Whichever way we look at the Internet, whether from the perspective of the network nodes, the transmission processes or the communication services, we see similar principles of organisation. The decentralised and open structure of the Internet is the determining factor at all levels of Net technology. The flow of data is basically controlled on the periphery or, more precisely, by the applications. The applications themselves also have a decentralised organisation. The host servers and the network´s transmission technology are as inaccessible to any form of central control as the communication services. A principle of self-administration reigns - with the result that the applications of the technology are gradually becoming more heterogeneous.

The common denominator in the evolution of Usenet, in the Unix "family trees" and in the further development of IP can be expressed in one maxim: support the free flow of data. The aim of the global connectivity project is the integration of all potential users and applications. This is evidenced by the renunciation of proprietary technologies in the area of standards development and by the enormous significance of "running code", the proof of functional efficiency and interoperability. The principle of open standards leaves it up to the users to decide how and to what purpose they apply these standards. Any comprehensive regulation of the transformation the Internet is undergoing in the area of services and data transmission is thus out of the question in reality.

But these transformations do not necessarily mean that the losses outweigh the gains in any of the areas under investigation. While prophesies of doom can be heard in reaction to the developments, the fact is that the possibilities for communication via the Internet are expanding rather than contracting. This is not only true of the number of people, organisations and data banks that can be reached, but also of the services that can be used for communication. A recent example of these new possibilities is Internet telephony.

If we compare the changes the Net has undergone in the individual areas studied during the period of observation, we perceive a pattern that can be described by the term "change through integration". The mutual interpenetration of WWW and Usenet is a good example of the principle of "incorporation", which could also serve as a model for the future establishment of new services or transmission technologies. It is above all by means of interoperability or mutual support, and not by direct substitution, that new applications or transmission procedures gain acceptance on the Net. IPv6 itself will only be able to gain a foothold on the Internet if it really is downwardly compatible and if it supports the current standard, IPv4. The installation basis of Internet technologies has become so large that any innovations at the level of host, transmission or service can only reckon with success if they are able to coexist with current procedures. Thus, even ever so purist Unix has acquired a graphical interface.

This tendency to integrate rather than replace is, incidentally, also found in the emergency measures that have been implemented on the Internet. An example is the reform of the address architecture being pursued since 1984. The new procedure for allocating Internet addresses (CIDR) can only be implemented as a modification of existing practices, i.e. it cannot radically change the previous allocation policy. The same pattern of development is visible in the emergence of the "alt" hierarchies on Usenet. The attempt to create an independent channel of distribution on Usenet only led to the coexistence of different hierarchies that obey different types of administration.

For the time being it is unlikely that this process of absorption between old and new will lead to a slower pace of development, but the conventions that hold the Net together are certainly likely to become more heterogeneous.

Because central forms of control are largely ruled out on the Net, innovations cannot be prescribed, but can only be arrived at through the active agreement of each individual site. This, if you will, is the virtual equivalent of "voting with one´s feet", where the expression of individual preferences amounts to a collective decision. The rapid spread of the World Wide Web illustrates how effective this form of decentralised coordination, based on nothing but free will, can be. The WWW established itself within a matter of months as a new Internet communication service without any formal agreements between developers and users ever being required. The development of the hypertext language and transmission protocol, of browsers, search engines and software agents, and of links to other Internet services are all occurring independently. Only the interoperability of new Net objects is subject to a general norm. In this sense the transformation of the Net cannot be attributed either to individuals or to institutions. Governance on the Internet is based instead on "distributed" forms of coordination, which manage to get by without hierarchies and with a minimum of centralised functions and rules.

The development of the host software Linux, of the Internet protocol IPv6 and of the administration of Usenet is something which happens in public, mainly or even exclusively on the Internet itself. Administrative and legal norms, as are characteristic of public- or private-sector standardisation bodies, are to a large extent incompatible with this system. The continuing development and administration of the Net is in fact subject to the same conditions as its use: everybody is free to participate actively and to try to influence matters according to their own interests. This governance by a kind of grassroots democracy does not, however, mean that there are not also mechanisms of exclusion of an informal kind in operation. A high level of technical competence and a knowledge of the context are the basic conditions for participation on Usenet, in the IETF and in the Unix community. The only voices likely to gain a hearing in discussions concerning the regulation of the Net are those of people who know how to relate to its cultural and technical traditions.

Can we expect, then, that in the short or long term the Net will lose its status as a self-organising space for communication? Are the sceptics right when they predict that this interactive, symmetrically organised communication device will eventually meet the same fate as the radio? One answer to such pessimistic forecasts is the observation that the decentralised architecture of the Internet has so far withstood all attempts at regulation and standardisation. It may be true that interaction is increasingly losing ground to "one-way" information services and that the Internet is thus becoming more like the traditional broadcast media; however, it is also to be expected that different forms of communication will coexist rather than replace each other.

In relation to the principles of transformation and governance that we found in the three areas investigated, we might add that attempts to control and normalise the Internet ultimately depend on the agreement of the users. Without their acceptance, technologies of encryption, authentication and control will have no effect. Whether the Internet will move towards traditional forms of social organisation and allow itself to be governed in the traditional sense is something which is still beyond the power of state authorities to decide, and thus remains to be seen.