only in the West. Moreover, the number of people who have used and liked computers and the internet is increasing. They may accept without resistance measures against uninteresting minorities like pedophiles (or users of the network). But if they would loose some own comfort - having a computer at home, with the ability to book various things in the internet from home, the situation may be quite different.
Let's, therefore, assume in the following that for the state it appears politically impossible to apply totalitarian methods, at least if they influence the everyday life of the majority. As a consequence, what remains is already a lot. Let's order these things in decreasing discomfort for the majority (thus, decreasing ability of democratic states to forbid them):
With all these assumptions fulfilled, it is technically possible for the network to survive: One can forbid the installation of the network software, but if it is installed on encrypted memory, it becomes impossible to prove that it has been installed. One can forbid the use of the network software. But if the network connections work only under the protection of an encrypted network protocol, one cannot prove the use of the forbidden software from outside the network. Thus, the network can work without leaving a possibility for police to detect it other than manipulation of the computer with keyloggers and similar spies. Once we have assumed that this is not allowed in case of simple participation in the network, simple participation would be sufficiently safe.
Once simple participation in the network and its use for legal or minor illegal activities is safe, and the advantages of use are strong, the network may be used by a sufficiently large number of people and can not only survive, but even increase and become more powerful.
Of course, between the purely totalitarian case, where the network does not survive, which is unrealistic because it hurts the democratic majority, and the non-totalitarian case, where the network can survive and grow, there is a large gray area, where the measures against the network hurt the democratic majority not very much or only in subtle ways, and where it is much less clear if the network can survive. In particular, mere participation in the network may be handled as an extremely serious crime, justifying all imaginable measures, which seems possible with sufficient demonization of the network users. Then, the use of encryption software may be possibly restricted without much resistance from the democratic majority.
There is an important defense strategy which works even if strong encryption is illegal.
The point of it is that non-totalitarian laws which forbid to own and use some particular software on the computer have to take into account the possibility of infection by a computer virus. As long as there is a possibility of infection, and the state is not totalitarian, or the police should be able to prove that the ownership and usage has not been caused by an infection, or the fine for violating the law cannot be large.
But to prove that the actual use was not caused by an infection may be hard, if the illegal software is implemented as a "nice virus", having all abilities of a virus if accidentally used by those who don't know about the virus, but behaving appropriately if used by those who know about it.
One needs some innocent basic program, "infects" it with the illegal software, and has now two possibilities to use it: If one applies the old, innocent use, it works like a program infected by a usual virus: It looks like working as usual for the user, but the virus sometimes copies itself, sometimes fakes some illegal use of the software. But there is also the second possibility: Those who know that the program is infected can call it in a slightly different way. Then it works like the forbidden program is supposed to work. Because the illegal activity of the program, faked or not, is encrypted (with hard encryption, even if it is illegal), the fake activitiy will be indistinguishable from real activity.
Now, police can detect the existence of the infected program, and also detect that it has worked some time, establishing, in particular, illegal network connections with other sites. But this is not sufficient to prove that the computer user has really used the network, instead of being an innocent victim of a virus infection.
Thus, as long as the state imprisons people only if their crime is "proven beyond reasonable doubt", to use the network software seems quite safe.
The network will be based on encryption technology. To use it may be, in principle, forbidden. But this will not prevent the network from using it. As a consequence, all what can be established without breaking the encryption is that some illegal software has been run on the computer. This is, as we have seen, not much. In particular, the poor infected user does not know any passwords to tell the police, even if the virus can create lots of encrypted files on the infected computer. Thus, laws requiring to give the police the keys to encrypted data do not work for the same reasons.
A counterstrategy would be to allow the police to manipulate the computer of everybody who shows illegal activity. That means, the police would have to invade the private life of everybody who has been really infected by the virus, thus, not only of those who have, by their behaviour, given some reason to suspect them of doing serious crimes, but of really innocent victims of a computer virus. This is already sufficiently close to a totalitarian method.
To put personal data into the net is one important way to improve the personal reliability in the network. But the verification of such personal data requires some contact which may be problematic in conditions of illegality.
There are two persons involved: The newbie who wants his personal data verified, and the notary who has to verify them. The problem is that the newbie has no reputation yet: He may be, as well, an undercover cop. The notary is, instead, a person with good reputation, and the newbie has chosen him out of his free decision. Thus, the security problem is one-sided: The notary has to care about the possibility that the newbie is an police spy.
Fortunately, it is not necessary for the newbie to have contact with the notary: All what is necessary is for the notary to get the information about the newbie.
A simple way to manage this is, for example, to ask the newbie to send lots of fotos by (network-internal pseudonymous) e-mail to the notary. Then, the notary can ask the newbie to do something: Go to some restaurant, sit down there some time, drink a cup of tea, and to go home. The verification - comparing his face with the fotos - can then be done by the notary in a completely undetectable way. All he has to do is to walk around the restaurant, or to have a camera somewhere which takes a foto of him. Even with support from lot's of other undercover policemen it would be impossible to identify the notary.
It is slightly more dangerous, but also possible, to obtain in similar ways fingerprints, iris scans or genetic data.
Another possibility is to use some state where the legal restrictions are less serious to do the verification.
Let's note the conceptual importance of these steps: It allows to verify that different persons identified in this way are really different persons. So, it becomes very hard for a single person to fake a large network.
The specific danger of undercover agents in a situation where the network itself is illegal is that previous successful cooperation does not decrease the dangers of future cooperation. Usually, the first cooperation with a newbie is risky, but once somebody has successfully finished a cooperation with him, future cooperation with him becomes less risky. In the case of an undercover agent, if one person has finished some cooperation successfully, there is no such increase in security: The undercover agent may tell about all his cooperations later, making the seemingly successful cooperation a failure.
This effect diminishes the advantages of the network, because the greatest advantage of the network this increases the security of successful cooperation with a stranger based on the information about his behaviour in the past. In the case of an undercover agent, the past behaviour does not tell the most important thing about the future: If he will tell all he knows about the whole network.
Fortunately this does not make the network entirely worthless: There are lot's of useful things which can be done in the network even if it is full of undercover agents. First of all, the world is full of small, informal local networks which work well. All these small local networks can become part of the global network without any additional risks, because becoming part of the network without establishing personal contacts is not risky. But the local networks gain advantages from participating even without such personal contacts: They can use the global blacklist. This increases the costs of violating the rules of the local networks for their participants, thus, adds security to these local networks. Then, they can use all the possibilities of interaction which do not require personal contact: File sharing, the banking system, uncensored discussion forums. Last but not least, some persons may establish personal contacts between different local networks.
This form of the network could be named a "localized global network". Undercover agents are a threat only if they appear inside the local networks. Nonetheless, the global network is much more powerful than the sum of the local networks, because of the various types of cooperation without personal real life contact.
Such a localization is an appropriate defense even in case of strong persecution of the network, that means, irrational high penalties for simple membership. In its final stage, the local network may as well reduce to a single person which refuses to establish any personal real life contacts with others. For such a single person, undercover cops are not dangerous. They see only what some pseudonym is doing - something which can be made open access anyway.
There is no way to distinguish an undercover policemen in the network until he uses the information he has obtained. But the larger the network, the more serious is the risk taken by the undercover agent: He will be discredited for his whole life in the whole world, because he appears on the black list with all his private information. The ability of police to give him a new identity is very restricted: It is beyond police power to change his genetic fingerprint. If the harm he has created has been serious enough, it may become even more dangerous to appear on the networks black list than on some interpol list of wanted criminals.
Given these dangers, it may become hard for the police to find people who agree to be undercover agents.
This effect works only if the network has had enough time to become sufficiently strong and powerful. Moreover, to rely on it in some way would be nonsensical: Some individuals can always be found, in particular among convicted criminals who would otherwise face long prison sentences. But will they play fair game with the police? A quite safe strategy for such undercover agents would be to report nothing of real interest, so that it does not make much sense for the police to make use of his information. As long as this does not happen, the undercover agent will not appear on the blacklist.
Another problem which appears because of the network is that one cannot make a job of being an undercover agent. In the past, an undercover agent could, after finishing his job in one organization, continue his job in another organization, with a different identity created by the police. But faking such an identity becomes much harder, close to impossible, if the network has the possibilities to use various types of biological data (fingerprints, genetric code, iris scans, biometric data). It seems beyond the ability of the police to fake the genetic fingerprint.
Thus, the undercover agent has to think about his future job after finishing his job as an undercover agent. This increases the problem created by his appearance on the black list.
If the network itself is legal, undercover agents do not impose any danger for the network. They define a problem only for a particular subgroup of people - those who want to use the network for serious illegal activities. As a consequence, the illegal activities in the network will tend to become localized, while for all legal activities one can take advantage of all possibilities of the network: One can, in particular, cooperate with strangers based only on the knowledge of their past promise-holding behaviour.
We have found that undercover agents are a serious problem in case of illegality of the whole network, because one of the main advantages given by the network - the information about his past promise-holding behaviour - does not give security. As a consequence, the network becomes less useful.
On the other hand, it remains useful: There remain advantages of the global network in comparison with the situation we have today - lot's of small local networks in real life. In particular, there are lot's of applications of the network not endangered by undercover agents, because they do not require personal contact: Uncensored forums, free exchange of files, an uncontrolled global banking system, and the global black list.
Once there remain lot's of advantages of using the network, the survival and increase of the network itself is not endangered by the problem of undercover cops.
Of course, all the private information about the members is encrypted. The current situation in encryption technology is that even extremely powerful state agencies seem to be unable to break the encryption. Evidence for this are cases where police has been unable to decrypt data of suspected criminals. And it is important that there is not only one such algorithm, but several different ones. To kill the network, one would have to be able to break them all.
Now, of course, one cannot exclude that one or another secret service finds a way to break one or another encryption algorithm. But this is a one-time weapon: This particular secret service is, in this case, able to read everything encoded with this particular algorithm. But once this becomes known, this ability becomes worthless, because nobody will encrypt anything worth to be encrypted with the compromised algorithm. As a consequence, it is quite obvious that they will not share their knowledge with the secret services of other states - they will prefer to use this one-time advantage alone. And they will prefer to use this one-time weapon for a really important case.
But in this case, it is quite probable that they will simply never use it. Most of the illegal activities in the network are consensual ones - those where the "victims" cooperate in the network with the "perpetrators" (drug market and prostitution), or where we have no "victims" at all, or where the victim is unable to find out that it is a victim (most copyright violations). To open the secret for persecuting these minor violations seems nonsensical.
Thus, the consequence of such a situation is that the secret service of some state knows more than one would expect, but this additional knowledge cannot be used to prove something in court, because this would reveal the source of that knowledge.
We ignore here, for the same reasons as before, the classical totalitarian possibility that one can be imprisoned based on information given by the secret service without any proof that this information is correct. But even in this case, the network would develop safely outside the particular state whose secret service has broken the encryption. Moreover, all those participants in this state who, by accident, use another, unbroken encryption algorithm, will be safe too. Once these are those who remain in freedom in the network, and newbies will do what the other guys in the net do, Darwinian evolution will automatically decrease the use of the broken encryption algorithm, even if nobody learns about it.
Thus, if only one (or a few) secret services can break only one (or a few) of the encryption algorithms, the network will survive this. The probability that almost all secret services can break almost all encryption algorithms seems purely theoretical: In this case, from time to time some public domain mathematician would find and publish such a loophole too. Moreover, in this case there would be much less incentive for a particular secret service to keep the fact that they can decrypt a particular code hidden: If others can decrypt this code too, it is useful to apply it first.
The most probable scenario is the safest one: Some encryption algorithms simply are safe, cannot be broken, and in this case no amount of money allows any secret service of the world to break it.
The only real danger is a psychological one: Many people will be afraid that the secret services can break any encryption, and, therefore, will be afraid of participating in the network. Such a fear may slow down the grow of the network. But not for a long time: Not everybody is afraid, those who are afraid have friends who are not, and once nothing happens with these friends, those afraid will become less afraid with time and participate too.