POLE-Papers
POLE PAPER SERIES ISSN 1370-4508 Vol. 2, No. 5, December 1996

Anarchy and Cooperation within the Internet

by Gustaaf Geeraerts en Anthony Antoine

  1. Introduction
  2. The Internet is really the topic of the day. Many lively debates are held and numerous journalistic articles are written by advocates and critics alike in attempts to persuade the public of the positive or negative implications of this very system of communication. Lacking, however, are more profound studies gauging the underlying mechanisms and deeper structural characteristics driving the system. In essence, the Internet is a highly decentralized construction through which millions of users communicate with each other. As a central authority is absent within the system and therefore nobody really controls it, the Internet can be considered anarchical. Major questions, then, are: How is order maintained in a similar system? Is it at all possible to regulate an anarchic system with so many users as the Internet?

    This paper explores some answers to these fundamental questions applying to the Internet the concept of anarchy and related theories dealing with the possibility of cooperation among states. The main purpose is to reveal the structural similarities and differences between the international system and the Internet. Important is where this will lead us as far as the problems of order and collective action are concerned.

  3. Anarchy and the problem of order
  4. Analysing the Internet one can discern at least three levels, i.e. the Internet as a network of networks, the autonomous networks themselves, and the users constituting them. In this paper we will focus on the user level as it is central to any deeper understanding of the network. Without users, no interactions take place, and the Net turns out to be nothing more than some pieces of hardware and infrastructure. The user, and in principle he alone, decides what he will read, when he will access the information, and when he will disseminate or send documents. The only rules he needs to comply with are those issued by the network he uses, but he is always free in his choice of the latter. In theory, his freedom is absolute, and virtually nothing can withhold him from being part of several networks. Nevertheless, he is not sovereign in all respects. Access to the Internet has an unmistakeably economic dimension: one has to join a network and thus pay for it. The user level stretches out to more profound questions in international relations, i.e. that of maintaining order in an anarchic, that is a decentralised, environment populated by autonomous or "sovereign" actors. Indeed, at least at first sight, the Internet appears to form a world comparable to the state system, albeit one with far more players. The sheer possibility of such a parallel, inspired us to investigate both the Internet's structure and the interaction patterns among its users from the perspective of leading theories about the implications of anarchy on state behaviour.

    1. Anarchy in International Relations

      The idea of international anarchy most commonly used in international relations is a Hobbesian one. It refers to the lack of a higher authority above the state. States are the only sovereign actors in international society. They are judges in their own causes. This means no prospect of completely enforceable international law, or of a universal moral code to guide the behaviour of states. The use of force may not always be right, but unless it is dealt with adequately it may prevail. States cannot escape from this anarchical setting, and their attempts to avoid some of its worst consequences account for the recurring patterns of arms racing and alliances. Anarchy can only be mitigated, not transcended. Since the structure of the system remains anarchical, states cannot but take care of their own security. Force, or the threat of its use, will stay a possible outcome of any international interaction.1

      Given international anarchy with its lack of trust, states face a security dilemma. The more a state arms to protect itself from other states, the more threatened these other states become and the more they are likely to start arming themselves to protect their own national security interests. The dilemma consists of these states' worst-case assumptions about the adversary's intentions leading to the sole rational response in a self-help system, namely engaging in a similar arms buildup. Even if the adversary state is sincerely arming only for defensive purposes, one can simply never be sure of the real intent. States cannot escape from this anarchical setting, and their attempts to avoid some of its worst consequences cause recurrent patterns of arms racing, alliances and war.

    2. Internet's Anarchic Structure

      What, then, do the Internet and the international system have in common? Internet's anarchic structure lies in its roots. Originally designed by the US military as ARPANET, the network of the late 1960s needed to ensure communication under the worst of conditions, namely nuclear attack. Whereas the military normally prefer centralized structures, they now opted for an alternative and effective way of communication through decentralization. Much the same as the street-network where drivers can use alternative routes when the main road is blocked, "traffic" on the Internet can use various "routes" to reach its destination. When part of the network is down, the rest can still function, which would not be the case with a centrally governed system. Disconnecting the main workstation would indeed completely cut off the whole system.

      Because of its decentralised design, the Internet is not governed by an overarching body. It consists of various smaller and autonomously operated networks that have no ruling board or hierarchical superior. As such, Internet is a "network of networks."

      It is tempting to draw a parallel between the autonomous networks in Internet and the states in the international community. However, the inter-connectivity transcends this parallel, as the constituting networks are only the physical frameworks for the interaction of individual users. Users of the Internet are not bound to the rules of the network on which they are working. In fact, they only use the latter as an intermediate for getting access to the more global Internet. As such, the "autonomous" network has no function comparable to that of the state. The virtual world of the Internet reaches further than the physical decentralisation of its infrastructure. After all, the technological infrastructure represents nothing more than the material environment in which the anarchical interaction develops. The Internet resembles a huge "market," with millions of users (survey of July 1995 shows more than 60 million users), thus making the user rather than the networks the primary object for analysis. When comparing the relations and interactions on Internet to the international state system, we need to allow for this.

      Moreover, although physically all servers are located in a specific country, this is not important for the overall functioning of the Internet. This "virtual world" has no nationality. It is anarchic, and with the lack of a global regulation of the virtual Internet world, top-down created rules can only be implemented post-factum. On the national level, this can result in policies affecting the local autonomous network or the individual user. Quite understandably, governments strive to control the Internet for more than one reason. For economic reasons rather than on moral grounds, they are eager to get a grip on the commercial transactions occurring on the Net. The censorship on pornographic material can then be seen as a test-case of gaining control over the Internet. The Bavarian authorities' decision to ban pornography on the Net - that is, a ban on pornographic material placed on servers within the boundaries of the German Land - is a case in point. Such actions, however, fail to consider the nature of the Net. First, no such thing as an Internet "territory" exists. Second - and more importantly - this regional regulation does not affect the global Internet. Pornography offered from Honolulu can still be viewed in Germany, or anywhere else in the world for that matter. It only affects the potential German distributors of the envisaged material, not its readers. Because of the global nature of the Internet, Germans who wish to do so, can put all material of their choice on a different - foreign - server, thus bypassing the German regulation. This was clearly seen in a second example: the French censorship on a book on President Mitterand. When the hard copy was forbidden in France, it was offered on-line in the States, making it available to download for millions of people all over the world (including France). The US Government's recently issued Communication Decency Act, imposing stiff penalties for posting "indecent" material on-line, is another example of failing national regulation on the Internet.2 Though its impact still has to be investigated, one can already predict that the outcome will not be as expected.

      The above actions should not surprise the international relations scholar, as he or she will know that it does not make too much sense to set up hierarchic (i.e. state-) measures in an anarchic system (here: the Internet). As Waltz pointed out, the ordering principles of hierarchy and anarchy are different. Whereas the first refers to a political system in which the units stand in a relationship to one another that is institutionally and legally organized as a hierarchy of power, the second applies to a system in which no such formal power relations are at work.3

      Another resemblance with the international system is that, like states, Internet users can be considered as "like-units", that is functionally undifferentiated. Clearly, Internet users have one thing in common: they are all "virtually" equal. However, this does not mean that they are equal in any strict sense. They are equal in the sense of sharing the same power instrument: communication. This characteristic shows in the use of various "services"4 on the Internet.

      A further point of comparison concerns the security dilemma. For instance, there are many instances of threatening and retaliatory behaviour on Internet. The 'Green Card' incident offers a case in point. The reaction of the Internet community to the dissident behaviour of two lawyers as they advertised on news groups that were not meant for that was overwhelming: from all over the world, people started to mail them "flames," messages in which they uttered their despise and malcontent, or in which they even threatened with more violent action. In the end, the lawyers were even cut off the Net.

      Still, other ways of making life difficult for aggressive Internet users exist. A person can block a fax by sending constantly messages to it; he can also block a user on the Internet by sending him tons and tons of mail. This will not only disable him from working on E-mail. It may even lead to a system crash, thus undermining his use of other Internet services (i.e. World Wide Web, Gopher, ...). In this way, large-scale flaming is comparable to a total boycott. However, whereas an economical boycott aims to seal off a country by cutting down interaction, large-scale flaming relies on vastly increased interaction to isolate dissident users. The outcome, however, is much the same: the dissident becomes isolated.

      On closer examination, however, one is bound to conclude that security threats among users of the Internet are also different. Whereas states can possibly affect the security of their counterparts, users on the Internet cannot influence their fellow-users in the same way. At best, they can make life very hard on their colleagues, or expel them from the Net. However, the physical condition of the latter is never under direct attack. Moreover, whereas the use of power and force can result in great damage (e.g. economic sanctions leading to lower production and affecting national income, physical threats and casualties by using military force, ...), the Internet user can only be affected "virtually". He can simply start "another" (virtual) life whenever he wants to do so. Having several "accounts" on different networks is common for individuals. If expelled from one server, another is readily available.

  5. Anarchy and the problem of cooperation
  6. So similarities between the Internet's composition and the anarchical structure of the international system are in evidence. However, some differences also exist. Which, then, are the implications of these findings for the possibility of cooperation and coordination within the Internet environment?

    An anarchical, self-help system obviously makes cooperation among the units in the system a difficult proposition. Yet, although states interact in an anarchic world, cooperation remains a possibility. They may, for example, agree to limit their arms acquisition or cooperate by establishing organizations. Under conditions of anarchy violence and war may be evident but so are periods of relative peace and stability. Hierarchic elements within international structures limit and restrain the exercise of sovereignty but only in ways strongly conditioned by the anarchy of the larger system. The anarchy of that order strongly affects the likelihood of cooperation or collective action.5 In essence, then, the main question is to find out under which conditions states can cooperate.

    1. Neorealism, neoliberalism and cooperation among states

      In international relations literature two different views exist regarding this problem. Neoliberalism approaches the problem of cooperation in terms of absolute gains. Each actor strives to maximize its own profit, and cares little about how much the others gain or lose in the deal. Put differently, each actor only considers the others to the extent that their behaviour affects its own prospects for achieving maximum profits.

      For neorealists, by contrast, states only seek to maximize their absolute gains with respect to their own survival. Otherwise, they assume that states in an anarchic world are concerned about the distribution of power and consequently seek relative gains. In other words, they are also concerned about how well they perform in comparison to others. This is particularly true for cooperation. Yet, cooperation is more difficult to achieve when states are attuned to relative-gains rather than absolute-gains logic. Indeed, states maximizing absolute gains need only to make sure that the pie is expanding and that they are getting at least some portion of the benefit. States interested in relative gains, by contrast, must also be concerned about the way the pie is divided, which complicates cooperative efforts.6

      Neoliberals contend that these pessimistic assumptions are unfounded because neorealists underestimate the influence of international institutions.7 To the neorealist argument that international institutions affect the behaviour of states only marginally because they are not an independent force, neoliberals reply that international institutions do not just reflect temporary interests of states but also shape those interests and practices. In their view, institutions can continue to promote international cooperation even when the state interests which led to their creation no longer exist.

      Despite these conflicting predictions, neorealism and neoliberalism share some fundamental assumptions. Both schools accept that regularities of international behaviour are best explained by the nature of the international system. Both theories start from the premise that states are the major players in world affairs and rational unitary actors trying to maximize their national interests. Finally, as no central authority can enforce compliance with international agreements, both schools agree that anarchy is an underlying force of the international system.

      Due to these shared assumptions, neoliberals and neorealists agree that international cooperation is difficult even if states have common interests. Anarchy allows states to defect from international agreements. Neoliberals claim, however, that states can solve this problem by establishing international institutions that diminish the incentives to cheat and make the compliance option more attractive. According to neoliberals, institutions do so by distributing information, by reducing the costs of monitoring individual compliance, and by making it more cost-effective for states to punish noncompliance. Accordingly, neoliberals argue that international institutions carry a potential for overcoming basic obstacles to international cooperation.8

      Neorealists argue that neoliberals overlook important impediments to cooperation since they fail to grasp the full implications of international anarchy. Two factors increase the risks of cooperation among states. States worry not only about cheating by others, but also about the possibility that their partners may gain more from cooperation than they themselves. In other words, states care not only about absolute but also about relative gains. From a realist point of view, states are not rational egoists with independent utility functions, but defensive positionalists who seek to prevent a decline of their relative capabilities. Accordingly, when states are confident about their partners' compliance, they may still forgo gains resulting from cooperation if they expect different gains to shift the relative distribution of capabilities in favour of their partners. This diminishes the value that states attach to the functioning and endurance of international institutions. Neoliberals admit that at times states pay more attention to relative gains than absolute gains. Still, they claim that states do so only when they expect others to be hostile and deceptive and when their margins of survival are small. States can afford to focus on absolute gains if they expect substantial mutual gains through cooperation and deem threats with force highly unlikely. These expectations depend on the nature and effectiveness of prevailing rules. In this way institutions create and maintain the preconditions of their own existence.

      Neorealists and neoliberals thus agree on the existence of a causal link between the importance of relative gains considerations and the functioning of international institutions, but they emphasize different aspects of that causal relationship. Neorealists focus on relative gains as an independent variable. For them, such considerations crucially affect the relevance and effectiveness of international institutions. Strong relative-gains considerations inhibit cooperation and so limit the independent influence of institutions. Neoliberals place more emphasis on international institutions as an independent variable that determines to what extent relative gains considerations matter. Stable institutions make states less worried about who gains more as they enhance the "shadow of the future." Indeed, the higher future gains are valued relative to current gains, the smaller the incentive to stop cooperation. If one player defects today, the others are likely to do the same tomorrow.9

      These differences between neorealists and neoliberals follow from their dissimilar views regarding the determinants of state interests. For neorealists the relative distribution of national capabilities in the international system determines how states define their interests. Under conditions of anarchy, security is their overriding goal. The most important determinant of a state's security is its overall capability compared with the capabilities of other states. States regard capabilities as the ultimate basis for their security because of the difficulties in gauging the intentions of others.

      According to neoliberals, states calculate their interests not only from the international distribution of capabilities but also with regard to international institutions. International institutions affect a state's interest in two closely interrelated ways: they alter incentives and influence expectations of other states' behaviour and increase the costs of defection. What is more important, each state is aware of its partners' incentives being affected in the same way. Consequently, institutions enhance a state's capacity to predict the behaviour of other states. By following the rules and standards of international institutions, states signal their willingness to continue patterns of cooperation, and therefore reinforce expectations of stability, and with it the very stability of the system. As a result, states have to pay less attention to the distribution of capabilities and their changes through unequal gains when defining their interests.

    2. Virtual cooperation on the user level

      If we project the two perspectives outlined above onto the Internet, which of the two, then, is most plausible? Even after a cursory look, neoliberal institutionalism appears to cover best what is happening on the information super highway. The overall distribution of power in the system does not seem to bother Internet users to the same extent as states with regard to the international system. Without any real threat to their physical survival, "internauts" are likely to guide their behaviour by absolute gains considerations. Moreover, to satisfy their information needs, they attach great value to the survival and maintenance of their beloved communication system. This, in turn, affects their expectations and strengthens their willingness for conditional cooperation to maintain the system. In other words, interacting through Internet is an iterated game, and on top of that there is a long shadow of the future.

      Upon analysing the interaction among Internet users, we need to focus on those services that are most relevant. Both news groups and E-mail conferences - especially the unmoderated ones10 - are platforms for collective action reaching far beyond the level of interaction through ordinary E-mail or browsing. For this paper, we will not analyse IRC, as by its volatile nature, the "chatting" involves a more thorough investigation. We will limit ourselves to the interaction within News groups and E-mail conferences. Services like Gopher and the World Wide Web are less interactive and thus of lesser importance for this analysis.

      The news groups, originally conceived as an academic architecture for leaving messages within a specific interest field, form an electronic version of the classic "bulletin board." News groups expanded over the years to a total of 7,000.11 The topics vary from the original scientific subjects over socio-political discussions to the alternative and the bizarre. How does it work? Somebody posts a message on a bulletin board, say, a news group about politics in Norway (e.g. soc.culture.norway). This can be a piece of information, a general document of interest to somebody else, or a request for more information. The user will choose the news group(s) that meet his requirements the best. If a bulletin board covering his desired topic does not exist yet, he can ask his server administrator to create it. Each Internet-provider will decide whether it will support this group (i.e. whether it will copy the contents of the news group on a daily basis to his own server), allowing his users to join its discussions.

      Conversely, people sharing the same interests will "subscribe"to the same news group (i.e. select them). Similarly, they can also decide to take part in the same E-mail conference. It is on this specific level that rules are created and common practices emerge. This varies from the use of specific language to commonly shared rules and incentives (so the emergence of a "netiquette"). Again the famous 'Green Card' incident offers an instructive example as it brought to the surface common, Internet-specific practices. Two lawyers, Canter and Siegel, posted commercials for themselves on bulletin boards and mail-conferences intended for other purposes.12 They thus "violated" the unwritten rule of privacy: on Internet, users post messages there where they belong; one does not want to receive unordered mail. The reaction of the Internet community on the lawyers' dissident behaviour was very intense: from all over the world, people started to mail them "flames", messages in which they uttered their despise and malcontent, or in which they even threatened with more violent action. The result was that Canter and Siegel were "cut off" from the Net because of collective action by "offended" Internet users.

      Numerous are instances in which users develop rules and define group boundaries. The very evolution of the Internet illustrates this dynamic. Most of the discussion groups and mail-conferences offer examples of ordered debates in an international conference centre. The users themselves seem to guard the specific character of both discussion and news groups. As illustrated already, those who defy the rules are sanctioned by the other members of the group. Either they start a flame war, or they attack the dissidents through bombing with electronic mail (the so-called "spamming" - machine generated junk mail).13 In still other cases they resort to different types of message management.

  7. Conclusion: "Internet is what its users make of it".
  8. Although still very impressionistic, the analysis nonetheless permits us to draw some preliminary conclusions.

    First, neorealism's propositions may be of less relevance for studying problems of order and collective action in the Internet system. We have argued that the security dilemma and the related assumption of the dominance of relative gains considerations are not really suited to explain the behaviour of "internauts". The overall distribution of power in the system does not seem to bother Internet users as much as states in the international system.

    Conversely, neoliberal institutionalism appears more appropriate. Its assumptions allow for the following general proposition: without any real threat to their physical survival, "internauts" are guided in their behaviour by absolute gains considerations. Moreover, to remain capable of satisfying their information needs, they develop a common interest regarding the survival and maintenance of their beloved communication system. This very want, in its turn, affects their expectations and strengthens their willingness for conditional cooperation to maintain the system. Neoliberalism thus shows why Internet users, motivated by self-interest, are willing to engage in collective actions to 'self-regulate' the net. Simultaneously it suggests how systems of rules, that is institutions, have an independent influence on the users' calculus. More specifically, institutions affect actors' interests in two closely interrelated ways:

    1. they change incentives and influence expectations about the other actors' behaviour;
    2. they increase the costs of defection.

    It is important to note that all participants are aware of the fact that their expectations are affected in a similar way. Therefore, institutions enhance capacities of Internet users to predict the behaviour of their fellow users. By following basic rules and standards of behaviour, the participants signal their willingness to continue patterns of cooperation, and therefore reinforce expectations of stability, and with it the very stability of the system. As a result, internauts have to pay less attention to the distribution of capabilities and their changes through unequal gains when defining their interests.

    However, neoliberals have less to say about how Internet rules of behaviour come about. Our expectation is that constructivist structural theory, as it was developed within the discipline of international relations,14 holds great potential for developing a deeper understanding of how the rule-making mechanisms of Internet come about and develop further. More specifically, this theory suggests how actors can develop rule systems through concrete processes of interaction. As such, it demonstrates how anarchic systems may develop into communities or societies. Starting from this theory and combining it with examples of common understandings and practices within Internet, it should in principle be possible to deduce general propositions about how rule systems are generated within the Internet. The main assumption then is: "Internet is what its users make of it."


    1. Martin Hollis and Steve Smith, Explaining and Understanding International Relations, Oxford, Oxford University Press, 1992, p. 102.
    2. See e.g. Michael D. Lemonick, "The Net's Strange Day," Time Magazine, Vol. 147, (1996), No. 8 (online via http://www.pathfinder.com).
    3. Kenneth N. Waltz, Theory of International Relations, Reading, Addison-Wesley, 1979.
    4. The Internet can be accessed through various "programs" which are generally known as "services." We can distinguish E-mail (and E-mail conferences which derives from this service), Gopher, World Wide Web, News groups, FTP (File Transfer Protocol) and IRC (Internet Relay Chat).
    5. Martin Hollis and Steve Smith, Explaining and Understanding International Relations, pp. 115-116. See also Joseph M. Grieco, "Anarchy and the Limits of Cooperation: A Realist Critique of the Newest Liberal Institutionalism," in International Organization, Vol. 42, (1988), No. 3, pp. 485-508.
    6. John J. Mearsheimer, "The False Promise of International Institutions," International Security, Vol. 19, (1994-1995), No. 3, pp. 12-13.
    7. Robert O. Keohane, International Institutions and State Power, Boulder, Westview, 1989.
    8. Robert Axelrod and Robert O. Keohane, "Achieving Cooperation Under Anarchy: Strategies and Institutions," World Politics, Vol. 38, (1985), No. 2, pp. 226-54.
    9. Ibid.
    10. The total amount of news groups is estimated to exceed 12,000. However, this figure includes the news groups only stored at local servers and the "private" news groups of individual networks
    11. Unmoderated refers to the fact that there is no entity who controls the e-mails. The latter can be sent by any participant of the conference to all others without any intermediary.
    12. For a broad description of the Canter and Siegel Affair, see (e.g.) Philip Elmer-Dewitt, "Battle for the soul of the Internet," in Time Domestic, July 25 (1994), Vol. 144, No 4 (online via http://www.pathfinder.com).
    13. See for example "I've been Spammed" by Philip Elmer-Dewitt in Time Magazine, Vol. 147, March 18 (1996), No. 12.
    14. See among others Alexander Wendt, "Constructing International Politics," International Security, Vol. 20, (1995), No. 1, p. 72.


    © 1996, Centrum voor Polemologie
    Vakgroep Politieke Wetenschappen, VUB